JP2018179977A - Condition monitoring apparatus, condition monitoring system, and condition monitoring method - Google Patents

Abstract

A condition monitoring device capable of calculating an evaluation value for accurately determining whether an object has an abnormality is provided. A data processing device (80) includes a peak detection section (103) that detects a peak from a frequency spectrum, and a map generation section (104) that generates an abnormality map for the frequency spectrum. The anomaly map includes, as anomalous components, the frequency of the detected target peak and the frequency of the peak that appears together with the target peak when it is assumed that the target peak is caused by an anomaly. The data processing device 80 includes an abnormal peak extraction unit 105 that extracts, as an abnormal peak, a frequency peak that matches any of the abnormal components included in the abnormal map, and an abnormal peak corresponding to the abnormal map based on the spectral density of the abnormal peak. and a first evaluation value calculation unit 106 for calculating a first evaluation value indicating the degree of occurrence. [Selection drawing] Fig. 2

Description

  The present invention relates to a condition monitoring apparatus, a condition monitoring system, and a condition monitoring method for monitoring the condition of an object.

  2. Description of the Related Art Conventionally, there has been known a state monitoring device that diagnoses an abnormality of a machine in order to detect and maintain the abnormality of the machine at an early stage. For example, JP 2009-20090 A (Patent Document 1) corresponds to the theoretical value of the abnormal frequency of the vibration caused by the abnormality of the rotating part among the frequency components of the vibration signal generated from the rotating part of the mechanical equipment A technique is disclosed that diagnoses the presence or absence of an abnormality by extracting a frequency component that The theoretical value of the abnormal frequency is calculated in advance according to a predetermined relational expression.

JP, 2009-20090, A

  In the prior art as described above, the presence or absence of abnormality is diagnosed based on the frequency component corresponding to the theoretical value of the abnormal frequency calculated in advance. However, if noise or the like is included in the frequency component corresponding to the theoretical value of the abnormal frequency, the abnormality may be erroneously detected. Furthermore, the actual abnormal frequency may deviate from the theoretical value due to the effects of tolerances of the machine to be monitored for state monitoring, assembly accuracy and the like. When the object is a rotary machine, the actual abnormal frequency may deviate from the theoretical value also due to the change in rotational speed during vibration data measurement. In such a case, an abnormality can not be detected correctly.

  Further, the theoretical value of the abnormal frequency is calculated in advance, using as a model an abnormality of a specific part of the rotating component to be an object of state monitoring. Therefore, it is not possible to detect the occurrence of an abnormality whose unknown site is unknown.

  The present invention has been made to solve the above problems, and an object thereof is a state monitoring device capable of calculating an evaluation value for accurately determining the presence or absence of an abnormality of an object, a state monitoring System and condition monitoring method

  The present disclosure relates to a state monitoring device that monitors the state of an object. The state monitoring apparatus detects a peak from a frequency spectrum obtained by performing frequency analysis on waveform data measured by a sensor installed on an object, and a peak detection unit, and at least one anomaly map for the frequency spectrum. And a map generation unit to generate. The at least one anomaly map includes the frequency of one target peak selected from among the detected peaks, and the peak that appears together with the target peak when the target peak is assumed to be a peak due to a target anomaly. The frequency is included as an abnormal component. The state monitoring apparatus selects one target map from at least one abnormality map, and a peak having a frequency difference with a predetermined value from the detected peak to any of the abnormal components included in the target map is regarded as an abnormal peak. It further comprises a first evaluation value calculator that calculates a first evaluation value indicating the degree of occurrence of an abnormality corresponding to the target map based on the extracted abnormal peak extraction unit and the spectrum density of the abnormal peak.

  Preferably, the frequency spectrum is indicated by a data arrangement in which unit data in which frequencies are associated with spectral densities at the frequencies are arranged in order according to the frequencies. The peak detection unit detects, as a peak, unit data having a spectral density that indicates a local maximum and exceeds the first threshold in the data array.

  Preferably, the peak detection unit sets a first threshold that is a constant value regardless of the frequency based on the frequency spectrum. Alternatively, the peak detection unit may set a first threshold that fluctuates according to the frequency based on the frequency spectrum.

  Preferably, in at least one anomaly map, a first anomaly map in which the frequency of the target peak is a fundamental frequency and the fundamental frequency and the frequency of the harmonic of the fundamental wave having the fundamental frequency are anomalous components; And the frequency of the sideband wave of the fundamental wave, the frequency of the harmonics, and the frequency of the sideband waves of the harmonics as the anomalous component, and the frequency included in the predetermined frequency band including the frequency of the target peak is abnormal It includes at least one of the component and the third anomaly map.

  Preferably, at least one anomaly map is information in which the frequency is associated with a value of 0 or 1 for each frequency, and a value corresponding to a frequency that is an anomaly component is set to 1 and is not an anomaly component. The value corresponding to the frequency is set to zero. The abnormal peak extraction unit performs AND processing on the unit data detected as a peak by the peak detection unit and the target map, and extracts an abnormal peak by masking a peak of a frequency that is not an abnormal component.

  Preferably, the first evaluation value is a sum of spectral densities of abnormal peaks. Alternatively, the first evaluation value may be the sum of multiplication values obtained by multiplying the spectral density of the abnormal peak by the weighting factor according to the frequency of the abnormal peak.

  Preferably, for each part of the object, the state monitoring device stores abnormal part information in which part information identifying the part, a frequency caused by an abnormality in the part, and a second threshold are associated. From the inside, the frequency difference with the target peak corresponding to the target map indicates frequencies within a predetermined value, and abnormal part information indicating a second threshold value less than the first evaluation value corresponding to the target map is extracted as identification information And an abnormal part estimation unit configured to generate first estimation result information indicating that an abnormality has occurred in a part identified by the part information of the identification information.

  Alternatively, the at least one anomaly map may include two or more anomaly maps generated for the target peak based on two or more anomaly models in which the appearances of the anomaly components are different from each other. The state monitoring device further associates, for each part of the object, model information for identifying an abnormal model, part information for identifying the part, a frequency caused by an abnormality for the part, and a second threshold. In the database storing the abnormal part information, model information for identifying an abnormal model corresponding to the target map is shown, and a frequency difference with a target peak corresponding to the target map indicates a frequency within a predetermined value, and the target First estimation result information indicating that abnormality is occurring in a portion identified by the portion information of the identification information by extracting abnormal portion information indicating a second threshold value less than the first evaluation value corresponding to the map as identification information And an abnormal part estimation unit that generates

  Preferably, the abnormal part estimation unit can not extract identification information for the target map, and when the first evaluation value corresponding to the target map exceeds the third threshold, an unregistered abnormality occurs in the database. Generating second estimated result information indicating that.

  Preferably, the at least one anomaly map comprises a plurality of anomaly maps. The first evaluation value calculation unit sequentially selects each of the plurality of abnormality maps as a target map, and calculates a first evaluation value for each of the plurality of abnormality maps. The state monitoring device further includes a second evaluation value calculation unit that calculates, as a second evaluation value, the sum of the first evaluation values of the target map for which the identification information has been extracted by the abnormal part estimation unit. Alternatively, in the state monitoring apparatus, the first evaluation value of the target map from which the identification information is extracted by the abnormal part estimation unit and the identification information are not extracted, and the corresponding first evaluation value exceeds the third threshold. You may further provide the 2nd evaluation value calculation part which calculates a total with a 1st evaluation value as a 2nd evaluation value.

  A method according to another aspect of the present disclosure is a status monitoring system including the above-described status monitoring device and a terminal device. The terminal device displays a graph indicating the frequency spectrum on the display unit.

  Preferably, the terminal device displays a portion of the abnormal peak corresponding to the abnormal map selected from the at least one abnormal map in the frequency spectrum in a display format different from the remaining portion.

  Preferably, the terminal device removes anomalous peaks corresponding to the anomalous map selected from at least one anomalous map in the frequency spectrum.

  Preferably, the terminal device displays only anomalous peaks corresponding to the anomalous map selected from at least one anomalous map in the frequency spectrum.

