A method and system for identifying transmission faults

Abstract

This invention discloses a method and system for identifying transmission faults, comprising: simulating and determining transmission measurement points; acquiring vibration acceleration signals from the measurement points; extracting features from the vibration acceleration signals to obtain time-domain and frequency-domain indices, and then calculating the transmission's energy proportion characteristic parameters and the order characteristic parameters of each gear pair, thereby obtaining the transmission's characteristic parameter matrix and sensitive characteristic parameters; obtaining the transmission's comprehensive coefficients and defining faulty transmissions; constructing a weight factor matrix based on the transmission's sensitive characteristic parameters and characteristic parameter weight factors; using the normalized weight factor matrix of the identified faulty transmission types as a benchmark, calculating the PSI values ​​of other faulty transmissions; and classifying faults based on the PSI values ​​of the faulty transmissions. This invention, in judging transmission faults, does not rely on the experience of technical personnel, and has advantages such as unified evaluation standards, high efficiency, and easy problem tracing.

Description

Technical Field

[0001] This invention belongs to the field of gearbox technology, specifically relating to a gearbox fault identification method and system. Background Technology

[0002] Gearboxes are widely used in automobiles, construction machinery, and other fields. They are typically composed of pumps, gears, and bearings. However, due to deficiencies in their design, manufacturing, and assembly, defective products often reach the customer, resulting in economic losses from rework, returns, and manual troubleshooting. Therefore, it is necessary to monitor defective products before they reach the customer and screen out faulty samples through running tests and testing benches.

[0003] Vibration testing and analysis is one of the most commonly used methods in mechanical fault diagnosis. Vibration signals can identify the characteristic frequencies of faulty components, thereby quickly locating the faulty component and the type of fault. However, when faults occur in construction machinery, technicians still rely on their experience to determine the approximate fault area and gear position through subjective auditory means. Then, they use the controlled variable method to replace gears, bearings, or shafts at each stage, and finally confirm the faulty component and its cause through laboratory testing. Because manual judgment is highly subjective, and the transmission has a complex structure, many parts, and is difficult to disassemble and assemble manually, fault identification presents difficulties and challenges. Therefore, there is an urgent need for a test bench, method, and system that can automatically identify transmission faults and determine the source of the fault. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method and system for identifying transmission faults. When determining transmission faults, it does not rely on the experience of technical personnel and has advantages such as unified evaluation standards, high efficiency, and easy traceability of problems.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] Firstly, a method for identifying transmission faults is provided, comprising: performing deformation analysis on the transmission using an established dynamic simulation model to determine the transmission measurement points; acquiring vibration acceleration signals of the transmission, wherein the vibration acceleration signals are measured by vibration acceleration sensors, which are installed on an established transmission off-line testing run and test bench according to the determined transmission measurement points; extracting features from the acquired vibration acceleration signals of the transmission to obtain time-domain and frequency-domain indices; calculating the energy proportion characteristic parameters of the transmission and the order characteristic parameters of each gear pair of the transmission based on the transmission structure and the obtained time-domain and frequency-domain indices, thereby obtaining the characteristic parameter matrix of the transmission; obtaining the sensitive characteristic parameters of the transmission based on the characteristic parameter matrix of the transmission; obtaining the characteristic parameter weighting factors based on the sensitive characteristic parameters of the transmission, thereby obtaining the comprehensive coefficient of the transmission, and defining the transmission with a comprehensive coefficient exceeding a set threshold as a faulty transmission; constructing a weighting factor matrix based on the sensitive characteristic parameters and the characteristic parameter weighting factors of the transmission, thereby obtaining a normalized weighting factor matrix; calculating the PSI values ​​of other faulty transmissions based on the normalized weighting factor matrix of the fault types of the identified faulty transmissions, and classifying the faults based on the PSI values ​​of the faulty transmissions.

[0007] Furthermore, the method for determining the transmission measurement points includes: inputting the gear pitch error and circular runout error of the transmission under normal operating conditions as excitation sources into the established dynamic simulation model for deformation analysis to obtain the first vibration response of the transmission housing; inputting the gear pitch error and circular runout error of the transmission under fault conditions as excitation sources into the established dynamic simulation model for deformation analysis to obtain the second vibration response of the transmission housing; and selecting the points in the first vibration response and the second vibration response whose vibration response acceleration meets the set conditions as transmission measurement points.

