JP2008164578A - Physical quantity detection device and rolling device - Google Patents

Abstract

The present invention provides a physical quantity detection device and a rolling device that can suppress a decrease in accuracy of abnormality diagnosis caused by a rapidly changing environment or an environment where noise or vibration noise is large.
In an abnormality diagnosis of a rolling device incorporated in a railway vehicle, a steel rolling mill, a machine tool, or the like, a detection vibration sensor and a reference vibration sensor are provided, and the detection vibration sensor and the reference vibration sensor are provided. These signals are digitally converted by an AD converter (ADC) 172, and the DSP circuit 173 estimates noise using an optimum filter based on the digital output signal, and subtracts the amount to output the signal. Perform abnormality diagnosis.
[Selection] Figure 4

Description

  The present invention relates to a physical quantity detection device attached to a rolling device (such as an axle bearing, a gear, a wheel, or a bearing) used in a machine device such as a railway vehicle or a machine tool for abnormality diagnosis, and in particular, an abnormality of the rolling device. The present invention relates to a physical quantity detection device configured to be able to remove noise generated during diagnosis and a rolling device provided with the physical quantity detection device.

  2. Description of the Related Art Conventionally, a rolling device incorporated in a machine device such as a railway vehicle or a machine tool is periodically inspected for abnormalities such as damage and wear after the rolling bearing and other rotating parts are used for a certain period. In this inspection, various devices have been proposed that can attach a physical quantity detection device such as a vibration sensor and diagnose the abnormality of the rolling device in an actual operation state without disassembling the mechanical device incorporating the rolling device. (For example, refer to Patent Documents 1 to 3). As a typical example, as described in Patent Document 1, for example, a vibration sensor is installed in a bearing portion and measured, and further, an FFT (Fast Fourier Transform) process is performed on the signal to generate a vibration generation frequency. A method for performing diagnosis by extracting component signals is known.

  Further, when performing these abnormality diagnoses, in order to remove noise that appears in the signal output from the vibration sensor, it is generally performed to apply a band-pass filter to improve the signal-to-noise ratio of the signal. (For example, refer to Patent Document 4).

  However, when using a band-pass filter, it is necessary to grasp the noise frequency band in advance. For example, the frequency band of noise is constantly being affected by factors such as driving conditions, traveling road conditions, and load conditions as in a railway vehicle. It may fluctuate and affect the accuracy of abnormality diagnosis in a complicated manner. In addition, the level of noise or electromagnetic noise is often high and in a high range in a machine tool, and the vibration sensor also has a high band to detect peeling or wear of a rolling device incorporated in the machine tool. Therefore, it is necessary to widen the band of the band pass filter, and in fact, even if the electromagnetic noise expressed as a high frequency can be removed, noise noise should be sufficiently removed. Was difficult.

  On the other hand, there is one that uses two sensors (pickups) without using a bandpass filter and corrects the signals detected by each sensor so that the phase difference is zero to remove noise (for example, And Patent Documents 5 and 6).

JP 2002-22617 A JP 2003-202276 A Japanese Patent Laid-Open No. 2004-257836 JP-A-9-113416 JP-A-6-288806 JP-A-9-33310

  However, those described in Patent Documents 5 and 6 do not consider dynamic characteristics (transmission characteristics in signal processing) such as rigidity and viscoelasticity that are interposed between sensors, and are not subject to abnormality diagnosis. This may affect the signal to be used, and there is room for improvement in improving the accuracy of abnormality diagnosis.

  The present invention has been made in view of the above-described circumstances, and an object of the present invention is to detect a physical quantity that can suppress a decrease in accuracy of abnormality diagnosis caused by a rapidly changing environment or an environment where noise or vibration is large. It is in providing a device and a rolling device.