  A method according to another aspect of the present disclosure is a state monitoring method of monitoring the state of an object. The state monitoring method comprises the steps of: detecting a peak from a frequency spectrum obtained by performing frequency analysis on waveform data measured by a sensor installed on an object; and generating at least one anomaly map for the frequency spectrum And step. In the state monitoring method, one target map is selected from at least one abnormal map, and a peak having a frequency difference within a predetermined value from a detected peak to any of the abnormal components included in the target map is regarded as an abnormal peak. The method may further include the steps of extracting and calculating a first evaluation value indicating the degree of occurrence of an abnormality corresponding to the target map based on the spectral density of the abnormal peak.

  According to the state monitoring apparatus or the state monitoring method of the present invention, it is possible to calculate an evaluation value for accurately determining the presence or absence of an abnormality of an object.

BRIEF DESCRIPTION OF THE DRAWINGS It is the figure which showed roughly the structure of the wind power generator with which the state monitoring apparatus based on embodiment of this invention was applied. It is a schematic block diagram showing functional composition of a data processor concerning an embodiment. It is a figure which shows an example of the vibration waveform measured by the state monitoring sensor. It is a figure which shows the vibration waveform after performing a filtering process with respect to the vibration waveform shown in FIG. It is a figure which shows an example of the frequency spectrum produced | generated by the frequency analysis part shown in FIG. It is a figure which shows the unit data from which PSD becomes local maximum. It is a figure which shows unit value which shows local maximum and has PSD larger than 1st threshold value. It is a figure which shows an example of a 1st abnormality map. It is a figure which shows an example of a 2nd abnormality map. It is a figure which shows the relationship of the detected peak, a 1st abnormal map, and an abnormal peak. It is a figure which shows the relationship between the detected peak, a 2nd abnormal map, and an abnormal peak. It is a figure which shows an example of the 1st evaluation value calculated by the 1st evaluation value calculation part shown in FIG. It is a figure which shows the example of the abnormal site | part information which the database shown in FIG. 2 memorize | stores. It is a figure which shows the example of the estimation result information produced | generated by the abnormal site | part estimation part shown in FIG. It is a flowchart which shows the flow of the process in the data processor which concerns on embodiment. It is a flowchart which shows the flow of the subroutine of the peak detection process shown in FIG. It is a figure which shows the relationship between the detected peak, a 3rd abnormality map, and an abnormal peak. An example of calculation results of the first evaluation value, abnormal part information stored in the database, and estimation result information in the case where the object may generate an abnormality according to the first to third abnormality models is shown. FIG. It is a figure which shows the example of the peak detected using the 1st threshold value which fluctuates according to frequency. It is a figure which shows the example of the abnormal peak extracted using the 2nd abnormal map. It is a schematic block diagram which shows the structure of a terminal device. It is a figure which shows an example of the screen displayed on the display part of a terminal device. It is a figure which shows another example of the screen displayed on the display part of a terminal device. It is a figure which shows another example of the screen displayed on the display part of a terminal device. It is a figure which shows the frequency spectrum of abnormal peak vicinity. It is a figure which shows an example of the method of removing an abnormal peak. It is a figure which shows another example which removes an abnormal peak. It is a figure which shows an example of the method of removing other than an abnormal peak.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the following drawings, the same or corresponding parts are denoted by the same reference numerals, and the description thereof will not be repeated. Also, the modifications described below may be selectively combined as appropriate.

  Below, the speed increasing gear of a wind power generator is demonstrated as an example of the target whose state is monitored by the state monitoring device. However, the object is not limited to the speed increaser of the wind power generator, and may be any one that causes changes in waveform data such as vibration, sound, acoustic emission (AE), etc. due to an abnormality. For example, the objects include various devices installed in factories and power plants, railway cars, and the like.

<Configuration of Wind Turbine>
FIG. 1 is a view schematically showing the configuration of a wind turbine generator to which a condition monitoring device according to the present embodiment is applied. Referring to FIG. 1, the wind turbine 10 includes a main shaft 20, a hub 25, a blade 30, a speed increasing device 40, a generator 50, a main shaft bearing 60, a state monitoring sensor 70, and data processing. And an apparatus 80. The speed increasing gear 40, the generator 50, the spindle bearing 60, the state monitoring sensor 70, and the data processing device 80 are stored in the nacelle 90. The nacelle 90 is supported by the tower 100.

  The main shaft 20 enters the nacelle 90 and is connected to the input shaft of the step-up gear 40, and is rotatably supported by the main shaft bearing 60. The main shaft 20 transmits the rotational torque generated by the blade 30 which receives the wind force to the input shaft of the step-up gear 40. The blades 30 are provided on the hub 25 and convert wind power into rotational torque and transmit it to the main shaft 20. The main shaft bearing 60 is fixed in the nacelle 90 and rotatably supports the main shaft 20.

  The speed increasing machine 40 is provided between the main shaft 20 and the generator 50, accelerates the rotational speed of the main shaft 20, and outputs the speed to the generator 50. As an example, the speed increasing gear 40 is configured by a gear speed increasing mechanism including a planetary gear, an intermediate shaft, a high speed shaft and the like. In the speed increasing gear 40, a plurality of bearings rotatably supporting a plurality of shafts are provided. The plurality of bearings are, for example, rolling bearings, and have an outer ring (fixed ring), a rolling element, and an inner ring (rotating ring).

  The state monitoring sensor 70 is fixed to the speed increasing machine 40 and measures waveform data indicating the state of the speed increasing machine 40. In the present embodiment, the state monitoring sensor 70 measures the vibration waveform of the step-up gear 40, and outputs the measured vibration waveform data to the data processing device 80. The state monitoring sensor 70 is configured of, for example, an acceleration sensor using a piezoelectric element.

  The generator 50 is connected to the output shaft of the speed increaser 40, and generates electric power by the rotational torque received from the speed increaser 40. The generator 50 is configured of, for example, an induction generator. In addition, also in the generator 50, the bearing which supports a rotor rotatably is provided.

  The data processing device 80 is provided in the nacelle 90, and receives vibration waveform data of the step-up gear 40 from the state monitoring sensor 70. The data processing device 80 functions as a condition monitoring device that monitors the condition of the transmission 40 based on the vibration waveform data received from the condition monitoring sensor 70. The data processing device 80 and the condition monitoring sensor 70 constitute a condition monitoring system that monitors the condition of the transmission 40.

  The data processing device 80 includes a central processing unit (CPU), a read only memory (ROM) for storing processing programs and the like, and a random access memory (RAM) for temporarily storing data, and further inputs and outputs various signals. Input / output ports etc. (all not shown). The data processing device 80 executes various data processing in accordance with the program stored in the ROM. The processing executed by the data processing device 80 is not limited to the processing by software, and may be processed by dedicated hardware (electronic circuit).

<Functional Overall Configuration of Data Processing Device>
FIG. 2 is a schematic block diagram showing the functional configuration of the data processing apparatus 80 according to the present embodiment. Referring to FIG. 2, data processing device 80 includes filter unit 101, frequency analysis unit 102, peak detection unit 103, map generation unit 104, abnormal peak extraction unit 105, and first evaluation value calculation unit 106. And a database (DB) 107, an abnormal part estimation unit 108, a second evaluation value calculation unit 109, and an output processing unit 110.

<Filter part>
The filter unit 101 performs filtering processing on the vibration waveform data received from the state monitoring sensor 70 to pass components of a predetermined frequency band and attenuate components of other frequency bands. As an example, the filter unit 101 includes a high pass filter that passes signal components higher than a predetermined frequency and blocks low frequency components. Furthermore, the filter unit 101 may perform envelope processing or the like to emphasize a peak in a frequency spectrum obtained by the frequency analysis unit 102 described later.

  FIG. 3 is a diagram showing an example of the vibration waveform measured by the state monitoring sensor 70. As shown in FIG. FIG. 4 is a diagram showing a vibration waveform after the filter unit 101 performs the filtering process on the vibration waveform shown in FIG. Referring to FIGS. 3 and 4, it can be understood from the vibration waveform data that the low frequency component is removed by the filter unit 101.

<Frequency analysis unit>
The frequency analysis unit 102 performs Fourier transform on the vibration waveform data output from the filter unit 101 to generate a frequency spectrum. The frequency spectrum is indicated by a data array in which unit data in which frequencies are associated with power spectral densities (PSDs) (hereinafter referred to as PSDs) at the frequencies are arranged in order according to the frequencies.

