A method for detecting circular runout fault of servo motor

Abstract

The present invention relates to the field of motor detection technology, and in particular to a method for detecting circular runout faults of servo motors. The present invention adopts the following technical scheme: a laser ranging sensor is used to perform circular runout detection on the rotating shaft of a servo motor in a non-contact manner to collect test data, a univariate variance analysis method and a multivariate covariance matrix analysis method are used to calculate and process the test data to determine whether the servo motor to be tested has a fault, and the specific fault type of the servo motor is determined by calculating the deviation rate of each indicator of the detection data and combining historical test records. The beneficial effects of the present invention are: by using a laser ranging sensor to collect circular runout data, the sampling speed and accuracy can be effectively improved; at the same time, by using a univariate variance analysis method and a multivariate covariance matrix analysis method, the overall distribution characteristics of the data can be fully considered, the assembly quality can be reflected, and the type of servo motor fault can be determined.

Description

Technical Field

[0001] The present invention relates to the technical field of motor detection, and in particular to a method for detecting circular runout fault of a servo motor. Background Art

[0002] Traditionally, the circular runout detection of the servo motor shaft usually uses contact measuring equipment such as micrometers. The use of contact measuring equipment has problems such as low test efficiency, susceptibility to human factors, and limited test accuracy. In addition, the current circular runout detection of servo motors only uses the maximum circular runout as the judgment standard, ignoring the overall distribution characteristics of the data, unable to fully reflect the assembly quality, and unable to determine the type of fault. Summary of the invention

[0003] The purpose of the present invention is to provide a method for detecting circular runout faults of a servo motor, and more particularly to provide a method for quickly and accurately detecting circular runout of a rotating shaft of a servo motor and determining the type of fault.

[0004] To achieve the above object, the present invention adopts the following technical solution: a servo motor circular runout fault detection method, comprising the following steps:

[0005] S01. First, use a laser distance measuring sensor to collect N circular runout data of the rotating shaft of the servo motor to be tested at a certain sampling frequency to obtain N groups of test data.

[0006] S02, using the univariate variance analysis method and the multivariate covariance matrix analysis method to calculate the multiple test data obtained in step S01, to determine whether the shaft of the servo motor has a fault, wherein the specific calculation steps of the univariate variance analysis method and the multivariate covariance matrix analysis method are as follows:

[0007] S21. Univariate analysis of variance method:

[0008] (1) Calculate the average value of N groups of test data respectively, recorded as: (i=1, 2, ..., N); where, is the average value of the i-th group of data, is the number of data in the i-th group of data, is the value of the jth data in the i-th group of data.

[0009] (2) Calculate the average value of all data in N groups of test data: .

[0010] (3) Calculate the horizontal error sum of squares: , and the horizontal error sum of squares The degrees of freedom are N-1.

[0011] (4) Calculate the sum of squares of the error terms: , and the degrees of freedom of the sum of squares of the error terms is: .

[0012] (5) Calculate the statistics of the F distribution: , and based on the degrees of freedom N-1 of the horizontal error sum of squares and the degrees of freedom of the error term sum of squares Determine the corresponding value F in the F distribution table A , compare the calculated statistic value F with the value F determined in the F distribution table A If F>F A , it is determined that the shaft of the servo motor is faulty.

[0013] S22. Multivariate covariance matrix analysis method:

[0014] (1) First, use the method in step S01 to collect circular runout data of a servo motor of the same model that has been determined to be operating normally, and obtain a set of benchmark data, which is recorded as , n is the number of benchmark data.

[0015] (2) Select any one group from the N groups of test data obtained in step S01, denoted as ,in Representative Set test data, For the The number of test data sets.

[0016] (3) Starting from the first data, windows of length w are generated in sequence with a step length of s for the benchmark data of step S22 (1) and the test data of step S22 (2). Each window is recorded as a vector, and a multidimensional data set is obtained for each set of data. The multidimensional data set obtained from the benchmark data is recorded as ,in , … is the vector of each window, p is the number of windows obtained from the benchmark data; the multidimensional data set obtained from the test data is ,in , … is the vector of each window, and q is the number of windows obtained by the test data.

[0017] (4) Multidimensional datasets using benchmark data Compute the probability density function of the multivariate normal distribution: , where w is the dimension of vector y, i.e., the length of the window in (3) of step S22, is the covariance matrix calculated from the multidimensional dataset of the benchmark data, is the average value of each window vector of the multidimensional dataset of the benchmark data, that is, , ; Vector y is a vector randomly selected from each window vector of the multidimensional data set Y of the test data, or is the mean vector of each window vector of the multidimensional data set Y of the test data.