The object of the present invention is achieved by the following constitution.
(1) a detection sensor for detecting any physical quantity of vibration, sound, ultrasonic wave, stress, displacement, or strain emitted from a rolling device incorporated in a mechanical device;
A reference sensor disposed at a predetermined position to detect the physical quantity;
A signal processing unit that takes in a signal from the detection sensor and a signal from the reference sensor and applies an optimal filter to remove a noise signal generated from other than the rolling device and estimate the physical quantity;
A physical quantity detection device comprising:
(2) The signal processing unit
further,
A diagnosis processing unit that performs frequency diagnosis of the rolling device based on the estimated physical quantity to perform abnormality diagnosis of the rolling device;
The physical quantity detection device according to (1) above.
(3) The physical quantity detection device according to (1) or (2), wherein the reference sensor is similarly attached to a member to which the detection sensor is attached.
(4) The reference sensor is disposed in the signal processing unit.
The physical quantity detection device according to (1) or (2) above.
(5) The reference sensor is disposed so as to detect the physical quantity in the same direction as the detection sensor.
The physical quantity detection device according to any one of (1) to (4) above.
(6) The optimum filter is at least one of a Kalman filter or a Wiener filter.
The physical quantity detection device according to any one of (1) to (5) above.
(7) The optimum filter is at least one of an LMS algorithm and an RLS algorithm.
The physical quantity detection device according to any one of (1) to (5) above.
(8) A rolling device to which the physical quantity detection device according to any one of (1) to (7) is attached.

  According to the present invention, it is possible to detect the physical quantity from the signal output by taking the signals of both the detection sensor and the reference sensor, estimating the noise using the optimum filter, and subtracting that amount. Since abnormality diagnosis can be performed based on the detection result, it is possible to suppress a decrease in accuracy of physical quantity detection and abnormality diagnosis which is caused by a rapidly changing environment or an environment where noise or vibration noise is large.

  In the present invention, it is preferable that the reference sensor is also attached to a member to which the detection sensor is attached (for example, a rolling device housing in the case of a railway vehicle). As a result, the accuracy of noise estimation can be further improved, and rapid physical quantity detection and abnormality diagnosis can be performed.

  Hereinafter, first to third embodiments according to the present invention will be described in detail with reference to the drawings.

FIG. 1A is a schematic plan view of a railway vehicle equipped with a rolling device provided with the physical quantity detection device according to the first embodiment of the present invention, and FIG. 1B is a schematic side view of the railway vehicle. 2 is a cross-sectional view for explaining the rolling device of the first embodiment, FIG. 3 is a block configuration diagram of a signal processing device of the physical quantity detection device, and FIG. 4 is a diagram of a filter processing unit of the signal processing device. FIG. 5 is a block diagram showing the operation of the filter processing unit, FIG. 6 is a block diagram showing another operation of the filter processing unit, and FIG. 7 is related to the second embodiment of the present invention. FIG. 8 is a schematic diagram of a steel rolling machine equipped with a physical quantity detection device, FIG. 8 is a schematic diagram of a machine tool equipped with a drill that is a cutting tool and equipped with a physical quantity detection device according to a third embodiment of the present invention; Fig. 9 shows the output signal of the detection sensor and the optimal filter processing Is a diagram showing a frequency spectrum comparison of the force signal, with (a) the rotational speed of the rolling device is 360rmp For ^{(min -1),} (b) the rotational speed of the rolling device is 88rpm is ^{(min -1)} FIG. 10 is a diagram showing another method of attaching the vibration sensor according to the first embodiment, and FIG. 11 is a block configuration diagram showing a modification of the signal processing unit according to the first embodiment.

(First embodiment)
First, a railway vehicle incorporating a rolling device (rotation support device) provided with a physical quantity detection device according to a first embodiment of the present invention will be described with reference to FIGS.

As shown in FIG. 1, one rail vehicle 100 is supported by two front and rear chassis, and four wheels 101 are attached to each chassis. The rotation support device (rolling device) 110 of each wheel 101 includes a detection vibration sensor (for example, vibration, ultrasonic wave, stress, displacement, distortion, or angular velocity) emitted from the rotation support device 110. A detection sensor (111) and a reference vibration sensor (reference sensor) 117 are provided.
The detection vibration sensor 111 detects vibration generated from the rotation support device 110 during operation in order to perform an abnormality diagnosis to be described later, and the reference vibration sensor 117 is also detected from the rotation support device 110. It is used to detect the generated vibration and remove noise contained in the detection signal detected by the detection vibration sensor 111.