  FIG. 5 is a diagram illustrating an example of the frequency spectrum generated by the frequency analysis unit 102. Referring to FIG. 5, the frequency spectrum is represented by a graph in which the horizontal axis is frequency and the vertical axis is PSD.

<Peak detection unit>
The peak detection unit 103 detects a peak sufficiently larger than the noise amount from the frequency spectrum. The peak detection unit 103 detects, as a peak, unit data indicating a local maximum in the frequency spectrum and having a PSD exceeding the first threshold.

FIG. 6 is a diagram showing unit data (unit data of frequency fi) in which the PSD has a maximum value. Referring to FIG. 6, in the three consecutive unit data (unit data of frequencies f _{i -1} , f _i and f _{i +1} ), the PSD of the unit data (unit data of frequency f _i ) in the middle is on both sides It is larger than the PSD of unit data (unit data of frequency f _{i -1} , unit data of frequency f _{i +1} ).

FIG. 7 is a diagram showing unit data having a maximum value and having a PSD larger than the first threshold. In the example shown in FIG. 7, the peak detection unit 103 detects a peak indicated by unit data of the frequency f _i and a peak indicated by unit data of the frequency f _j .

  The peak detection unit 103 sets a first threshold, which is a constant value, based on the PSDs of all unit data constituting the frequency spectrum. For example, the peak detection unit 103 sets a value 10 times the median of the PSDs of all unit data as the first threshold.

The peak detection unit 103 generates a matrix P of the following formula (1) indicating the frequency and the PSD of the peak detected from the frequency spectrum. In Equation (1) below, in the matrix P, N peaks are detected by the peak detection unit 103, the frequency of the n-th (n = 1 to N) peak is f _n , and the value of PSD is ps d _n To indicate that

<Map generation unit>
The map generation unit 104 generates at least one anomaly map for the frequency spectrum obtained by the frequency analysis unit 102. The abnormality map indicates the frequency of one target peak selected from the peaks detected by the peak detection unit 103 and the target when it is assumed that the target peak is a peak caused by the abnormality of the speed increasing device 40. The peak and the frequency of the peak appearing in the frequency spectrum are included as anomalous components.

  The map generation unit 104 generates, as an abnormality map, information in which the frequency and the value of 0 or 1 are associated with each other for each frequency. In the abnormality map, a value corresponding to a frequency that is an abnormal component is set to 1, and a value corresponding to a frequency that is not an abnormal component is set to 0.

  When an abnormality occurs in any part of the speed increaser 40, a peak occurs at a frequency corresponding to the part in the frequency spectrum. The appearance of the peak in the frequency spectrum differs depending on the location of the abnormality and the type of the abnormality.

  For example, when damage occurs to the outer ring which is a fixed ring of a bearing included in the speed increasing gear 40, it has a fundamental wave according to the damage and a frequency that is an integral multiple of the frequency (basic frequency) of the fundamental wave. Harmonics are generated. Below, the abnormal model which a fundamental wave and a harmonic generate | occur | produce is called "1st abnormal model."

  When damage occurs to the inner ring which is a rotating wheel of the bearing included in the speed increasing gear 40, in addition to the fundamental wave and its harmonics according to the damage, sideband waves of each of the fundamental wave and harmonics are generated . Below, the abnormal model which a fundamental wave, a harmonic, and each sideband wave of a fundamental and a harmonic generate is called "the 2nd abnormal model."

  The map generation unit 104 generates two anomaly maps for the target peak according to the first anomaly model and the second anomaly model. Hereinafter, the anomaly map generated according to the first anomaly model is referred to as a "first anomaly map", and the anomaly map generated according to the second anomaly model is referred to as a "second anomaly map".

  When the number of peaks detected by the peak detection unit 103 is N as shown in the above equation (1), the map generation unit 104 selects one target peak in order from the N peaks, Two anomaly maps are generated for the selected target peak. In this case, the map generation unit 104 generates 2N anomaly maps for the frequency spectrum obtained by the frequency analysis unit 102.

FIG. 8 is a diagram showing an example of the first abnormality map. Referring to FIG. 8, the map generation unit 104, the frequency _{f n} of the peak of interest, an integral multiple of the frequency _{f n} (2-fold, 3-fold, ...) frequency _2f n of, _3f n, ... And A as an anomalous component to generate a first anomalous map. In other words, the map generation unit 104 generates a first anomaly map including the fundamental frequency and the harmonic frequency of the fundamental wave having the fundamental frequency as the anomalous component, with the frequency f _{n of the} target peak as the fundamental frequency. .

The map generation unit 104 generates a first anomaly map M_n (1) as shown in the following equation (2) for the target peak of the frequency f _n .
M_n (1) = [.. 0010 ... 010 ... 010 ...] equation (2)
In the first abnormality map M_n (1), a value corresponding to a frequency that is an abnormal component is set to 1, and a value corresponding to a frequency that is not an abnormal component is set to 0.

FIG. 9 is a diagram showing an example of the second anomaly map. Referring to FIG. 9, the map generation unit 104, the frequency _{f n} of the peak of interest, an integral multiple of the frequency _{f n} (2-fold, 3-fold, ...) frequency _2f n of, _3f n, ... When, the including frequency _{f n ± f r, f n} ± 2f r, 2f n ± f r, 2f n ± 2f r, 3f n ± f r, 3f n ± 2f r, and ... as abnormal component 2 Generate an anomaly map. In other words, with the frequency f _{n of the} target peak as the fundamental frequency, the map generation unit 104 determines the fundamental frequency, the harmonic frequency of the fundamental wave having the fundamental frequency, the frequency of the sideband wave of the fundamental wave, and the harmonics A second anomalous map is generated which includes the sideband wave frequency of the wave as the anomalous component.

Here, the difference f _r between the frequency of the fundamental wave (fundamental frequency) and the frequency of the sideband wave is predetermined. Alternatively, the map generation unit 104 may calculate the difference between the frequency f _n of the target peak and the frequency (f _n-1 or f _{n + 1} ) of another peak closest to the target peak (for example, | f _n-1 −f _{n + 1} | ) _May be set as f _r .

The map generation unit 104 generates a second anomaly map M_n (2) as shown in the following equation (3) for the target peak whose frequency is f _n .
M_n (2) = [.. 00100100100001010... 01001..]] Equation (3)
In the second abnormality map M_n (2), the value corresponding to the frequency that is the abnormal component is set to 1, and the value corresponding to the frequency that is not the abnormal component is set to 0.

<Abnormal peak extraction unit>
The abnormal peak extraction unit 105 sequentially selects one abnormality map as an object map from among the abnormality maps generated by the map generation unit 104, and for each object map, an abnormality is detected from among the peaks detected by the peak detection unit 103. Extract the peaks. Here, the abnormal peak is a peak whose frequency difference with any of the abnormal components included in the target map is within a predetermined value. Note that the abnormal peak extraction unit 105 may extract only peaks having a frequency difference of 0 with any of the abnormal components included in the target map (that is, peaks matching with any of the abnormal components) as abnormal peaks. A peak whose frequency difference is equal to or less than a predetermined value th (th> 0) may be extracted as an abnormal peak. By extracting the abnormal peak using the predetermined value th, the influence of the measurement error by the state monitoring sensor 70 or the error generated in the Fourier transform by the frequency analysis unit 102 can be removed, and the extraction accuracy of the abnormal peak is improved. It can be done.

  The abnormal peak extraction unit 105 uses the kth abnormal map M_n (k) (k = 1 or 2) shown in the above equation (2) or the equation (3) in the matrix P shown in the above equation (1). An abnormal peak can be extracted by masking the peaks of frequencies that are not abnormal components by AND processing.

The abnormal peak extraction unit 105 uses the kth abnormal map M_n (k) generated for the detected n-th peak (peak of frequency f _n ) to extract the frequency and PSD value of the abnormal peak extracted. The matrix Pex_n (k) to be shown is generated (see Equation (4) below). In the matrix Pex_n (k), fex_n _m (k) indicates the frequency of the m-th anomalous peak, and psdex_n _m (k) indicates the value of the PSD of the m-th anomalous peak.