[0018] (5) According to the probability density calculated in (4), , by judging whether the probability density exceeds the set threshold, if it exceeds the set threshold, it is judged that the shaft of the servo motor is faulty.

[0019] Further, step S03 is included. When the method of S21 and / or S22 in step S02 determines that the shaft of the servo motor has a fault, step S03 is used to determine the fault type of the servo motor, specifically:

[0020] S31. Calculate the overall indicators of the circular runout data of the servo motor to be tested, including: standard deviation, maximum deviation value, quarter digit, kurtosis value of normal test and skewness value of normal test. The specific calculation method is as follows:

[0021] (1) Overall standard deviation std. Calculate the standard deviations of the N groups of test data in step S01 respectively, record them as std1, std2, ..., stdN, and use them to calculate the overall standard deviation: .

[0022] (2) The overall maximum deviation value max is calculated by calculating the maximum deviation values ​​of the N groups of test data in step S01. The maximum deviation value is the difference between the maximum value and the minimum value in the group of test data, which are recorded as max1, max2, ..., maxN. The overall maximum deviation value is calculated in this way: .

[0023] (3) Overall quarter digit Q. Calculate the quarter digits of the N groups of test data in step S01 respectively, record them as Q1, Q2, ..., QN, and calculate the overall quarter digit in this way: .

[0024] (4) The kurtosis value qu of the overall normality test is to perform normal distribution on the N groups of test data in step S01 respectively, and calculate the kurtosis value of each group of test data, which is recorded as qu1, qu2, ..., quN, and then calculate the overall kurtosis value: .

[0025] (5) The skewness value sk of the overall normality test is: N groups of test data in step S01 are normally distributed, and the skewness value of each group of test data is calculated, recorded as sk1, sk2, ..., skN, and the overall skewness value is calculated in this way: .

[0026] S32, respectively calculating the deviation rate of each indicator in step S31, specifically:

[0027] (1) Deviation rate of standard deviation , wherein std0 is a standard deviation reference value calculated by the method of step S01 and step S31 for a servo motor of the same model that has been determined to be operating normally;

[0028] (2) Deviation rate of maximum deviation value , wherein max0 is a maximum deviation reference value calculated by the method of step S01 and step S31 for a servo motor of the same model that has been determined to be operating normally.

[0029] (3) One-quarter digit deviation rate , wherein Q0 is a quarter-digit reference value calculated by the method of step S01 and step S31 for a servo motor of the same model that has been determined to be operating normally.

[0030] (4) Deviation rate of kurtosis value .

[0031] (5) Deviation rate of skewness value .

[0032] S33. For the deviation rate A0, B0, C0, D0, E0 of each indicator calculated in step S32, if the deviation rate is greater than 50%, it is considered that the indicator corresponding to the deviation rate is abnormal, otherwise it is normal. According to the abnormality or normality of each indicator and combined with the historical fault records, the fault type of the servo motor to be tested is determined.

[0033] The beneficial effects of the present invention are as follows: by using a laser ranging sensor to perform circular runout detection on the rotating shaft of a servo motor in a non-contact manner, the sampling speed and accuracy of the circular runout detection of the servo motor can be effectively improved; at the same time, a univariate variance analysis method and a multivariate covariance matrix analysis method are used in conjunction to perform calculation processing on the detection data to determine whether the rotating shaft of the servo motor has a fault, and by calculating the deviation rate of each indicator of the detection data, the specific fault type of the servo motor is judged, thereby improving the efficiency and accuracy of the circular runout detection of the servo motor. DETAILED DESCRIPTION

[0034] Embodiment 1, a servo motor circular runout fault detection method, comprising the following steps:

[0035] S01. First, use a laser ranging sensor to collect N circular runout data of the shaft of the servo motor to be tested at a certain sampling frequency to obtain N groups of test data. Specifically, when using a laser ranging sensor to collect circular runout data of the shaft of the servo motor, it is necessary to select a laser ranging sensor with higher precision, and generally select a ranging sensor with a measurement accuracy of micrometer level; during the specific operation, fix the servo motor on a standard test platform, fix the laser ranging sensor and align it with the outer circle of the servo motor shaft, and align the laser emitted by the laser ranging sensor with the axis of the servo motor shaft, control the servo motor to rotate at a lower speed, use the laser ranging sensor to collect data at a certain sampling frequency during the rotation process, and control the servo motor to rotate at different speeds to collect N circular runout data, so as to obtain N groups of test data. The detection distance of the laser ranging sensor is the distance between the laser ranging sensor and the surface of the shaft of the servo motor. As the shaft rotates, the runout of the shaft can be judged by the change in the distance between the laser ranging sensor and the surface of the shaft. In this embodiment, the sampling frequency of the laser ranging sensor is set to 1000 Hz, and three circular runout data collections are performed to obtain three sets of test data, namely N=3. Of course, the above sampling frequency and collection times can be adjusted accordingly according to actual needs.

