TWI728535B - Monitor system and method thereof - Google Patents

Abstract

A monitor system is applied to a linear actuator, and the monitor system includes a sensing element, a monitor analysis module, a dynamic characteristic database and a comparing element. The sensing element is for sensing at least one signal formed via the linear actuator. The monitor analysis module is connected to the sensing element and for receiving the signal, and the signal is transformed into at least one characteristic value via a calculus software. The dynamic characteristic database includes at least one predetermined characteristic value. The comparing element is connected to the monitor analysis module and the dynamic characteristic database, and for comparing the characteristic value and the predetermined characteristic value. When the characteristic value matches the predetermined characteristic value, an alert signal is provided. Therefore, it is favorable for promoting an accuracy of the linear actuator.

Description

Monitoring system and its method

The present disclosure relates to a monitoring system and method thereof, and in particular to a monitoring system and method applied to a linear transmission mechanism.

The linear transmission mechanism is a mechanism in which steel balls roll and circulate between the slider and the linear slide rail, and a rolling interface is formed between the steel balls and the linear slide rail. Generally speaking, the rolling interface is considered to be the key position that affects the dynamic and static characteristics of the feed system structure. In detail, the load rigidity of the contact interface is directly related to the damping properties and the steel ball preload level. The rigidity and damping properties of the contact interface will directly affect the static load force and dynamic rigidity of the linear transmission mechanism.

However, the rolling elements or ball grooves in the linear transmission mechanism are prone to wear during long-term operation, which reduces the amount of interference between the balls and the linear slide rail and reduces the preload of the balls, which affects the rigidity and damping properties of the linear transmission mechanism. Therefore, a device that can monitor the linear transmission mechanism in real time is necessary, and the user can further use the device for real-time monitoring of the linear transmission mechanism to instantly improve and maintain the performance of the linear transmission mechanism.

The present disclosure provides a monitoring system and method thereof, which are applied to a linear transmission mechanism. Thereby, the performance of the linear transmission mechanism can be accurately monitored, and the user is immediately reminded to solve the abnormality of the linear transmission mechanism.

According to an embodiment of the present disclosure, a monitoring system is provided, which is applied to a linear transmission mechanism, and includes a sensing element, a monitoring analysis module, a dynamic feature database, and a comparison element. The sensing element is used for sensing at least one signal generated by the linear transmission mechanism. The monitoring analysis module is connected with the sensing element and used to receive a signal, and convert the signal into at least one characteristic value through a calculation software. The dynamic feature database includes at least one preset feature value. The comparison component is connected to the monitoring analysis module and the dynamic feature database, and is used to compare the feature value with the preset feature value, and provide a warning signal when the feature value matches the preset feature value.

The monitoring system according to the embodiment described in the preceding paragraph may further include a computer for carrying the monitoring analysis module, the dynamic feature database, and the comparison component.

According to the monitoring system of the embodiment described in the preceding paragraph, the sensing element can be a microphone, and the signal can be a voiceprint.

According to the monitoring system of the embodiment described in the preceding paragraph, the sensing element may include a three-axis accelerometer or a three-uniaxial accelerometer, and the signal may be a vibration.

According to the monitoring system of the embodiment described in the preceding paragraph, the calculation software can be a type of neural network model, and includes an input layer, a hidden layer, and an output layer. The signal is input to the input layer, and at least one input neuron is generated. The hidden layer is connected to the input layer, and it contains a plurality of hidden neurons, hidden neurons The element is used to count and transmit the input neurons of the input layer. The output layer is connected to the hidden layer, and the input neuron obtains the feature value after conversion through the hidden neuron. The input neuron can be a vibration root mean square value or a sound pressure root mean square value. The characteristic value can be a preload rigidity change or a preload failure value of the linear transmission mechanism.

The monitoring system according to the embodiment described in the preceding paragraph may further include a warning element, which is a warning light, which is used to generate a bright light after receiving the warning signal.

The monitoring system according to the embodiment described in the preceding paragraph may further include a warning element, which is a buzzer, which is used to generate a sound after receiving the warning signal.

The present disclosure provides a monitoring method applied to a linear transmission mechanism, which includes a step of establishing a dynamic characteristic database, a step of sensing the linear transmission mechanism, a step of signal conversion, and a step of output comparison. The step of establishing a dynamic characteristic database is to obtain at least one preset characteristic value through the complex load rigidity values of the plurality of sliders in the linear transmission mechanism. The step of sensing the linear transmission mechanism is to sense at least one signal in each of the three axial directions of a positioning platform of the linear transmission mechanism. In the signal conversion step, the signal received in the step of sensing the linear transmission mechanism is converted into at least one characteristic value through a neural network model. The output comparison step compares the preset feature value in the step of establishing the dynamic feature database with the feature value in the signal conversion step.