FIG. 10 is a diagram showing a relationship between a detected peak, a first abnormality map, and an abnormal peak. Referring to FIG. 10, using the first anomaly map M_n (1) generated for the peak at frequency f _n , the anomaly peak extraction unit 105 determines the frequency fex_n ₁ (1), fex_n ₂ (1), Extract _three abnormal peaks of fex_n ₃ (1). The frequency fex_n ₁ (1) coincides with the frequency f _n . The PSD values of the three abnormal peaks are psdex_n ₁ (k), psdex_n ₂ (k), and psdex_n ₃ (k), respectively.

FIG. 11 is a diagram showing a relationship between a detected peak, a second abnormality map, and an abnormal peak. Referring to FIG. 11, using the second anomaly map M_n (2) generated for the peak at frequency f _n , the anomaly peak extraction unit 105 generates the frequencies fex_n ₁ (2) and fex_n ₂ (2), Extract multiple abnormal peaks. The frequency fex_n ₃ (2) coincides with the frequency f _n . The PSD values of the plurality of abnormal peaks are psdex_n ₁ (k), psdex_n ₂ (k),.

<First evaluation value calculation unit>
The first evaluation value calculation unit 106 calculates, for each target map, a first evaluation value indicating the degree of occurrence of an abnormality corresponding to the target map based on the PSD of the abnormal peak. The first evaluation value calculation unit 106 calculates the sum of the PSDs of the abnormal peaks extracted by the abnormal peak extraction unit 105 as a first evaluation value. Specifically, when the first evaluation value calculation unit 106 sets the k th anomaly map generated for the peak of the frequency f _n as the target map, the first evaluation value En (En ( Calculated as k).

  FIG. 12 is a diagram illustrating an example of the first evaluation value calculated by the first evaluation value calculation unit 106. Referring to FIG. 12, the first evaluation value calculation unit 106 calculates, for each of the detected peaks, first evaluation values corresponding to two abnormality maps generated for the peaks. Therefore, when the peak detection unit 103 detects N peaks, the first evaluation value calculation unit 106 calculates 2N first evaluation values.

<Database>
The database 107 includes, for each part of the speed increasing device 40, part information for identifying the part, model information for identifying an abnormal model corresponding to the abnormality in the part, and a fundamental wave generated due to the abnormality in the part. The abnormal region information in which the second frequency (basic frequency) is associated with the second threshold is stored.

  FIG. 13 is a diagram showing an example of the abnormal part information stored in the database 107. As shown in FIG. Referring to FIG. 13, in the database 107, for example, with respect to the outer ring of the bearing included in the speed increasing device 40, the abnormal model corresponding to the abnormality of the outer ring is the first abnormal model, and The abnormal site information indicating that the frequency of the generated fundamental wave is 37.2 Hz and the second threshold is 100 is stored.

  The abnormal part information stored in the database 107 is created in advance by an experiment or simulation using the speed increaser 40 as the object. The second threshold is the total value of the PSDs of the fundamental and harmonics (fundamental, harmonics and sidebands in the case of the second anomaly model) generated when an anomaly occurs in the corresponding part in the experiment or simulation It is set to a value less than and sufficiently larger than noise.

<Abnormal part estimation part>
The abnormal part estimation unit 108 estimates a part where an abnormality has occurred in the transmission 40, based on the first evaluation value calculated for each target map and the abnormal part information stored in the database 107.

  Specifically, the abnormal part estimation unit 108 extracts abnormal part information satisfying the following conditions (a) to (c) as identification information from the database 107 for each target map. The condition (a) is a condition that the abnormal model identified by the model information of the abnormal part information matches the abnormal model corresponding to the target map. Condition (b) is a condition that the difference (frequency difference) between the fundamental frequency of the abnormal part information and the frequency of the target peak corresponding to the target map is within a predetermined value. The condition (c) is a condition that the second threshold value of the abnormal part information is less than the first evaluation value corresponding to the target map.

  If the abnormal part estimation unit 108 can extract identification information corresponding to the target map, the abnormal part estimation unit 108 estimates that an abnormality occurs in the part identified by the part information of the extracted identification information. Furthermore, the abnormal part estimation unit 108 generates estimation result information including the part information, the model information and the fundamental frequency in the extracted identification information, and the first evaluation value corresponding to the target map, and the generated estimation result information Are output to the second evaluation value calculation unit 109 and the output processing unit 110.

If the abnormal part estimation unit 108 can not extract the identification information corresponding to the target map, the abnormal part estimation unit 108 compares the first evaluation value corresponding to the target map with the third threshold. The third threshold is preset to a value sufficiently larger than noise. If the first evaluation value exceeds the third threshold, the abnormal part estimation unit 108 estimates that an abnormality has occurred in a part not registered in the database 107. Furthermore, the abnormal part estimation unit 108 determines part information indicating that the part where the abnormality is occurring is unknown, model information indicating an abnormal model corresponding to the target map, and the frequency f of the target peak corresponding to the target map. _The estimation result information including _n and the first evaluation value corresponding to the target map is generated, and the generated estimation result information is output to the second evaluation value calculation unit 109 and the output processing unit 110. The said estimation result information is information which shows that abnormality generate | occur | produces in the site | part which is not registered into the database 107. FIG.

  FIG. 14 is a diagram showing an example of the estimation result information generated by the abnormal part estimation unit 108. As shown in FIG. FIG. 14 shows an example of estimation result information when the first evaluation value shown in FIG. 12 is calculated, the abnormal part information shown in FIG. 13 is stored in the database 107, and the third threshold is set to 300. Indicated. The abnormal part estimation unit 108 determines that the first evaluation value corresponding to the abnormality map generated using the first abnormality model for the peak with a frequency of 37.2 Hz has the frequency as a fundamental frequency and the abnormality model. It confirms that it is larger than the 2nd threshold of abnormal part information shown. Thereby, the abnormal part estimation unit 108 estimates “the outer ring of the bearing” as a part where the abnormality occurs, and generates estimation result information (see the second line in FIG. 14) indicating the result of the estimation.

  Furthermore, the abnormal part estimation unit 108 generates a second evaluation value corresponding to the abnormality map generated using the first abnormality model for the peak at a frequency of 87.9 Hz and a second evaluation value for the peak at a frequency of 191.6 Hz. It is confirmed that the first evaluation value corresponding to the abnormality map generated using the abnormality model is larger than the third threshold "300". Thereby, the abnormal part estimation unit 108 estimates that an abnormality occurs at two unknown parts, and estimation result information corresponding to each of the two parts (the third and fourth rows in FIG. 14). Create an eye).

<Second evaluation value calculation unit>
The second evaluation value calculation unit 109 calculates, for each frequency spectrum, the sum of the first evaluation values included in the estimation result information generated for the frequency spectrum as a second evaluation value. The second evaluation value calculation unit 109 outputs the calculated second evaluation value to the output processing unit 110.

<Output processing unit>
The output processing unit 110 outputs the estimation result information and the second evaluation value to the terminal device of the user who is at a remote place using a wired or wireless communication system. Thereby, the terminal device can notify, for example, the display unit of the estimation result information and the second evaluation value. As a result, the user can easily confirm the estimation result information and the second evaluation value.

<Flow of processing of data processing device>
Next, the flow of processing in the data processing device 80 will be described with reference to FIG. FIG. 15 is a flowchart showing the flow of processing in the data processing apparatus 80.

  First, in step S1, the data processing device 80 performs preprocessing such as filtering processing on the vibration waveform data received from the state monitoring sensor 70. The pre-processing is performed by the filter unit 101.

  Next, in step S2, the data processing device 80 generates a frequency spectrum by performing Fourier transform as frequency analysis processing on the vibration waveform data on which the preprocessing has been performed. The frequency analysis process is performed by the frequency analysis unit 102.

  Next, in step S3, the data processing device 80 detects a peak sufficiently larger than noise from the frequency spectrum. The peak detection process is performed by the peak detection unit 103. Here, it is assumed that the data processing device 80 detects N peaks. The data processing apparatus 80 sequentially selects one target peak from the detected N peaks, and performs the processes of steps S4 to S6 on the selected target peak. The processes of steps S4 to S6 are performed on each of the N peaks detected in step S3.

  In step S4, the data processing device 80 generates a first anomaly map and a second anomaly map for the target peak. The process of generating the anomaly map is executed by the map generation unit 104.