[0036] S02, using the univariate variance analysis method and the multivariate covariance matrix analysis method to calculate the multiple test data obtained in step S01, to determine whether the shaft of the servo motor has a fault, wherein the specific calculation steps of the univariate variance analysis method and the multivariate covariance matrix analysis method are as follows:

[0037] S21. Univariate analysis of variance method:

[0038] (1) Calculate the average value of N groups of test data respectively, recorded as: (i=1, 2, ..., N); where, is the average value of the i-th group of data, is the number of data in the i-th group of data, is the value of the jth data in the i-th group of data.

[0039] (2) Calculate the average value of all data in N groups of test data: .

[0040] (3) Calculate the horizontal error sum of squares: , and the horizontal error sum of squares The degrees of freedom are N-1.

[0041] (4) Calculate the sum of squares of the error terms: , and the degrees of freedom of the sum of squares of the error terms is: .

[0042] (5) Calculate the statistics of the F distribution: , and based on the degrees of freedom N-1 of the horizontal error sum of squares and the degrees of freedom of the error term sum of squares Determine the corresponding value F in the F distribution table A , compare the calculated statistic value F with the value F determined in the F distribution table A If F>F A , it is determined that the shaft of the servo motor to be tested has a fault. A The standard value is determined by querying from the F distribution table. When the actual calculated F distribution statistic is greater than the standard value found in the F distribution table, it can be determined that the shaft of the servo motor to be tested has a fault. The F distribution table is a table commonly used in the art. The standard value of the F distribution statistic can be determined by querying to quickly determine whether the data is abnormal.

[0043] S22. Multivariate covariance matrix analysis method:

[0044] (1) First, use the method in step S01 to collect circular runout data of a servo motor of the same model that has been determined to be operating normally, and obtain a set of benchmark data, which is recorded as , n is the number of benchmark data.

[0045] (2) Select any one group from the N groups of test data obtained in step S01, denoted as ,in Representative Set test data, For the The number of test data sets.

[0046] (3) Starting from the first data, windows of length w are generated in sequence with a step length of s for the benchmark data of step S22 (1) and the test data of step S22 (2). Each window is recorded as a vector, and each group of data obtains a multidimensional data set. The multidimensional data set obtained by the benchmark data is recorded as ,in , … is the vector of each window, p is the number of windows obtained from the benchmark data; the multidimensional data set obtained from the test data is ,in , … is the vector of each window, and q is the number of windows obtained by the test data.

[0047] (4) Multidimensional datasets using benchmark data Compute the probability density function of the multivariate normal distribution: , where w is the dimension of vector y, i.e., the length of the window in (3) of step S22, is the covariance matrix calculated from the multidimensional dataset of the benchmark data, is the average value of each window vector of the multidimensional dataset of the benchmark data, that is, , ; Vector y is a vector randomly selected from each window vector of the multidimensional data set Y of the test data, or is the mean vector of each window vector of the multidimensional data set Y of the test data.

[0048] (5) Based on the probability density calculated in (4) of step S22 , by judging whether the probability density exceeds the set threshold, if it exceeds the set threshold, it is judged that the shaft of the servo motor to be tested has a fault. The set threshold can be set according to the experience of the tester, or it can be determined by the value obtained by performing the same steps on a servo motor of the same model that has been determined to be operating normally.

[0049] Specifically, when any of the methods in step S21 and S22 indicates that the servo motor to be tested has a fault, it is considered that the servo motor to be tested has a fault. At this time, step S03 is used to determine the fault type of the servo motor, specifically:

[0050] S31. Calculate the overall indicators of the circular runout data of the servo motor to be tested, including: standard deviation, maximum deviation value, quarter digit, kurtosis value of normal test and skewness value of normal test. The specific calculation method is as follows:

[0051] (1) Overall standard deviation std. Calculate the standard deviations of the N groups of test data in step S01 respectively, record them as std1, std2, ..., stdN, and use them to calculate the overall standard deviation: .

[0052] (2) The overall maximum deviation value max is calculated by calculating the maximum deviation values ​​of the N groups of test data in step S01. The maximum deviation value is the difference between the maximum value and the minimum value in the group of test data, which are recorded as max1, max2, ..., maxN. The overall maximum deviation value is calculated in this way: .