10‧‧‧Monitoring system

110‧‧‧Sensing components

111‧‧‧Single Axial Accelerometer

120‧‧‧Monitoring and Analysis Module

121‧‧‧like neural network model

121a‧‧‧Input layer

121b‧‧‧Hidden layer

121c‧‧‧Output layer

130‧‧‧Dynamic feature database

140‧‧‧Comparison components

150‧‧‧Warning element

20‧‧‧Linear transmission mechanism

210‧‧‧Base

220‧‧‧Linear slide

221‧‧‧Slider

230‧‧‧Positioning platform

30‧‧‧Monitoring method

S301‧‧‧Steps to create dynamic feature database

S302‧‧‧Sensing the steps of linear transmission mechanism

S303‧‧‧Signal conversion steps

S304‧‧‧Output comparison step

X‧‧‧Input neuron

W‧‧‧Hidden Neuron

Y‧‧‧Eigenvalue

Figure 1 shows the method of the monitoring system according to an embodiment of the present disclosure Block diagram

Fig. 2 is a schematic diagram of a neural network model according to the embodiment in Fig. 1;

Fig. 3 shows a schematic diagram of the linear transmission mechanism according to the embodiment in Fig. 1;

Figure 4 shows a schematic flow diagram of a monitoring method performed by the monitoring system according to the embodiment shown in Figure 1;

Fig. 5 is a schematic diagram of the time-domain vibration response of the positioning platform according to the monitoring method of the embodiment in Fig. 4;

Fig. 6 is a schematic diagram showing the relationship between the vibration of the positioning platform and the preloading degree of the slider according to the monitoring method of the embodiment in Fig. 4; and

Fig. 7 shows the scatter diagram of the actual value of the load rigidity of the positioning platform and the predicted value of the load rigidity of the neural network-like model in the monitoring method according to the embodiment of Fig. 4.

Please refer to FIG. 1. FIG. 1 is a block diagram of a monitoring system 10 according to an embodiment of the present disclosure. The present disclosure provides a monitoring system 10 applied to a linear transmission mechanism 20, which includes a sensing element 110, a monitoring analysis module 120, a dynamic feature database 130, a comparison element 140, and a warning element 150 , One of the computers (not shown in the figure) is equipped with a monitoring analysis module 120, a dynamic feature database 130, and a comparison component 140.

In detail, the sensing element 110 is used to sense at least one signal generated by the linear transmission mechanism 20. The monitoring analysis module 120 is connected to the sensing element 110, and is used to receive a signal, and convert the signal into at least one characteristic value Y (as shown in FIG. 2) through a calculation software.

The dynamic feature database 130 includes at least one preset feature value. The comparison element 140 is connected to the monitoring analysis module 120 and the dynamic feature database 130, and is used to compare the feature value Y with the preset feature value, and provide a warning signal when the feature value Y matches the preset feature value. Thereby, the dynamic characteristics of the linear transmission mechanism 20 can be monitored in real time. Through the calculation, analysis and comparison of the characteristic value Y, the changes of the linear transmission mechanism 20 can be predicted, or the state of the linear transmission mechanism 20 can be monitored in real time as a basis for maintenance. Maintain its effectiveness.

Furthermore, the sensing element 110 may be a microphone. When the sensing element 110 is a microphone, the signal received by the sensing element 110 is a voiceprint. The sensing element 110 may include a three-axis accelerometer or a three-axis accelerometer. When the sensing element 110 includes a triaxial accelerometer or a three uniaxial accelerometer, the signal received by the sensing element 110 is a vibration, but it is not limited to this. Thereby, the sensing element 110 can capture the signal of the linear transmission mechanism 20 in real time, and the collection effect is improved, and the state of the linear transmission mechanism 20 can be monitored more accurately.