The control panel 115 of the railway vehicle 100 takes in sensor signals output from the detection vibration sensor 111 for four channels and the reference vibration sensor 117 for four channels simultaneously (substantially simultaneously), and diagnoses an abnormality. Two signal processing devices (signal processing units) 150 for performing signal processing are installed. That is, the output signals of the eight vibration sensors 111 and 117 provided in each chassis are continuously input to different signal processing devices 150 for each chassis via signal lines (wired) 116. The signal processing device 150 also receives a rotation speed pulse signal from a rotation speed sensor (not shown) that detects the rotation speed of the wheel 101. Note that, instead of the signal line 116 being wired, wireless signal transmission may be performed.
In this manner, the physical quantity detection device 50 includes the detection vibration sensor 111, the reference vibration sensor 117, and the signal processing device 150, and is attached to the rotation support device 110.

  As shown in FIG. 2, the physical quantity detection device 50 according to the present invention, that is, the rotation support device 110 of the railway vehicle 100 including the vibration sensors 111 and 117 rotates the axle 213 connected to the wheels 101 at both ends. In order to support freely, a double row tapered roller bearing 211 that is a rotating part and a bearing box 212 that is a stationary member that constitutes a part of the carriage of the railway vehicle 100 are provided.

  The double row tapered roller bearing 211 rotatably supports an axle 213 of a railway vehicle, which is a rotating shaft, and has a pair of inner rings 214 and 214 having inner ring raceway surfaces 215 and 215 inclined like a cone outer surface on the outer circumferential surface. A single outer ring 216 having a pair of outer ring raceways 217 and 17 inclined like a conical inner surface on the inner peripheral surface, inner ring raceways 215 and 215 of the inner rings 214 and 214, and outer ring raceways 217 and 217 of the outer ring 216, Tapered rollers 218 and 218 that are a plurality of rolling elements arranged in a double row between them, annular punching cages 219 and 219 that hold the tapered rollers 218 and 218 in a freely rolling manner, and both axial ends of the outer ring 216 And a pair of seal members 220 and 220 respectively attached to the parts.

  The bearing box 212 includes a housing 221 that constitutes a side frame of the bogie of the railway vehicle. The housing 221 is formed in a cylindrical shape having a flange portion at an intermediate portion so as to cover the outer peripheral surface of the outer ring 216. ing. In addition, the reference vibration sensor 117 is attached to the outside of the housing 221 on the outer end side in the axial direction (left direction in the drawing) in order to capture noise generated from the periphery. A vibration sensor 111 for detection is attached to the outside so as to capture the radial vibration of the double row tapered roller bearing 11. Both vibration sensors 111 and 117 are arranged so as to be held in a posture capable of detecting vibration acceleration in the same direction, that is, in the direction of gravity, and are attached to the same member (housing 221).

  An inner ring spacer 224 is disposed between the pair of inner rings 214 and 214. An axle 213 is press-fitted into the pair of inner rings 214, 214 and the inner ring spacer 224, and the outer ring 216 is fitted in the housing 221. The double-row tapered roller bearing 211 is loaded with a radial load due to the weight of various members and an arbitrary axial load, and the upper side in the circumferential direction of the outer ring 216 is loaded in the load zone (the load is applied to the rolling elements). Area).

  One seal member 220 disposed on the front end side of the axle shaft 213 is assembled at an axial end portion of the outer ring 216 so as to close a gap space defined by the inner ring 214 and the outer ring 216.

The vibration sensors 111 and 117 are vibration measuring elements such as piezoelectric elements, and the inner and outer ring raceway surfaces 215, 215, 217, and 217 of the double row tapered roller bearing 211 are peeled off, the gears are lost, the wheels are worn flat, and the like. Used to detect.
For the detection vibration sensor 111, various sensors such as an acceleration sensor, an AE (acoustic emission) sensor, an ultrasonic sensor, a shock pulse sensor, and the like can be adopted. In consideration of easiness, a MEMS acceleration sensor is preferable. The vibration sensors 111 and 117 are composite sensors in which a vibration system sensor capable of detecting at least one vibration of an AE (Acoustic Emission) sensor, an acoustic sensor, and an ultrasonic sensor and a temperature sensor are integrally housed in a housing. You may comprise as.