  Next, in step S5, the data processing device 80 determines, from among the N peaks detected in step S3, the difference with the frequency of any abnormal component of the first abnormality map generated in step S4 to a predetermined value. Within the peak is extracted as a first abnormal peak. Similarly, among the N peaks detected in step S3, the data processing device 80 sets the difference with the frequency of any abnormal component of the second abnormality map generated in step S4 within a predetermined value. The peak is extracted as a second abnormal peak. The abnormal peak extraction process is performed by the abnormal peak extraction unit 105.

  Next, in step S6, the data processing device 80 calculates the sum of PSDs of the first abnormal peak extracted using the first abnormal map as a first evaluation value corresponding to the first abnormal map. Similarly, the data processing device 80 calculates the sum of PSDs of the second abnormal peak extracted using the second abnormal map as a first evaluation value corresponding to the second abnormal map. The calculation process of the first evaluation value is executed by the first evaluation value calculation unit 106.

  The processes of steps S4 to S6 are executed for each of the N peaks to generate 2N abnormality maps, and the 2N first evaluation values corresponding to the 2N abnormality maps are generated. It is calculated.

  When the processes of steps S4 to S6 for each of the N peaks are completed, the data processing apparatus 80 sequentially selects one anomaly map as a target map from the 2N anomaly maps, and selects the selected target map. The process of step S7 is performed. In step S7, the data processing device 80 compares the first evaluation value corresponding to the target map, the fundamental frequency of the target map, and the abnormal model corresponding to the target map with the abnormal part information stored in the database 107 and compares them. Generate estimation result information based on the result. As described above, the estimation result information includes the site information of any abnormal site information stored in the database 107 or the site information indicating “unknown”. The abnormal part estimation process is performed by the abnormal part estimation unit 108.

  When the process of step S7 for all the 2N abnormality maps is finished, next, in step S8, the data processing device 80 calculates the sum of the first evaluation values included in the estimation result information as a second evaluation value. The calculation process of the second evaluation value is executed by the second evaluation value calculation unit 109.

  Finally, the data processing device 80 outputs the estimation result information and the second evaluation value to the terminal device of the user (step S9), and ends the processing. The output processing is performed by the output processing unit 110.

<Flow of peak detection process>
Next, with reference to FIG. 16, the flow of a subroutine of the peak detection process (step S3) shown in FIG. 15 will be described. FIG. 16 is a flowchart showing the flow of a subroutine of peak detection processing.

  First, in step S31, the data processing device 80 specifies unit data that constitutes the apex of the convex portion in the frequency spectrum. Specifically, the data processing apparatus 80 specifies unit data whose PSD is the maximum value among unit data constituting a data array indicating a frequency spectrum (see FIG. 6).

  Next, in step S32, the data processing device 80 sets a first threshold based on the values of all the PSDs of the unit data identified in step S31. For example, the data processing device 80 sets 10 times the median of all PSDs of the unit data identified in step S31 as the first threshold.

  Next, in step S33, the data processing apparatus 80 extracts unit data having a PSD exceeding the first threshold from the unit data identified in step S31, and detects the extracted unit data as a peak. Thus, the subroutine of step S3 is completed.

<Advantage>
The data processing apparatus (state monitoring apparatus) 80 according to the above embodiment includes a peak detection unit 103, a map generation unit 104, an abnormal peak extraction unit 105, and a first evaluation value calculation unit 106. The peak detection unit 103 detects at least one peak from a frequency spectrum obtained by performing frequency analysis on vibration waveform data. The map generation unit 104 generates at least one anomaly map for the vibration waveform data. The abnormal peak extraction unit 105 selects one abnormal map as a target map from at least one abnormal map, and the frequency difference between any of the detected peaks and any of the abnormal components included in the target map is within a predetermined value. Peaks are extracted as abnormal peaks. The first evaluation value calculation unit 106 calculates a first evaluation value indicating the degree of occurrence of an abnormality corresponding to the target map, based on the spectral density of the abnormal peak.

  According to the above configuration, for each abnormality map generated based on the peak detected from the frequency spectrum, the first evaluation value indicating the degree of occurrence of the abnormality corresponding to the target map is calculated. That is, as in the prior art, not only frequency components corresponding to theoretical values of the abnormal frequency calculated in advance but all peaks appearing in the frequency spectrum are considered, and for each error map generated for each peak, 1 Evaluation value is calculated. Thereby, even if a peak is generated at a frequency deviated from the theoretical value due to some influence, the first evaluation value corresponding to the anomaly map generated for the peak is confirmed by checking the first evaluation value. The presence or absence of an abnormality causing a peak can be accurately determined. Furthermore, by checking the first evaluation value, it is possible to determine the presence or absence of an abnormality whose cause is unknown.

  The abnormality map includes not only the frequency of the target peak but also the frequency of the peak appearing together with the target peak when it is assumed that the target peak is a peak caused by the abnormality of the speed increasing device 40 as an abnormal component. Therefore, the first evaluation value is calculated in consideration of not only the target peak but also the peak attached to the target peak. Thereby, even if noise or the like is included in the target peak, the abnormality corresponding to the target peak is confirmed by confirming the first evaluation value calculated in consideration of the peak associated with the target peak. The presence or absence of can be accurately determined.

  As described above, the data processing device 80 having the above-described configuration can output the first evaluation value for accurately determining the presence or absence of abnormality of the speed increasing device 40 which is the object.

  Further, the frequency spectrum is indicated by a data arrangement in which unit data in which the frequency and the spectral density at the frequency are associated are sequentially arranged according to the frequency. The peak detection unit 103 detects, as a peak, unit data having a spectral density that indicates a local maximum and exceeds the first threshold value in the data array. The first threshold is set, for example, to a value larger than noise. This can suppress noise from being erroneously detected as a peak.

  Furthermore, the peak detection unit 103 sets a first threshold, which is a constant value, based on the frequency spectrum. Thereby, the load of processing of peak detection by the peak detection unit 103 can be reduced.

  Furthermore, the map generation unit 104 sets the frequency of the target peak as a fundamental frequency, the first anomaly map in which the fundamental frequency and the frequency of the harmonic of the fundamental wave having the fundamental frequency are anomalous components, the fundamental frequency and the fundamental frequency A second anomaly map is generated in which the frequency of the sideband wave of the wave, the frequency of the harmonic wave, and the frequency of the sideband wave of the harmonic wave are anomalous components. Generally, when an abnormality occurs in a fixed ring of a rolling bearing, a fundamental wave of a frequency according to the abnormality and its harmonics are generated. On the other hand, when an abnormality occurs in the rotating wheel, a fundamental wave of a frequency according to the abnormality, its harmonics, and sideband waves of the fundamental wave and the harmonics are generated. Many other anomalies also produce the fundamental and its harmonics, or the fundamental and its harmonics and their sidebands. Therefore, according to the above configuration, by checking the first evaluation value corresponding to the generated abnormality map, it is possible to determine the presence or absence of abnormality in various parts.

  Furthermore, the abnormality map is information in which the frequency is associated with the value of 0 or 1 for each frequency, the value corresponding to the frequency that is the abnormal component is set to 1, and it corresponds to the frequency that is not the abnormal component. The value is set to 0. The abnormal peak extraction unit 105 performs an AND process on the unit data detected as a peak by the peak detection unit 103 and the target map, and extracts an abnormal peak by masking a peak of a frequency that is not an abnormal component. Thereby, the abnormal peak extraction unit 105 can easily extract an abnormal peak.

  Furthermore, the first evaluation value calculation unit 106 calculates the sum of the spectral density of the abnormal peak as a first evaluation value. In general, as the degree of abnormality increases, the spectral density of peaks generated due to the abnormality increases. Therefore, the degree of abnormality can be easily determined by confirming the first evaluation value which is the sum of the spectral density.

  Furthermore, the data processing device 80 includes a database 107 and an abnormal part estimation unit 108. The database 107 includes, for each part of the speed increasing device 40, a second threshold, part information for identifying the part, fundamental frequency of a fundamental wave caused by an abnormality for the part, and model information for identifying an abnormality model. The associated abnormal part information is stored. The abnormal part estimation unit 108 indicates, from the database 107, a second threshold value less than the first evaluation value corresponding to the target map, and indicates a fundamental frequency whose frequency difference with the target peak corresponding to the target map is within a predetermined value. Abnormal site information indicating model information identifying an abnormal model corresponding to the target map is extracted as identification information. The abnormal part estimation unit 108 generates estimation result information (first estimation result information) indicating that an abnormality occurs in the part identified by the part information of the extracted identification information. As a result, the data processing apparatus 80 can accurately estimate the part where the abnormality has occurred based on the abnormal part information registered in the database 107 in advance.