[0053] (3) Overall quarter digit Q. Calculate the quarter digits of the N groups of test data in step S01 respectively, record them as Q1, Q2, ..., QN, and calculate the overall quarter digit in this way: .

[0054] (4) The kurtosis value qu of the overall normality test is to perform normal distribution on the N groups of test data in step S01 respectively, and calculate the kurtosis value of each group of test data, which is recorded as qu1, qu2, ..., quN, and then calculate the overall kurtosis value: .

[0055] (5) The skewness value sk of the overall normality test is: N groups of test data in step S01 are normally distributed, and the skewness value of each group of test data is calculated, recorded as sk1, sk2, ..., skN, and the overall skewness value is calculated in this way: .

[0056] S32, respectively calculating the deviation rate of each indicator in step S31, specifically:

[0057] (1) Deviation rate of standard deviation , wherein std0 is a standard deviation reference value calculated by the method of step S01 and step S31 for a servo motor of the same model that has been determined to be operating normally;

[0058] (2) Deviation rate of maximum deviation value , wherein max0 is a maximum deviation reference value calculated by the method of step S01 and step S31 for a servo motor of the same model that has been determined to be operating normally.

[0059] (3) One-quarter digit deviation rate , wherein Q0 is a quarter-digit reference value calculated by the method of step S01 and step S31 for a servo motor of the same model that has been determined to be operating normally.

[0060] (4) Deviation rate of kurtosis value Under ideal conditions, the fluctuation of circular runout data is normally distributed, which is a zero kurtosis state with a kurtosis value of 3. When the fluctuation of circular runout data is abnormal, it will be a positive kurtosis state with a kurtosis value greater than 3. Therefore, 3 is used as the reference value for calculating the kurtosis value deviation rate.

[0061] (5) Deviation rate of skewness value Similarly, under ideal conditions, the skewness value is 1. Whether the skewness value is greater than 1 or less than 1, it indicates that there is a deviation. Therefore, 1 is used as the benchmark value for calculating the skewness value deviation rate.

[0062] S33. For the deviation rates A0, B0, C0, D0, and E0 of each indicator calculated in step S32, if the deviation rate is greater than 50%, it is considered that the indicator corresponding to the deviation rate is abnormal, otherwise it is normal. According to the abnormality or normality of each indicator, combined with the historical fault records, the fault type of the servo motor to be tested is judged. When judging the fault type of the servo motor to be tested, it is necessary to combine the fault types existing in the servo motor circular runout detection obtained by past statistics and the indicators of abnormal deviation rates corresponding to the faults. The following table is a statistical table of the fault types of the servo motor and which indicators corresponding to the faults are abnormal:

[0063]

[0064] In the table, "1" represents that the deviation rate of the corresponding indicator of the fault type is abnormal, and "0" represents that the deviation rate of the corresponding indicator of the fault type is normal. When judging the fault of the servo motor, if at least four of the above five indicators have the same abnormal / normal state of the deviation rate as the fault type in the table, and the deviation rate of the indicator with abnormal deviation rate in the fault type is also abnormal in the servo motor to be tested, it is judged that the servo motor to be tested has the fault type. For example, after calculation, the state of the deviation rate of each indicator of the servo motor to be tested is: 1, 1, 0, 1, 1. After comparison, the fault types that meet the conditions that the state of the deviation rate of at least four indicators is consistent with the fault type in the table include: motor vibration and others. However, the deviation rate of each indicator of the fault type "other" is 1 (that is, abnormal), and the deviation rate of the indicator of the quarter digit of the servo motor to be tested is normal. Therefore, "other" is excluded, and the fault type of the servo motor to be tested is judged to be "motor vibration".