  As shown in FIG. 3, the signal processing device 150 includes a sensor signal processing unit 150A and an abnormality diagnosis processing unit (MPU) 150B. The sensor signal processing unit 150A includes one filter processing unit 151 and one waveform shaping circuit 155, and the output signals of the four vibration sensors 111 and the output signals of the four reference vibration sensors 117 are collectively displayed. Then, a rotational speed signal such as a speed proportional sine wave or a rotational speed pulse signal from the rotational speed sensor is input to the filter processing unit 151 and the waveform shaping circuit 155, respectively. The filter processing unit 151 has the function of an analog amplifier and the function of an anti-aliasing filter, and appropriately amplifies and filters the input 8-channel output signal (analog signal) to convert it into a digital signal. The noise removal calculation described later is performed. These calculated 8-channel digital signals are taken into the abnormality diagnosis processing unit (MPU) 150B.

On the other hand, after the rotation speed signal is shaped by the waveform shaping circuit 155, the number of pulses per unit time is counted by a timer counter (not shown), and the value is sent to the abnormality diagnosis processing unit (MPU) 150B as the rotation speed signal. Entered. The abnormality diagnosis processing unit (MPU) 150B diagnoses an abnormality based on a vibration waveform signal (estimated signal S _n ′ described later) detected by the vibration sensors 111 and 117 and a rotation speed signal detected by the rotation speed sensor. Execute. The diagnosis result by the abnormality diagnosis processing unit (MPU) 150B is output to the communication line 120 (see FIG. 1) via the line driver (LD) 156. The communication line 120 is connected to an alarm device, and an appropriate alarm operation is performed when an abnormality such as a flatness of the wheel 101 occurs.

  The signal processing device 150 includes a static random access memory (SRAM) 162 having a backup battery (Batt) 161 in order to retain data even when power is not supplied.

  The abnormality diagnosis processing unit 150B performs frequency analysis of vibration generated in the double-row tapered roller bearing 211 in the rotation support device 110 based on output signals from the filter processing unit 151 and the waveform shaping circuit 155. Specifically, after an absolute value process is performed on the output signal of the filter processing unit 151, a frequency spectrum of vibration is calculated based on an FFT algorithm. Note that as preprocessing for performing FFT, envelope processing for obtaining an envelope of a vibration signal may be performed.

  In general, the abnormal frequency band of vibration caused by the rotation of the double row tapered roller bearing 211 is determined depending on the bearing specifications such as the size of the bearing and the number of rolling elements. For example, Table 1 shows the relationship between defects in each member of an axle bearing such as the inner ring 214, the outer ring 216, the tapered roller (rolling element) 218, and the punched cage (retainer) 219 and the abnormal vibration frequency generated in each member. It is shown. Therefore, the vibration processing unit 150B calculates a frequency spectrum from the output signal of the filter processing unit 151 described above, and on the basis of the rotational speed signal and the bearing specifications based on the arithmetic expression shown in Table 1, a double row cone is obtained. The abnormal vibration frequency in each member of the roller bearing 211 is calculated as a reference value and stored in a storage medium such as a RAM.

  Then, the abnormality diagnosis processing unit 150B compares the calculation result of the frequency spectrum with the reference value to determine whether abnormal vibration has occurred. For this abnormality diagnosis, for example, the peak frequency is calculated from the calculation result of the frequency spectrum, and the determination is made from the degree of coincidence and the number of coincidences with the reference value. The determination may be made from the number of times, and in this case, it is not necessary to detect the rotation speed signal.