[0008] Furthermore, the time-domain indicators include: peak value, peak-to-peak value, mean, variance, kurtosis, skewness, waveform indicator, margin indicator, peak value indicator, impulse indicator, kurtosis indicator, and root mean square value; the frequency-domain indicators include: input energy, low-frequency energy ratio, the sum of the first three orders of input gear meshing frequency, the sum of the first three orders of planetary gear meshing frequency, and the sum of the first three orders of output gear meshing frequency.

[0009] Furthermore, the energy proportion characteristic parameter of the gearbox is obtained through the following method: The vibration acceleration signal of the gearbox is subjected to Fourier transform. Based on the frequency domain characteristics of the research object, the fundamental frequency and harmonic energy of the pump gear that account for a set value of the energy signal are extracted. First, the characteristic frequency and harmonics of the pump gear are calculated. Then, the energy value of 1 / 3 bandwidth with the characteristic frequency of the pump gear as the center frequency is calculated to obtain the frequency domain pump gear energy and select it as the frequency domain pump energy characteristic parameter. Second, the frequency band is further divided into low frequency according to the bandwidth and the low frequency energy proportion is calculated and selected as the energy proportion characteristic parameter. Among them, the low frequency refers to the frequency of 0-1000Hz.

[0010] Furthermore, the order characteristic parameters of each gear pair in the gearbox are obtained by the following method: taking the speed of the output gear as a reference, the meshing frequency and harmonics of each gear pair are converted to obtain the order characteristics of each pair of gear pairs. The first three orders of each pair of gear pairs are selected and summed to obtain the order sum of each pair of gear pairs, which is used as the order characteristic parameters of each gear pair.

[0011] Furthermore, the characteristic parameter matrix of the gearbox is as follows:

[0012]

[0013] Where i represents the sample label, i.e., the number of gearboxes; m represents the number of channels for the gearbox vibration acceleration signal; n represents the number of feature parameters contained in each channel, X imn The result is the characteristic parameter of all gearboxes.

[0014] Furthermore, the sensitive characteristic parameters of the transmission are obtained based on the characteristic parameter matrix of the transmission, including:

[0015] If the characteristic parameters corresponding to the time-domain and frequency-domain indices of a certain non-faulty transmission are used to form the first diagonal matrix, then the inverse of the first diagonal matrix is:

[0016]

[0017] Among them, X mn The result is the characteristic parameter of a qualified transmission.

[0018] Multiplying the inverse of the first diagonal matrix by the characteristic parameter matrix of the gearbox yields the sensitivity coefficient matrix of the gearbox:

[0019]

[0020] Among them, X imn These are the characteristic parameter results for transmissions other than qualified transmissions.

[0021] Calculate the mean of the sensitivity coefficients in the sensitivity coefficient matrix of a set number of gearboxes, sort them from largest to smallest, and select t parameters whose sensitivity coefficients satisfy the set conditions as the sensitive feature parameters of the gearbox.

[0022] Furthermore, based on the sensitive characteristic parameters of the transmission, a feature parameter weighting factor is obtained, which in turn yields the transmission's comprehensive coefficient. Transmissions with a comprehensive coefficient exceeding a set threshold are defined as faulty transmissions, including:

[0023] Taking a faulty transmission as an example, the sensitive characteristic parameters of the faulty transmission are arranged into a second diagonal matrix:

[0024]

[0025] Among them, Y mn These are the characteristic parameter results for the faulty transmission.

[0026] Take the reciprocal of each feature parameter in the gearbox's feature parameter matrix and multiply it by the second diagonal matrix to obtain the weight coefficient matrix for each channel:

[0027]

[0028] Calculate the mean of the t weighted coefficients for i prototypes:

[0029]

[0030] Among them, Q n X represents the mean of the weight coefficients of the nth feature parameter. imn -1 *Y mn The result is the sum of each element in the weight coefficient matrix. Since each gearbox contains m channels, the mean is the sum of m*i elements.

[0031] The weight coefficients of the gearbox are calculated and normalized based on the weight coefficient matrix of each channel to obtain the feature parameter weight factor for each feature parameter:

[0032] Normalize the n weight coefficients to obtain the feature parameter weight factor for each feature parameter:

[0033]

[0034] Among them, Z n The weight factor represents the normalized feature parameter weighting factor for normalization.