  Furthermore, the abnormal part estimation unit 108 can not extract identification information for the target map, and when the first evaluation value corresponding to the target map exceeds the third threshold, an unregistered abnormality occurs in the database 107. Generating estimation result information (second estimation result information) indicating that the As described above, the data processing device 80 can also detect an abnormality whose cause is unknown.

  Furthermore, the data processing device 80 includes a second evaluation value calculation unit 109. The second evaluation value calculation unit 109 does not extract the first evaluation value of the target map from which the identification information has been extracted by the abnormal part estimation unit 108 and the identification information, and the corresponding first evaluation value corresponds to the third threshold. A total with the first evaluation value of the exceeding object map is calculated as a second evaluation value. By confirming the second evaluation value, it is possible to comprehensively determine the degree of abnormality of all parts of the speed increasing gear 40.

  As shown in FIG. 15, the state monitoring method according to the present embodiment includes step S3, step S4, step S5, and step S6. In step S3, at least one peak is detected from the frequency spectrum obtained by frequency analysis of the vibration waveform data. At step S4, at least one anomaly map is generated for the vibration waveform data. In step S5, one abnormality map is selected as a target map, and a peak having a frequency difference with any of the abnormal components included in the target map within a predetermined value is extracted as an abnormal peak from the peaks detected in step S3. . In step S6, a first evaluation value indicating the degree of occurrence of an abnormality corresponding to the target map is calculated based on the spectral density of the abnormal peak. Also by the above-described method, it is possible to output the first evaluation value for accurately determining the presence or absence of abnormality of the speed increasing device 40 which is the object.

<Modification 1>
In the above description, the map generation unit 104 generates two anomaly maps for one target peak. However, the number of anomaly maps generated for one target peak may be one, or three or more.

  Depending on the target of state monitoring, an anomaly according to one type of anomaly model (for example, a first anomaly model or a second anomaly model) may be dominant. In this case, the map generation unit 104 may generate only one anomaly map (for example, the first anomaly map or the second anomaly map) for one target peak. The database 107 may store only abnormal part information including model information for identifying an abnormal model used when generating the abnormal map. Therefore, the abnormal part estimation unit 108 indicates, from the database 107, a second threshold value less than the first evaluation value corresponding to the target map, and a fundamental frequency whose frequency difference with the target peak corresponding to the target map is within a predetermined value. The abnormal site information indicating the above may be extracted as identification information.

  Alternatively, depending on the target of the state monitoring, in addition to the first and second anomaly models described above, another anomaly model's anomaly may occur. For example, if the bearing is heavily worn, the PSD may increase in a predetermined frequency band. The frequency band in which the PSD increases varies depending on the specifications of the bearing, the location where wear is occurring, and the like. Hereinafter, an abnormal model in which the PSD increases in a predetermined frequency band is referred to as a “third abnormal model”.

  When there is a possibility that the object may generate an abnormality according to the third abnormality model, the map generation unit 104 stores in advance a predetermined frequency band corresponding to the third abnormality model. When any of the peaks detected by the peak detection unit 103 is included in the predetermined frequency band of the third abnormality model, the map generation unit 104 sets the frequency in the predetermined frequency band as an abnormality component. Generate a map.

The map generation unit 104 generates a third anomaly map M (3) as shown in the following equation (6).
M (3) = [· · · · · · · · · · · · formula (6)
In the third abnormality map M (3), the value corresponding to the frequency that is the abnormal component is set to 1, and the value corresponding to the frequency that is not the abnormal component is set to 0. The map generation unit 104 may generate a plurality of third anomaly maps different in frequency band in which the PSD increases. For example, the map generation unit 104 may generate x types of third anomaly maps of M_1 (3) to M_x (3). Each of M_1 (3) to M_x (3) is represented by the same equation as the above equation (6).

  FIG. 17 is a diagram showing a relationship between a detected peak, a third abnormality map, and an abnormal peak. Referring to FIG. 17, abnormal peak extracting section 105 selects a predetermined frequency band (band of frequencies fa to fb) shown in third abnormal map M (3) among the peaks detected by peak detecting section 103. The peaks contained in are extracted as abnormal peaks.

  Even when the map generation unit 104 generates the first to third abnormality maps, the data processing device 80 performs processing in accordance with the flowchart shown in FIG. However, the frequency of the abnormal component of the third abnormality map is predetermined and does not depend on the frequency of the peak detected by the peak detection unit 103. Therefore, when a plurality of peaks having frequencies within the predetermined frequency band of the third abnormal model are detected in step S3, the data processing device 80 sets any one of the plurality of peaks as a target peak. The third anomaly map may be generated only in step S4. Only in step S6 following step S4, the first evaluation value for the third abnormality map is calculated.

  Furthermore, in the case where the object may generate an abnormality according to the third abnormality model, the database 107 stores abnormal part information including model information indicating the third abnormality model.

  FIG. 18 shows the calculation result of the first evaluation value, the abnormal part information stored in the database 107, and the estimation result information when there is a possibility that the object may generate an abnormality according to the first to third abnormality models. And FIG. 18 (a) shows an example of the calculation result of the first evaluation value, FIG. 18 (b) shows an example of the abnormal part information, and FIG. 18 (c) shows an example of the estimation result information. Be

  In the example shown to Fig.18 (a), the map production | generation part 104 produces | generates the 3rd abnormality map which makes 100-200 Hz an abnormality component, and the 3rd abnormality map which makes 200-500 Hz an abnormality component. The first evaluation value calculation unit 106 calculates a first evaluation value “600” for a third abnormality map having 100 to 200 Hz as an abnormal component, and for a third abnormality map having 200 to 500 Hz as an abnormal component. A first evaluation value "150" is calculated.

  The database 107 stores abnormal part information corresponding to the third abnormal model in which the predetermined frequency band is 100 to 200 Hz and abnormal part information corresponding to the third abnormal model in which the predetermined frequency band is 200 to 500 Hz. It is done.

  In the example shown in FIG. 18, the first evaluation value “600” calculated for the third abnormality map having 100 to 200 Hz as the abnormal component corresponds to the third abnormality model in which the predetermined frequency band is 100 to 200 Hz. The second threshold “300” of the abnormal part information to be Therefore, as shown in FIG. 18C, the abnormal part estimation unit 108 identifies the bearings identified by the part information of the abnormal part information corresponding to the third abnormality model in which the predetermined frequency band is 100 to 200 Hz, Estimated as a part where wear abnormality has occurred.

<Modification 2>
In the above description, the peak detection unit 103 sets the first threshold which is a constant value. However, the peak detection unit 103 may set a first threshold that fluctuates according to the frequency based on the frequency spectrum. For example, peak detection section 103 compares the first threshold to compare the PSD of unit data of a certain frequency with the PSD of unit data having a frequency within a frequency band including that frequency (for example, a band within that frequency ± 50 Hz). Set to 10 times the median.

  FIG. 19 is a diagram illustrating an example of a peak detected using a first threshold that varies with frequency. As shown in FIG. 19, even when the noise level fluctuates according to the frequency, the first threshold fluctuates according to the noise level, so that false detection of noise as a peak can be prevented.

<Modification 3>
Generally, the greater the degree of anomaly, the greater the number of harmonics of the anomalous peak that appear in the frequency spectrum. However, the higher the order, the smaller the PSD of the harmonics. Therefore, the PSDs of the harmonics may be weighted according to the order so that the higher harmonics are not underestimated.

  For example, when generating the first abnormality map and the second abnormality map, the map generation unit 104 sets a weighting factor for each abnormal component. Specifically, the map generation unit 104 sets a weighting coefficient that is larger as the order is higher for the harmonic and the abnormal component that is the frequency of the sideband wave thereof. For example, the map generation unit 104 sets a weighting factor Wp = p for the harmonic of the order p (p th harmonic) and an abnormal component that is the frequency of the sideband wave. The map generation unit 104 sets the weighting factor W1 = 1 for the fundamental wave and the abnormal component that is the frequency of its sideband wave.