Next, the filter processing unit 151 will be described in detail with reference to FIG.
As shown in FIG. 4, the filter processing unit 151 includes a multiplexer (MUX) 171 and an AD converter (ADC) 172 with a DSP circuit 173 as a center. The input 8-channel analog signal is switched to a signal for each channel by a multiplexer (MUX) 171 and converted into a digital signal by an AD converter (ADC) 172. After the converted signals are temporarily stored in a storage medium such as a RAM, the DSP circuit 173 appropriately takes out these digital signals from the storage medium, detects the digital signals output from the reference vibration sensor 117, and detects them. An optimal filter is applied based on the digital signal output from the vibration sensor 111 and noise is removed, and then the digital signal is output to the diagnostic processing unit 150B. Note that the processing of the optimum filter is performed for each combination unit of vibration sensors 111 and 117 attached to the same rotation support device 110 (for example, every dotted frame in FIG. 4).

Further, a basic configuration of the optimum filter processing (noise removal calculation) will be described with reference to FIG.
In the detection vibration sensor 111, in each of the double-row tapered roller bearings 211, a noise signal x _n (where n = 1, 2, 3, 4, and so on) generated from a specific noise source is, for example, the noise. It is detected as a noise signal H _n · x _n multiplied by an unknown transfer function H _n such as a viscoelastic characteristic interposed between the source and the detection vibration sensor 111. Therefore, a desired signal S _n output signals d _n should be subject to the original diagnosis of each detecting vibration sensor 111, and the noise signal H _n · x _n is detected as added signals. Here, the reference vibration sensor 117 is also attached to a location where the noise _xn generated from the noise source can be separately detected, that is, the member to which the detection vibration sensor 111 is attached in the present embodiment.

For this reason, the filter processing unit 151 takes in the output signals of the four reference vibration sensors 117 and the output signals of the four detection vibration sensors 111 and is a combination of the vibration sensors 111 and 117 attached to the same member. Each time, by applying an optimum filter to both signals, the noise signal x _n can be estimated, and the estimated signal S _n ′ from which the noise has been removed can be calculated. As the optimum filter, a Kalman filter or a Wiener filter, which is known to be sequentially executable in the time domain, can be used, but in this embodiment, as shown in FIG. 5, an algorithm using a coefficient variable filter, that is, is preferable to employ an algorithm coefficients of variable coefficient filter in the direction difference e _n of the output signal d _n and the output y _n of the variable coefficient filter _(the estimated signal S _{n 'equivalent)} is minimized is updated is there.
Note that the coefficient of the coefficient variable filter is updated each time the optimum filter is sequentially applied to the signals of the vibration sensors 111 and 117, and the update is 2 of the difference (error) between the desired signal S _n and the estimated signal S _n ′. It is carried out so that the root mean square value is minimized.

  As an algorithm for updating a coefficient using such a coefficient variable filter, a gradient type algorithm or an algorithm based on a least square method is known. In this embodiment, an LMS (minimum mean square) algorithm or a particularly practical utility is used. Although an optimal filter such as an RLS (Sequential Least Squares) algorithm can be employed, the RLS (Sequential Least Squares) algorithm and a gradient coupled process algorithm have been confirmed to be more suitable as a result of experiments.

Any of these algorithms can cope with non-stationarity well by incorporating a forgetting coefficient into the coefficient in order to reduce the influence on the estimation result of the past estimated signal S _n ′. Among these algorithms, the RLS algorithm is particularly adaptable when a non-stationary phenomenon occurs, and the gradient type coupling process algorithm is also highly practical. Note that the RLS algorithm belongs to the Kalman filter, and the gradient combination process algorithm belongs to the Wiener filter, and can be realized as various optimum filters by devising (for example, combining) the implementation methods of these algorithms.

  Furthermore, a typical gradient type algorithm is the LMS algorithm, which incorporates a forgetting factor into the factor and a normalization algorithm that makes the step size parameter variable according to the amplitude of the input signal. By using the normalized LMS algorithm that has been adopted, non-stationarity can be dealt with, and a practical level can be achieved.

  Therefore, according to the present embodiment, in an abnormality diagnosis for a rolling device incorporated in a railway vehicle or the like, noise is estimated from the output signals of both the detection vibration sensor and the reference vibration sensor using an optimum filter, Since physical quantity detection and abnormality diagnosis are performed with the signal output by subtracting the minute, the degradation of detection accuracy caused by a rapidly changing environment or an environment with a lot of noise or vibration is suppressed, and the detection accuracy is improved. Can do.