In detail, the calculation software can be a type of neural network model 121. Please refer to FIG. 2. FIG. 2 is a schematic diagram of a neural network model 121 according to the embodiment in FIG. 1. The neural network-like model 121 includes an input layer 121a, a hidden layer 121b, and an output layer 121c. Specifically, a signal is input to the input layer 121a, and at least one input neuron X is generated. Hidden layer 121b is connected to the input layer 121a and includes a plurality of hidden neurons W, and the hidden neurons W are used to count and transmit the input neurons X of the input layer 121a. The output layer 121c is connected to the hidden layer 121b, and the input neuron X is transformed by the hidden neuron W to obtain the feature value Y. The input neuron X may be a vibration root mean square value or a sound pressure root mean square value, and the characteristic value Y may be a preload rigidity change or a preload failure value of the linear transmission mechanism 20. In this way, a more accurate judgment basis is provided for the monitoring system 10, and the prediction error value is reduced.

The warning element 150 may be a warning light or a buzzer. The warning light is used to generate a bright light after receiving the warning signal, and the buzzer is used to generate a sound after receiving the warning signal. In this way, the user is clearly reminded that there is an abnormal condition in the linear transmission mechanism 20, so that the user can immediately eliminate the abnormality.

Please refer to FIG. 3. FIG. 3 is a schematic diagram of the linear transmission mechanism 20 according to the embodiment in FIG. 1. The linear transmission mechanism 20 includes a base 210, a plurality of linear slide rails 220 and a positioning platform 230, wherein the linear slide rail 220 is disposed above the base 210, and the positioning platform 230 is disposed above the linear slide rail 220. At least one single-axis accelerometer 111 is provided on the three axes of the positioning platform 230 respectively. In detail, the linear slide 220 includes a plurality of sliders 221, the slide 221 is disposed between the linear slide 220 and the positioning platform 230, the linear transmission mechanism 20 is made of carbon steel, and the total travel length of the linear slide 220 is 1000 mm , But not limited to this.

Please refer to FIG. 4, which shows a schematic flow diagram of a monitoring method 30 performed by the monitoring system 10 according to the embodiment in FIG. 1. The monitoring method 30 includes a step S301 of establishing a dynamic feature database, a step S302 of sensing a linear transmission mechanism, a step of signal conversion S303, and an output comparison step S304.

In step S301 of establishing a dynamic characteristic database, a preset characteristic value is obtained through the complex load rigidity value of the slider 221 in the linear transmission mechanism 20.

Furthermore, first obtain the load rigidity of the sliding block 221 with three different preload levels according to the Hertz theory, and then obtain the load rigidity of the positioning platform 230 with the sliding block 221 of three different preload levels, and then obtain the preset characteristic value. The accelerometer 111 measures in the three directions of X, Y, and Z, and the number of the uniaxial accelerometer 111 of the monitoring system 10 of the present invention is one in the three directions of X, Y, and Z, and the signal is vibration. But it is not limited to this.

In detail, the slider 221 is divided into a high preload slider (505N/μm), a medium preload slider (424N/μm), and a low preload slider (294N/μm) based on load rigidity. The positioning platform 230 is equipped with four sliders 221, and is divided into a high preload platform (2020N/μm), a medium preload platform (1696N/μm) and a low preload platform (1176N/μm) according to the combination of the slider 221, and the slider 221 The load rigidity, quantity and combination are not limited to this. In addition, the measurement and positioning platform 230 is divided into different modes, including rolling mode (Rolling), swing mode (Yawing) and tilting mode (Pitching), but not limited to this.

Table 1 shows the modal frequencies of the high preload slider, the medium preload slider, and the low preload slider in the rolling mode, the rocking mode, and the tilt mode detected by the sensing element 110.

It can be seen from Table 1 that the high preload slider has a higher modal frequency in the three modes, and the change from the high preload slider to the medium preload slider will decrease. 7.7% to 11.5% of the modal frequency, changing from a medium preload slider to a low preload slider reduces the modal frequency by 21 to 23%. When the rolling elements or the chute are worn, the amount of interference decreases, so the load rigidity of the positioning platform 230 is further reduced. Therefore, replacing the slider 221 with a higher load rigidity value with a slider 221 with a lower load rigidity value can be used to simulate the wear of the rolling elements or the chute.