[0065] Of course, the above are only preferred embodiments of the present invention, and are not intended to limit the scope of use of the present invention. Therefore, any equivalent changes made to the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for detecting circular runout fault of a servo motor, characterized in that: The steps include: S01, first use a laser ranging sensor to collect N circular runout data of the rotating shaft of the servo motor to be tested at a certain sampling frequency to obtain N groups of test data; S02, using the univariate variance analysis method and the multivariate covariance matrix analysis method to calculate the multiple test data obtained in step S01, to determine whether the shaft of the servo motor has a fault, wherein the specific calculation steps of the univariate variance analysis method and the multivariate covariance matrix analysis method are as follows: S21. Univariate analysis of variance method: (1) Calculate the average value of N groups of test data respectively, recorded as: , i=1, 2, ..., N; where, is the average value of the i-th group of data, is the number of data in the i-th group of data, is the value of the jth data in the i-th group of data; (2) Calculate the average value of all data in N groups of test data: ; (3) Calculate the horizontal error sum of squares: , and the horizontal error sum of squares The degrees of freedom are N-1; (4) Calculate the sum of squares of the error terms: , and the degrees of freedom of the sum of squares of the error terms is: ; (5) Calculate the statistics of the F distribution: , and based on the degrees of freedom N-1 of the horizontal error sum of squares and the degrees of freedom of the error term sum of squares Determine the corresponding value F in the F distribution table A , compare the calculated statistic value F with the value F determined in the F distribution table A If F>F A , it is determined that the shaft of the servo motor is faulty; S22. Multivariate covariance matrix analysis method: (1) First, use the method in step S01 to collect circular runout data of a servo motor of the same model that has been determined to be operating normally, and obtain a set of benchmark data, which is recorded as , n is the number of benchmark data; (2) Select any one group from the N groups of test data obtained in step S01, denoted as ,in Representative Set test data, For the The number of test data in the group; (3) Starting from the first data, windows of length w are generated in sequence with a step length of s for the benchmark data of step S22 (1) and the test data of step S22 (2). Each window is recorded as a vector, and each group of data obtains a multidimensional data set. The multidimensional data set obtained by the benchmark data is recorded as ,in , … is the vector of each window, p is the number of windows obtained from the benchmark data; the multidimensional data set obtained from the test data is ,in , … is the vector of each window, and q is the number of windows obtained by the test data; (4) Multidimensional datasets using benchmark data Compute the probability density function of the multivariate normal distribution: , where w is the dimension of vector y, i.e., the length of the window in (3) of step S22, is the covariance matrix calculated from the multidimensional dataset of the benchmark data, is the average value of each window vector of the multidimensional dataset of the benchmark data, that is, , ; Vector y is a vector randomly selected from each window vector of the multidimensional dataset Y of the test data, or is the mean vector of each window vector of the multidimensional dataset Y of the test data; (5) Based on the probability density calculated in (4) of step S22 , by judging whether the probability density exceeds the set threshold, if it exceeds the set threshold, it is judged that the shaft of the servo motor is faulty.

2. A servo motor circular runout fault detection method according to claim 1, characterized in that: The method further includes step S03. When the method of S21 and / or S22 in step S02 determines that the shaft of the servo motor has a fault, step S03 is used to determine the fault type of the servo motor. Specifically, the method includes: S31. Calculate the overall indicators of the circular runout data of the servo motor to be tested, including: standard deviation, maximum deviation value, quarter digit, kurtosis value of normal test and skewness value of normal test. The specific calculation method is as follows: (1) Overall standard deviation std. Calculate the standard deviations of the N groups of test data in step S01 respectively, record them as std1, std2, ..., stdN, and use them to calculate the overall standard deviation: ; (2) The overall maximum deviation value max is calculated by calculating the maximum deviation values ​​of the N groups of test data in step S01 respectively. The maximum deviation value is the difference between the maximum value and the minimum value in the group of test data, which is recorded as max1, max2, ..., maxN. The overall maximum deviation value is calculated in this way: ; (3) Overall quarter digit Q. Calculate the quarter digits of the N groups of test data in step S01 respectively, record them as Q1, Q2, ..., QN, and calculate the overall quarter digit in this way: ; (4) The kurtosis value qu of the overall normality test is to perform normal distribution on the N groups of test data in step S01 respectively, and calculate the kurtosis value of each group of test data, which is recorded as qu1, qu2, ..., quN, and then calculate the overall kurtosis value: ; (5) The skewness value sk of the overall normality test is: N groups of test data in step S01 are normally distributed, and the skewness value of each group of test data is calculated, recorded as sk1, sk2, ..., skN, and the overall skewness value is calculated in this way: ; S32, respectively calculating the deviation rate of each indicator in step S31, specifically: (1) Deviation rate of standard deviation , wherein std0 is a standard deviation reference value calculated by the method of step S01 and step S31 for a servo motor of the same model that has been determined to be operating normally; (2) Deviation rate of maximum deviation value , wherein max0 is a maximum deviation reference value calculated by the method of step S01 and step S31 for a servo motor of the same model that has been determined to be operating normally; (3) One-quarter digit deviation rate , wherein Q0 is a quarter-digit reference value calculated by the method of step S01 and step S31 for a servo motor of the same model that has been determined to be operating normally; (4) Deviation rate of kurtosis value ; (5) Deviation rate of skewness value ; S33. For the deviation rate A0, B0, C0, D0, E0 of each indicator calculated in step S32, if the deviation rate is greater than 50%, it is considered that the indicator corresponding to the deviation rate is abnormal, otherwise it is normal. According to the abnormality or normality of each indicator and combined with the historical fault records, the fault type of the servo motor to be tested is determined.

Images