In addition to this effect, according to the present embodiment, the optimum filter is equivalent to estimating the noise transfer function H _n , and the reference sensor 117 is arranged in the vicinity of the detection sensor 111, that is, the same member. Since the order of the transfer function H _n to be estimated and calculated can be suppressed low, the calculation load of the DSP circuit 173 and the like can be reduced, and the estimated signal S _n ′ subject to abnormality diagnosis can be calculated quickly and accurately. You can also

(Second Embodiment)
Next, a second embodiment according to the present invention will be described with reference to FIG.
Note that in the embodiments described below, the same or equivalent components in the drawings are simplified or omitted by giving the same or equivalent reference numerals to the components that are the same as those in the first embodiment described above and functionally similar components. .

  In the present embodiment, a physical quantity detection device 350 including a detection vibration sensor 111, a reference detection sensor 117, and a signal processing device 150 is applied to a steel rolling mill 300. The steel rolling mill 300 according to the present embodiment includes an upper work roll 301 and a lower work roll 302, is rotatably supported by a support bearing 303, and is coupled to a spindle 304. The spindle 304 is connected to the speed reducer 306 by a joint 305 such as a coupling. The speed reducer 306 is rotationally driven by a motor 308 having a joint 307 mounted on the motor shaft. That is, when the motor 308 is driven to rotate, the upper and lower work rolls 301 and 302 rotate in a coordinated manner, and the material to be rolled 309 such as metal passes between the upper and lower work rolls 301 and 302 and is rolled. The As the support bearing 303 of this embodiment, for example, a four-row tapered roller bearing, a four-row cylindrical roller bearing, or a double-row tapered roller bearing is used, but it is not particularly limited and is within the scope of the present invention. Arbitrarily selected.

The detection vibration sensor 111 is attached to each support bearing 303. Furthermore, in this embodiment, since the main noise source during operation is expected to be the speed reducer 306, the reference vibration sensor 117 is attached to the speed reducer 306. The 5-channel analog signals detected by the detection vibration sensor 111 and the reference detection vibration 117 are input to the abnormality diagnosis device 150.
Other aspects, that is, abnormality diagnosis based on signals detected from the detection vibration sensor 111 and the reference vibration sensor 117 are the same as those in the first embodiment.

  Therefore, according to the present embodiment, the same effect as that of the first embodiment can be obtained in the steel rolling mill.

(Third embodiment)
A third embodiment will be described with reference to FIG.
In the present embodiment, a physical quantity detection device 450 including a detection vibration sensor 111, a reference detection sensor 117, and a signal processing device 150 is applied to a drill grinding machine 400. A drill grinding machine 400, which is a machine tool, includes a rotation shaft (rolling device) 401, a main shaft 402 including a support bearing (not shown) that rotatably supports the rotation shaft, and a housing that holds the main shaft 402. 403, a main body 406 including a table 405 that supports the housing 403 via a linear guide 404 and fixes and supports the workpiece W, and a motor 408 that rotationally drives the rotary shaft 401 via a transmission mechanism 407. I have. The motor 408 is attached to the main shaft 402 by a bracket 409 so as to be parallel to the rotating shaft 401, and the transmission mechanism 407 includes a belt 407 a and a pulley 407 b and transmits the driving force of the motor 408 to the main shaft 402. ing.

  Then, while the drill T is mounted on the tip of the rotating shaft 401 and guided and supported by the linear guide 404, the rotating shaft 401 to which the drill T is mounted moves forward and backward with respect to the workpiece W arranged on the table 405 ( Grinding is performed by making the vertical direction in the figure.

  The reference vibration sensor 117 is attached to the bracket 409 so as to be arranged in the vicinity of the motor 408, and the detection vibration sensor 111 is attached to the table.

  Therefore, according to the present embodiment, the same effect as that of the first embodiment can be obtained in the machine tool.