Next, perform five preload conditions of the positioning platform 230 with rigid load and adjust the motor feed rate of the linear transmission mechanism 20. The five preloading conditions of the positioning platform 230 are mode 1 to mode 5, among which mode 1 is the four slides 221 are low preload slides, the load rigidity of the positioning platform 230 is 1176N/μm; mode 2 is the four slides 221 are all medium preloaded sliders, the load rigidity of the positioning platform 230 is 1696N/μm; mode 3 is four sliders 221 are all high preloaded sliders, the load rigidity of the positioning platform 230 is 2020N/μm; mode 4 is three The slider 221 is a high-preload slider and one slider 221 is a low-preload slider. The load rigidity of the positioning platform 230 is 1809N/μm; mode 5 is two sliders 221 is a high preload slider and two sliders 221 It is a low preload slider, and the load rigidity of the positioning platform 230 is 1598N/μm. The feed rate of the motor is 500 revolutions/min, 1000 revolutions/min, 1500 revolutions/min, 2000 revolutions/min, 3000 revolutions/min, 4000 revolutions/min, 6000 revolutions/min and 8000 revolutions/min.

Please refer to Figures 5 and 6. Figure 5 shows a schematic diagram of the time-domain vibration response of the positioning platform 230 according to the monitoring method 30 of the embodiment in Figure 4, where (a) is mode 1, (b) is mode 2, and (c) is mode 3, (d) is mode 4, and (e) is mode 5. FIG. 6 is a schematic diagram showing the relationship between the vibration of the positioning platform 230 and the preloading degree of the slider 221 according to the monitoring method 30 of the embodiment in FIG. 4. It can be seen from Fig. 5 and Fig. 6 that the The amount of vibration increases with the increase of the feed rate of the motor, so the vibration caused by mode 3 is more significant than that of mode 1, and the vibration energy generated by mode 1 is lower than that of mode 3. Furthermore, mode 4 and mode 5 are a hybrid high-preload slider and low-preload slider, so the amount of vibration generated is between mode 1 and mode 3. In this way, the preset characteristic values of the preload conditions of different sliders 221 can be obtained, and the changes in the actual preload conditions of the linear transmission mechanism 20 can be compared with the preset characteristic values.

In step S302 of sensing the linear transmission mechanism, at least one signal of the three axial directions of the linear transmission mechanism 20 is sensed. In detail, the linear transmission mechanism 20 performs sensing in the X, Y, and Z directions through the single-axis accelerometer 111, and the detected signal is vibration, but it is not limited to this.

In the signal conversion step S303, the signal received in the step S302 of the sensing linear transmission mechanism is converted into the characteristic value Y through the neural network model 121.

In detail, the hidden layer 121b in the neural network-like model 121 uses Tan-Sigmoid as the activation function, and the output layer 121c in the neural network-like model 121 uses Linear as the activation function, but not Is limited.

Specifically, the signals input by the input layer 121a are the slider 221 initial preload level (Initial preload level), platform initial load rigidity (Initial rigidity), X-axis vibration, Y-axis vibration, Z-axis vibration, and The amount of vibration in the axis of rotation (rolling mode, rocking mode, and tilt mode).

Table 2 shows the initial preload level of the slider 221 and the preload failure state of the positioning platform 230 for the five positioning platform 230 modes.

It can be seen from Table 2 that the positioning platform 230 mode is mode I to mode V, where mode I is that the preload is reduced by 42%, which belongs to heavy wear; mode II is that the preload is reduced by 15%, which is moderate wear; the preload of mode III There is no reduction in pressure, which is a normal state; Mode IV is a 10% reduction in preload, which means that one of the sliders 221 is a serious failure and belongs to a moderate abnormal state; Mode V is that preload is reduced by 21%, which is the second slide 221 It is a serious failure and belongs to a severe abnormal state.

Then, the five positioning platform 230 modes are applied to the neural network model 121 such as the monitoring system 10 of the present invention for analysis, and six MLP modes are obtained. Table 3 shows the sensitivity levels of the six MLP modes.

The MLP in Table 3 is a multi-layer function connection reverse transfer neural network. The meaning of the number is MLP 11-4-1 as an example. 11 means that the input layer 121a has 11 input neurons X, and 4 means that the hidden layer 121b has 4 The hidden neuron W, 1 represents that the output layer 121c has 1 feature value Y. The larger the sensitivity value in Table 3, the less important its impact, on the contrary, the smaller the value, the higher the impact degree. Then, through Sensitivity Analysis, all the characteristic values Y are regarded as missing data one by one. Specifically, when the more important feature value Y is selected as the missing value, the error value of the neural network model 121 will increase significantly; when the less important feature value Y is selected as the missing value, the class The increase in the error value of the neural network model 121 is very limited. In this way, the prediction error value of the similar neural network model 121 can be obtained, so as to more accurately predict whether the linear transmission mechanism 20 is abnormal.

After the signal conversion step S303 is completed, an output comparison step S304 is performed. In the output comparison step S304, the data in the step S301 of establishing a dynamic feature database and the data in the signal conversion step S303 are compared and analyzed.