FIG. 20 is a diagram showing an example of the abnormal peak extracted using the second abnormality map. In Figure 20, the abnormal peak frequency _fex ¹ 0, PSD is psdex ₁ ⁰ is a peak corresponding to the abnormal component of the basic wave of the second abnormality map. The anomalous peak whose frequency is fex _p ⁰ and PSD is psdex _p ⁰ is a peak corresponding to the anomalous component indicating the p-th harmonic in the second anomalous map. An abnormal peak whose frequency is fex ₁ ^b and PSD is psdex ₁ ^b (b = ..., -2, -1, ₁ , 2, ...) is a sideband wave of the fundamental wave in the second anomalous map It is a peak corresponding to the anomalous component shown. An abnormal peak whose frequency is fex _p ^b and PSD is psdex _p ^b (b = ..., -2, -1, 1, 2, ...) is the side band of the p th harmonic in the second anomalous map It is a peak corresponding to an abnormal component showing a wave.

  As shown in FIG. 20, the PSDs of the p-th harmonic and its sidebands are smaller than the PSDs of the fundamental wave and their sidebands. Therefore, the first evaluation value calculation unit 106 may calculate the first evaluation value E for the second abnormality map according to the following equations (7) and (8).

  Note that the first evaluation value calculation unit 106 may calculate the first evaluation value according to Expressions (7) and (8) with b = 0 in Expression (8) for the first abnormality map in which there is no sideband wave.

<Modification 4>
In the above description, the state monitoring sensor 70 measures vibration waveform data. However, the condition monitoring sensor 70 is an AE (Acoustic Emission) sensor, and may measure waveform data in the ultrasonic region. Alternatively, the state monitoring sensor 70 may measure waveform data of sounds outside the ultrasonic region. In addition, the state monitoring sensor 70 may be a sensor that measures waveform data such as shaft rotation speed, load torque, and motor power. Since these waveform data also change according to the abnormality of the object, the presence or absence of the abnormality can be determined based on the waveform data.

<Modification 5>
In the above description, the data processing device 80 is installed inside the wind turbine 10. However, the data processing device 80 may be installed outside the wind turbine generator 10, receive the vibration waveform data measured by the state monitoring sensor 70 by a wired or wireless communication system, and execute the above data processing.

<Modification 6>
The database 107 may be installed outside the data processing device 80 instead of inside the data processing device 80, and can communicate with the data processing device 80 by a wired or wireless communication system. In this case, the abnormal part estimation unit 108 may access the database 107 via the communication system.

  Furthermore, the data processing device 80 may not include the abnormal part estimation unit 108, the database 107, the second evaluation value calculation unit 109, and the output processing unit 110. In this case, the data processing device 80 may output the first evaluation value to an external device including the abnormal part estimation unit 108, the database 107, the second evaluation value calculation unit 109, and the output processing unit 110. Alternatively, the user may check the first evaluation value calculated by the data processing device 80 to determine the presence or absence of abnormality in the speed increasing device 40.

<Modification 7>
In the above description, the second evaluation value calculation unit 109 does not extract the first evaluation value of the target map from which the identification information has been extracted by the abnormal part estimation unit 108 and the identification information, and the corresponding first evaluation value The sum total with the 1st evaluation value of the object map which exceeds 3rd threshold value shall be computed as the 2nd evaluation value. However, the second evaluation value calculation unit 109 may calculate, as the second evaluation value, the sum of the first evaluation values of the target map from which the identification information has been extracted by the abnormal part estimation unit 108. This configuration is effective in the case of an object having a low possibility of occurrence of an abnormality whose cause is unknown.

<Modification 8>
The output processing unit 110 may display a screen indicating the estimation result information generated by the abnormal part estimation unit 108 on the display unit (display device) of the terminal device.

  FIG. 21 is a schematic block diagram showing the configuration of the terminal device. The data processing device 80, the condition monitoring sensor 70, and the terminal device 95 shown in FIG. 21 constitute a condition monitoring system. The terminal device 95 includes a CPU, a ROM for storing processing programs and the like, a RAM for temporarily storing data, and further includes an input / output port for inputting / outputting various signals (all not shown). . The terminal device 95 executes various data processing in accordance with the program stored in the ROM. As shown in FIG. 21, the terminal device 95 includes a storage unit 96, a display unit 97, and a selection unit 98.

  The storage unit 96 stores the frequency spectrum acquired from the data processing device 80, estimation result information, and the like. Display unit 97 is, for example, a touch panel or a liquid crystal display. The selection unit 98 switches the screen displayed on the display unit 97 according to an input instruction from the user.

  FIG. 22 is a diagram showing an example of a screen displayed on the display unit 97 of the terminal device 95. As shown in FIG. As shown in FIG. 22, the selection unit 98 causes the display unit 97 to display a table indicating estimation result information and a graph indicating at least a part of the frequency spectrum obtained by the frequency analysis unit 102. Thus, the user (worker) can confirm the analysis result obtained from the vibration waveform measured by the state monitoring sensor 70.

  As shown in FIG. 22, the selection unit 98 displays the portion of the abnormal peak corresponding to the abnormal map selected from the plurality of abnormal maps in the frequency spectrum in a display format different from the remaining portion. In the example shown in FIG. 22, the selection unit 98 displays the selected one anomaly map with white triangle marks, and displays the selected another anomaly map with black circles. . As a result, the abnormal peak corresponding to the selected abnormal map is emphasized, and the user can easily grasp the size and the like of the abnormal peak corresponding to the selected abnormal map.

  The selection unit 98 receives an instruction to select an abnormality map from the user. For example, the selection unit 98 can receive the first evaluation value, the part information, and the model information in a drop-down or the like, and selects the abnormality map in accordance with the input instruction. For example, the selection unit 98 selects an anomaly map in which the first evaluation value is 500 or more, selects an anomaly map in which the part information is "unknown", or selects an anomaly map in which the model information is "1". .

  Alternatively, the selection unit 98 may automatically select the anomaly map. For example, the selection unit 98 automatically selects the anomaly map having the highest first evaluation value. Alternatively, the selection unit 98 may automatically select an abnormality map of specific model information.

  In the example shown in FIG. 22, the display format is made different by marking the selected anomaly map. However, the method of changing the display format is not limited to this. For example, the color, the thickness of the line, the line type (such as a broken line or a dotted line) may be made different between the portion of the abnormal peak corresponding to the selected abnormal map and the remaining portion.

  FIG. 23 is a view showing another example of the screen displayed on the display unit 97 of the terminal device 95. As shown in FIG. As shown in FIG. 23, the selection unit 98 may remove anomalous peaks corresponding to the anomalous map selected from among the plurality of anomalous maps in the frequency spectrum. In the example shown in FIG. 23, the abnormal peak corresponding to the abnormal map with model information “1” is removed.

  FIG. 24 is a diagram showing still another example of the screen displayed on the display unit 97 of the terminal device 95. As shown in FIG. As illustrated in FIG. 24, the selection unit 98 may display only abnormal peaks corresponding to the abnormal map selected from among the plurality of abnormal maps in the frequency spectrum. In the example illustrated in FIG. 24, the selection unit 98 displays only the abnormal peak corresponding to the abnormal map whose model information is "1". That is, among the frequency spectrum, models other than the abnormal peak corresponding to the abnormal map of “1” are removed.

  According to the screens shown in FIGS. 22-24, even if the harmonics and sidebands of a plurality of anomaly maps are complexly intricate, the anomaly peaks corresponding to the selected anomaly map are emphasized or eliminated. it can. Thus, the user can easily determine the presence or absence of an abnormality.

  The selection unit 98 makes the display format of the specific abnormal peak different, removes the specific abnormal peak, or removes other than the specific abnormal peak as follows.

  FIG. 25 is a diagram showing a frequency spectrum near the abnormal peak. The selection unit 98 specifies unit data Pa having the maximum frequency among unit data whose frequency is lower than the abnormal peak corresponding to the selected abnormal map and in which the PSD has a local minimum value. Furthermore, the selecting unit 98 specifies unit data Pb having the lowest frequency among unit data whose frequency is higher than the abnormal peak corresponding to the selected abnormal map and in which the PSD has a local minimum value.