[0035] The normalized characteristic parameter weighting factor is calculated for each characteristic parameter of other typical fault prototypes in turn. Then, the normalized characteristic parameter weighting factors of each characteristic parameter are averaged to obtain the final comprehensive characteristic parameter weighting factor B. n ;

[0036] The comprehensive feature parameter weighting factor of each channel's feature parameters is multiplied by the corresponding feature parameter and summed to obtain the comprehensive coefficient of each channel of the gearbox:

[0037]

[0038] Among them, F m This is the composite coefficient for the m-th channel;

[0039] A transmission whose overall coefficient exceeds a set threshold is defined as a faulty transmission.

[0040] Furthermore, a weight factor matrix is ​​constructed based on the sensitive characteristic parameters and weight factors of the transmission, thereby obtaining a normalized weight factor matrix. Using the normalized weight factor matrix of the identified faulty transmission types as a benchmark, the PSI values ​​of other faulty transmissions are calculated. Fault classification is then performed based on the PSI values ​​of the faulty transmissions, including:

[0041] Construct a weight factor matrix based on the sensitive feature parameters and feature parameter weight factors of the gearbox:

[0042]

[0043] The columns of the weight factor matrix are normalized to obtain a normalized weight factor matrix. Using the normalized weight factor matrix of the fault types of the identified faulty transmissions as a benchmark, the PSI values ​​of other faulty transmissions are calculated. Transmissions with PSI values ​​less than the set PSI benchmark value are grouped into one category. If the PSI values ​​of all fault types are greater than the set PSI benchmark value, they are classified as new fault samples and their fault types are analyzed and defined. Subsequent fault samples will use the new fault sample library as a benchmark to calculate the PSI, thereby realizing fault source identification.

[0044] Secondly, a transmission fault identification system is provided, comprising: a fault simulation analysis module for performing deformation analysis on the transmission using an established dynamic simulation model to determine transmission measurement points; a data acquisition module for acquiring vibration acceleration signals of the transmission, wherein the vibration acceleration signals of the transmission are measured by vibration acceleration sensors, which are installed on an established transmission off-line testing run and test bench according to the determined transmission measurement points; and a fault alarm module for extracting features from the acquired vibration acceleration signals of the transmission to obtain time-domain and frequency-domain indicators, and calculating the transmission's energy ratio characteristic parameters and shift parameters based on the transmission's structure and the obtained time-domain and frequency-domain indicators. The transmission's characteristic parameters are obtained by analyzing the order characteristic parameters of each gear pair in the gearbox, thus deriving the transmission's characteristic parameter matrix. Based on this matrix, the transmission's sensitive characteristic parameters are obtained. Then, the characteristic parameter weighting factors are derived from these sensitive characteristic parameters, leading to the transmission's overall coefficient. Transmissions with overall coefficients exceeding a set threshold are defined as faulty transmissions. The PSI fault source analysis module is used to construct a weighting factor matrix based on the transmission's sensitive characteristic parameters and characteristic parameter weighting factors, resulting in a normalized weighting factor matrix. Using the normalized weighting factor matrix of the identified faulty transmission types as a benchmark, the PSI values ​​of other faulty transmissions are calculated, and fault classification is performed based on the PSI values ​​of the faulty transmissions.

[0045] Compared with the prior art, the beneficial effects achieved by the present invention are as follows:

[0046] This invention employs an established dynamic simulation model to perform deformation analysis on the gearbox and determine the gearbox measurement points. It collects vibration acceleration signals from the gearbox, extracts features from these signals to obtain time-domain and frequency-domain indices, and calculates the gearbox's energy proportion characteristic parameters and the order characteristic parameters of each gear pair based on the gearbox's structure and the obtained time-domain and frequency-domain indices, thus obtaining the gearbox's characteristic parameter matrix. Based on the gearbox's characteristic parameter matrix, it obtains the gearbox's sensitive characteristic parameters. Based on these sensitive characteristic parameters, it obtains characteristic parameter weighting factors, thus obtaining the gearbox's comprehensive coefficient. Gearboxes with comprehensive coefficients exceeding a set threshold are defined as faulty gearboxes. Based on the gearbox's sensitive characteristic parameters and characteristic parameter weighting factors, it constructs a weighting factor matrix, thus obtaining a normalized weighting factor matrix. Using the normalized weighting factor matrix of the identified faulty gearboxes as a benchmark, it calculates the PSI values ​​of other faulty gearboxes and classifies faults based on the PSI values. In judging gearbox faults, it does not rely on the experience of technical personnel, offering advantages such as unified evaluation standards, high efficiency, and easy problem tracing. Attached Figure Description