Table 4 shows the correlation coefficients and root mean square errors of the six MLP modes.

The present invention uses Table 4 to evaluate the accuracy of the five positioning platform 230 modes embedded in the six MLP modes of the neural network model 121, and the correlation coefficient (Coefficient of Correlation) and the root mean square error (Root Mean Square) in Table 4 Square Error) contains three parameters: training, testing, and verification. There are 35 data sets in 230 modes of five positioning platforms. 70% of the data are randomly selected for training, 15% for testing, and 15% for testing. Verification, but the above data ratio is not limited to this. Through the calculation of the error value, the error value of the six MLP modes is less than 3%, and the formula for the error value is [(load rigidity prediction value-load rigidity actual value)/load rigidity actual value]*100%. Please refer to FIG. 7. FIG. 7 is a scatter diagram of the actual load rigidity value of the positioning platform 230 and the predicted load rigidity value of the neural network model 121 in the monitoring method 30 according to the embodiment in FIG. 4. It can be seen from Fig. 7 that the linear correlation between the six MLP modes and the actual value is above 0.97. It can be seen that the monitoring system 10 and the neural network model 121 of the present invention can accurately predict the linear transmission mechanism 20. The failure status.

Although the present invention has been disclosed in the above embodiments, it is not intended to limit the present invention. Anyone with ordinary knowledge in the relevant technical field can make some changes and modifications without departing from the spirit and scope of the present invention. The protection scope of the present invention shall be subject to those defined by the attached patent application scope.

10‧‧‧Monitoring system

110‧‧‧Sensing components

120‧‧‧Monitoring and Analysis Module

130‧‧‧Dynamic feature database

140‧‧‧Comparison components

150‧‧‧Warning element

20‧‧‧Linear transmission mechanism

210‧‧‧Base

220‧‧‧Linear slide

230‧‧‧Positioning platform

Claims (7)

A monitoring system, which is applied to a linear transmission mechanism, includes: a sensing element used to sense at least one signal generated by the linear transmission mechanism; a monitoring analysis module connected to the sensing element and used In order to receive the at least one signal, convert the at least one signal into at least one characteristic value through a calculation software; a dynamic characteristic database including at least one predetermined characteristic value; and a comparison element connected to the monitoring analysis module Group and the dynamic feature database, and used to compare the at least one feature value with the at least one preset feature value, and provide a warning signal when the at least one feature value matches the at least one preset feature value ; Wherein, the calculation software is a type of neural network model, and includes: an input layer, the at least one signal is input to the input layer, and generates at least one input neuron; a hidden layer, which is connected to the input layer, and It includes a plurality of hidden neurons, the hidden neurons are used to count and transmit the at least one input neuron of the input layer; and an output layer, which is connected to the hidden layer, and the at least one input neuron passes through the hidden The neuron obtains the at least one characteristic value after conversion; Wherein, the at least one input neuron is a vibration root mean square value or a sound pressure root mean square value; wherein, the at least one characteristic value is a preload rigidity change or a preload failure value of the linear transmission mechanism. For example, the monitoring system described in item 1 of the scope of patent application further includes a computer for carrying the monitoring analysis module, the dynamic feature database and the comparison component. According to the monitoring system described in claim 1, wherein the sensing element is a microphone, and the at least one signal is a voiceprint. In the monitoring system described in item 1 of the scope of patent application, the sensing element includes a three-axis accelerometer or a three uniaxial accelerometer, and the at least one signal is a vibration. The monitoring system described in item 1 of the scope of patent application further includes a warning element, which is a warning light, which is used to generate a bright light after receiving the warning signal. The monitoring system described in item 1 of the scope of patent application further includes a warning element, which is a buzzer, which is used to generate a sound after receiving the warning signal. A monitoring method is applied to a linear transmission mechanism, which includes: a step of establishing a dynamic characteristic database, obtaining at least one preset characteristic value through the complex load rigidity values of the plural sliders in the linear transmission mechanism; and a sensing linearity The transmission mechanism step is to sense at least one signal in each of the three axial directions of a positioning platform of the linear transmission mechanism; a signal conversion step is to convert the at least one signal received in the sensing linear transmission mechanism step through a neural network model A signal is at least one characteristic value; and an output comparison step is to compare the at least one predetermined characteristic value in the step of establishing a dynamic characteristic database with the at least one characteristic value in the signal conversion step.

Images