Here, a test was conducted to confirm the effectiveness of the optimum filter when the rolling device of the first embodiment of the present invention was used. The frequency spectrum obtained by this test result is shown in FIG. In this test, an artificial flaw is processed on the inner ring raceway of the double row cone rolling bearing, the vibration is detected in this state, and the signal as it is output from the detection vibration sensor 111 (observation signal: lower stage in the figure) ) d _n and filtering unit 151 at the optimal filter after estimation signal (optimum filter process after having been subjected output: were compared taking the frequency spectrum in Fig upper) S n _'and the by FFT transform, respectively. Note that the RLS (Sequential Least Squares) algorithm was used as the optimal filter algorithm.

FIG. 9A shows a test result when the rotational speed of the rolling device is 360 rpm (min ⁻¹ ). In the signal to which the optimum filter is applied, noise is almost removed, and a peak indicating abnormality and its harmonics are shown. It can be confirmed that the wave peak is emphasized. FIG. 9B shows the test result when the rotation speed is 88 rpm (min ⁻¹ ). Only the noise can be confirmed in the signal waveform output as it is from the vibration sensor for detection 111, and the peak indicating abnormality is shown in FIG. Although it is completely buried in noise and cannot be confirmed at all, the noise-removed peak is confirmed in the signal subjected to the optimum filter, and an abnormal peak is confirmed, and the fundamental wave to the third harmonic can be confirmed with high accuracy.
That is, the knowledge shown in Table 2 could be obtained through such a test.

  Therefore, by this test, noise is estimated from the output signals of both the detection vibration sensor 111 and the reference vibration sensor 117 using an optimum filter, and the abnormality is diagnosed with the output signal after subtracting the noise. Thus, it can be seen that abnormality diagnosis can be performed quickly and accurately even at medium and low rotational speeds, which has been considered difficult in the past.

  This is the end of the description of specific embodiments. However, aspects of the present invention are not limited to these embodiments, and modifications, improvements, and the like can be made as appropriate. The vibration sensor for detection and the vibration sensor for reference do not need to be the same type of sensor. For example, a piezoelectric ceramic sensor having excellent durability is used as the vibration sensor for detection, and the acceleration detection direction is used as the vibration sensor for reference. A MEMS acceleration sensor that is excellent in performance and can be easily incorporated into an arithmetic circuit may be used.

  The rolling device may be a rolling bearing, a gear, an axle, a rotating part such as a machine tool or a ball screw, or a part such as a linear guide or a linear ball bearing, and generates periodic vibrations due to damage. Any parts can be used. Further, as the speed signal for calculating the frequency component due to the damage of the rotating part, the rotating speed signal has been exemplified, but a moving speed signal can also be used as the speed signal in the case of the sliding part.

  In the above embodiment, most of the digital processing is performed by software, but part or all of the digital processing may be realized by hardware such as an FPGA (Field Programmable Gate Array). Furthermore, in order to improve the signal-to-noise ratio of the input signal, an arithmetic unit that performs amplification processing or anti-aliasing filter processing may be provided before the filter processing unit.

  Moreover, the mechanical apparatus should just be provided with the rolling device which is an abnormality diagnosis object, In addition to what was mentioned above, the bearing apparatus for windmills, the bearing apparatus for machine tool main spindles, etc. can be included.

  Furthermore, although the detected signal has been exemplified and described as vibration, instead of this, it includes mechanical quantities and physical quantities such as sound, ultrasonic waves (AE), stress, displacement, and distortion. When there is a defect or abnormality in the mechanical equipment including the moving parts, a signal component indicating the defect or abnormality is included, so that it can be a target signal for abnormality diagnosis.

  In the first embodiment, the detection vibration sensor 111 and the reference vibration sensor 117 are described as being attached to the same member. However, the reference vibration sensor 117 is arranged to detect a signal including noise. For example, the object of the present invention can be achieved even when the signal processing device 150 is disposed on the control board.