  FIG. 26 is a diagram illustrating an example of a method of removing an abnormal peak. FIG. 27 is a diagram showing another example of removing an abnormal peak. As shown in FIG. 26, the selecting unit 98 eliminates anomalous peak by setting the PSD between the specified unit data Pa and Pb to zero. Alternatively, as shown in FIG. 27, the selecting unit 98 may remove the abnormal peak by connecting the specified unit data Pa and Pb with a line segment.

  FIG. 28 is a diagram showing an example of a method for removing other than the abnormal peak. As shown in FIG. 28, the selection unit 98 sets the PSD of the region other than the area between the specified unit data Pa and the unit data Pb to zero.

  It should be understood that the embodiments disclosed herein are illustrative and non-restrictive in every respect. The scope of the present invention is indicated not by the description of the embodiment described above but by the scope of the claims, and is intended to include all modifications within the meaning and scope equivalent to the scope of the claims.

  DESCRIPTION OF SYMBOLS 10 wind power generator, 20 spindles, 25 hubs, 30 blades, 40 speed-up machines, 50 generators, bearings for 60 spindles, 70 state monitoring sensors, 80 data processing devices, 90 nacelle, 95 terminal devices, 96 storage units, 97 Display unit 98 selection unit 100 tower 101 filter unit 102 frequency analysis unit 103 peak detection unit 104 map generation unit 105 abnormal peak extraction unit 106 first evaluation value calculation unit 107 database 108 abnormal part estimation Part, 109 Second evaluation value calculation part, 110 Output processing part.

Claims (18)

A condition monitoring device for monitoring the condition of an object, comprising:
A peak detection unit that detects a peak from a frequency spectrum obtained by performing frequency analysis on waveform data measured by a sensor installed on the object;
And a map generator for generating at least one anomaly map for the frequency spectrum,
The at least one anomaly map includes the frequency of one target peak selected from among the detected peaks, and the target peak when it is assumed that the target peak is a peak due to an abnormality of the target. Containing the frequency of the peak that appears with
The condition monitoring device
One target map is selected from the at least one abnormal map, and a peak having a frequency difference with any of the abnormal components included in the target map within a predetermined value is extracted as an abnormal peak from the detected peaks. An abnormal peak extraction unit,
And a first evaluation value calculator configured to calculate a first evaluation value indicating a degree of occurrence of an abnormality corresponding to the target map based on a spectral density of the abnormal peak. The frequency spectrum is indicated by a data array in which unit data in which a frequency and a spectral density at the frequency are associated are sequentially arranged according to the frequency,
The state monitoring device according to claim 1, wherein the peak detection unit detects, as a peak, unit data having a spectral density that indicates a maximum value and exceeds a first threshold value from the data array.   The state monitoring device according to claim 2, wherein the peak detection unit sets the first threshold that is a constant value regardless of a frequency based on the frequency spectrum.   The state monitoring device according to claim 2, wherein the peak detection unit sets the first threshold that fluctuates according to a frequency based on the frequency spectrum.   The at least one anomaly map has a frequency of the target peak as a fundamental frequency, and a first anomaly map in which the fundamental frequency and the frequency of the harmonic of the fundamental wave having the fundamental frequency are anomalous components; A second anomaly map in which the frequency of the sideband wave of the fundamental wave, the frequency of the harmonic wave, and the frequency of the sideband wave of the harmonic wave are included in a predetermined frequency band including the frequency of the target peak The condition monitoring device according to any one of claims 1 to 4, including at least one of a third anomaly map in which the frequency is an anomaly component. The at least one anomaly map is information in which the frequency is associated with a value of 0 or 1 for each frequency, and a value corresponding to a frequency that is an anomaly component is set to 1 and a frequency that is not an anomaly component. The corresponding value is set to 0
The abnormal peak extraction unit extracts the abnormal peak by ANDing the unit data detected as a peak by the peak detection unit and the target map to mask a peak of a frequency that is not an abnormal component. The state monitoring device according to Item 2.   The state monitoring device according to any one of claims 1 to 6, wherein the first evaluation value is a sum of spectral densities of the abnormal peak.   The first evaluation value is a sum of multiplication values obtained by multiplying the spectral density of the abnormal peak by a weighting factor according to the frequency of the abnormal peak. State monitoring device as described.   From the database that stores abnormal part information that associates the part information identifying the part, the frequency caused by the abnormality of the part, and the second threshold for each part of the target in the target map The abnormal region information indicating the frequency within the predetermined value and the frequency difference with the corresponding target peak indicates the second threshold value less than the first evaluation value corresponding to the target map is extracted as identification information, and the identification is performed. The state monitoring according to any one of claims 1 to 8, further comprising: an abnormal part estimating unit that generates first estimation result information indicating that an abnormality occurs in a part identified by part information of information. apparatus. The at least one anomaly map includes two or more anomaly maps generated for the target peak based on two or more anomaly models in which the appearance of an anomaly component is different from each other,
The state monitoring apparatus further corresponds, for each part of the object, model information for identifying an abnormal model, part information for identifying the part, a frequency caused by an abnormality for the part, and a second threshold. Shows the model information for identifying the abnormal model corresponding to the target map from the database storing the attached abnormal part information, and the frequency difference with the target peak corresponding to the target map indicates the frequency within the predetermined value And, abnormal part information indicating the second threshold value less than the first evaluation value corresponding to the target map is extracted as identification information, and abnormality occurs in the part identified by the part information of the identification information The state monitoring device according to any one of claims 1 to 8, further comprising an abnormal part estimation unit that generates first estimation result information indicating.   The abnormal part estimation unit can not extract the identification information for the target map, and when the first evaluation value corresponding to the target map exceeds a third threshold, an unregistered abnormality occurs in the database. The condition monitoring device according to claim 9 or 10, further comprising: generating second estimated result information indicating that The at least one anomaly map comprises a plurality of anomaly maps,
The first evaluation value calculation unit sequentially selects each of the plurality of abnormality maps as the target map, and calculates the first evaluation value for each of the plurality of abnormality maps.
The state monitoring apparatus further includes a second evaluation value calculation unit configured to calculate, as a second evaluation value, a sum of the first evaluation values of the target map for which the identification information has been extracted by the abnormal part estimation unit. The condition monitoring device according to 9 or 10. The at least one anomaly map comprises a plurality of anomaly maps,
The first evaluation value calculation unit sequentially selects each of the plurality of abnormality maps as the target map, and calculates the first evaluation value for each of the plurality of abnormality maps.
In the state monitoring apparatus, the first evaluation value of the target map from which the identification information is extracted by the abnormal part estimation unit and the identification information are not extracted, and the corresponding first evaluation value is the third. The state monitoring device according to claim 11, further comprising: a second evaluation value calculation unit configured to calculate, as a second evaluation value, a sum of the target map exceeding the threshold value and the first evaluation value. A state monitoring system comprising the state monitoring device according to any one of claims 1 to 13 and a terminal device,
The state monitoring system, wherein the terminal device displays a graph indicating the frequency spectrum on a display unit.   The state monitoring according to claim 14, wherein the terminal device displays a portion of the abnormal peak corresponding to the abnormal map selected from the at least one abnormal map in the frequency spectrum in a display format different from the remaining portion. system.   The state monitoring system according to claim 14, wherein the terminal device removes the abnormal peak corresponding to the abnormal map selected from the at least one abnormal map among the frequency spectra.   The state monitoring system according to claim 14, wherein the terminal device displays only the abnormal peak corresponding to the abnormal map selected from the at least one abnormal map among the frequency spectra. A state monitoring method for monitoring the state of an object, comprising:
Detecting a peak from a frequency spectrum obtained by performing frequency analysis on waveform data measured by a sensor installed on the object;
Generating at least one anomaly map for the frequency spectrum;
The at least one anomaly map includes the frequency of one target peak selected from among the detected peaks, and the target peak when it is assumed that the target peak is a peak due to an abnormality of the target. Containing the frequency of the peak that appears with
The state monitoring method is
One target map is selected from the at least one abnormal map, and a peak having a frequency difference with any of the abnormal components included in the target map within a predetermined value is extracted as an abnormal peak from the detected peaks. Step and
Calculating a first evaluation value indicating the degree of occurrence of an abnormality corresponding to the target map based on the spectral density of the abnormal peak.

Images