  Furthermore, in the first embodiment, abnormality diagnosis is performed with one detection vibration sensor 111. However, in consideration of the load range of the double row tapered roller bearing 211, as shown in FIG. 10, the double row tapered roller bearing 211 is provided. In order to be able to face the inner and outer ring raceway surfaces 215, 215, 217, and 217, two detection vibration sensors may be provided for each rotation support device 110. Abnormalities can be diagnosed more quickly and accurately.

  In the above embodiment, the signal of all the vibration sensors 111 and 117 is acquired by one filter processing 151 and the optimum filter processing is performed. However, as shown in FIG. You may make it arrange | position individually according to the combination of the sensor 111 and the vibration sensor 117 for a reference. In this case, the amount of calculation can be increased, and an algorithm with quick convergence, for example, an RLS algorithm can be implemented.

  In addition, although the filter processing units 151 and 251 and the abnormality diagnosis processing unit 150B are provided separately, they may be configured as a single unit using these common CPUs. In this case, the manufacturing cost can be suppressed.

(A) is a schematic plan view of the railway vehicle carrying the abnormality diagnosis device of the present invention, and (b) is a schematic side view of the railway vehicle. It is sectional drawing for demonstrating the rolling device of 1st Embodiment. It is a block block diagram of the signal processing apparatus of a physical quantity detection apparatus. It is a block block diagram of the filter process part of a signal processing apparatus. It is a block diagram which shows operation | movement of a filter process part. It is a block diagram which shows another operation | movement of a filter process part. It is the schematic of the steel rolling mill carrying the physical quantity detection apparatus which concerns on 2nd Embodiment of this invention. It is the schematic of the machine tool carrying the physical quantity detection apparatus which concerns on 3rd Embodiment of this invention, and with which the drill which is a cutting tool was mounted | worn. It is a figure which shows the frequency spectrum comparison of the output signal of a sensor for detection, and the output signal after an optimal filter process, (a) is the rotational speed of 360 rolling (min < ^-1 >), (b) is rolling The rotation speed of the apparatus is 88 rpm (min ⁻¹ ). It is a figure which shows another attachment method of the vibration sensor which concerns on 1st Embodiment. It is a block block diagram which shows the modification of the signal processing part which concerns on 1st Embodiment.

Explanation of symbols

50, 350, 450 Physical quantity detection device 100 Railway vehicle 101 Wheel 110 Rotation support device (rolling device)
111 Detection Vibration Sensor (Detection Sensor)
117 Reference vibration sensor (reference sensor)
150 Signal processor 151,551 Filter processor 150A Sensor signal processor 150B Diagnosis processor 173 DSP circuit 211 Double row tapered roller bearing 221 Housing 300 Steel rolling mill 303 Support bearing (rolling device)
400 Drill grinding machine 401 Rotating shaft (rolling device)

Claims (8)

A detection sensor for detecting any physical quantity of vibration, sound, ultrasonic wave, stress, displacement, or strain emitted from a rolling device incorporated in a mechanical device;
A reference sensor disposed at a predetermined position to detect the physical quantity;
A signal processing unit that takes in a signal from the detection sensor and a signal from the reference sensor and applies an optimal filter to remove a noise signal generated from other than the rolling device and estimate the physical quantity;
A physical quantity detection device comprising: The signal processing unit
further,
A diagnosis processing unit that performs frequency diagnosis of the rolling device based on the estimated physical quantity to perform abnormality diagnosis of the rolling device;
The physical quantity detection device according to claim 1. The physical quantity detection device according to claim 1, wherein the reference sensor is also attached to a member to which the detection sensor is attached. The reference sensor is disposed in the signal processing unit.
The physical quantity detection device according to claim 1 or 2. The reference sensor is disposed so as to detect the physical quantity in the same direction as the detection sensor.
The physical quantity detection apparatus as described in any one of Claims 1-4. The optimal filter is at least one of a Kalman filter or a Wiener filter.
The physical quantity detection apparatus as described in any one of Claims 1-5. The optimal filter is at least one of an LMS algorithm or an RLS algorithm,
The physical quantity detection apparatus as described in any one of Claims 1-5.   A rolling device to which the physical quantity detection device according to any one of claims 1 to 7 is attached.

Images