[0047] Figure 1This is a simplified diagram of the gearbox transmission mechanism in an embodiment of the present invention;

[0048] Figure 2 This is a schematic diagram of the main process of a transmission fault identification method provided in an embodiment of the present invention;

[0049] Figure 3 This is a schematic diagram of the overall structure of the offline testing run and test bench in an embodiment of the present invention;

[0050] Figure 4 These are the results of the gearbox sensitivity coefficient analysis in the embodiments of the present invention;

[0051] Figure 5 This refers to the gearbox weighting factor results in the embodiments of the present invention;

[0052] Figure 6 This refers to the overall coefficient result of the gearbox in this embodiment of the invention;

[0053] Figure 7 These are the PSI fault source analysis results in the embodiments of the present invention;

[0054] Figure 8 This is a system block diagram of a gearbox fault identification system provided in an embodiment of the present invention. Detailed Implementation

[0055] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.

[0056] Example 1:

[0057] like Figures 1 to 7As shown, a method for identifying transmission faults includes: performing deformation analysis on the transmission using an established dynamic simulation model to determine transmission measurement points; acquiring vibration acceleration signals of the transmission, wherein the vibration acceleration signals are measured by vibration acceleration sensors, which are installed on an established transmission off-line testing run and test bench according to the determined transmission measurement points; extracting features from the acquired vibration acceleration signals of the transmission to obtain time-domain and frequency-domain indices; calculating the energy proportion characteristic parameters of the transmission and the order characteristic parameters of each gear pair of the transmission based on the transmission structure and the obtained time-domain and frequency-domain indices, thereby obtaining the characteristic parameter matrix of the transmission; obtaining the sensitive characteristic parameters of the transmission based on the characteristic parameter matrix; obtaining the characteristic parameter weighting factors based on the sensitive characteristic parameters of the transmission, thereby obtaining the comprehensive coefficient of the transmission, and defining the transmission with a comprehensive coefficient exceeding a set threshold as a faulty transmission; constructing a weighting factor matrix based on the sensitive characteristic parameters and characteristic parameter weighting factors of the transmission, thereby obtaining a normalized weighting factor matrix; calculating the PSI values ​​of other faulty transmissions based on the normalized weighting factor matrix of the fault types of the identified faulty transmissions, and classifying the faults based on the PSI values ​​of the faulty transmissions.

[0058] This embodiment uses Figure 1 Taking the gearbox shown as an example, the following technical solution is adopted to achieve this, such as... Figure 2 As shown.

[0059] The established dynamic simulation model was used to perform deformation analysis on the gearbox and determine the gearbox measurement points.

[0060] S1. Establish a dynamic simulation model of the transmission components.

[0061] S11. Create a 3D model of the gearbox housing in CAD software and export it as a .xt file.

[0062] S12. Import the 3D gearbox housing into Ansys and mesh it. Set the condensed nodes of the bearings and condense the stiffness and mass matrices on these nodes. Finally, solve the model to obtain its cdb file.

[0063] S13. Establish a dynamic analysis model of the gearbox in Masta. Import the gearbox housing cdb file from step S12 into the dynamic model and couple it at the condensed nodes to generate the condensed mass matrix and stiffness matrix. Establish the final bearing-gear-carrier gearbox dynamic model.

[0064] S2. Statistically analyze the distribution of common manufacturing errors in the transmission. Using the gear pitch error and circular runout error of the transmission under normal operating conditions as excitation sources, input them into the established dynamic simulation model and run advanced system deformation analysis to obtain the first vibration response of the transmission housing. Using the gear pitch error and circular runout error of the transmission under fault conditions as excitation sources, input them into the established dynamic simulation model and run advanced system deformation analysis to obtain the second vibration response of the transmission housing. The vibration responses of the transmission housing under these two conditions are obtained. Based on the vibration response acceleration results, four transmission measurement points with large responses and easy installation are initially selected. This invention proposes a transmission vibration monitoring position optimization method based on multibody dynamics simulation, solving the problem of difficult vibration sensor positioning and achieving accurate positioning of vibration measurement points.

[0065] S3, Establish as follows Figure 3 The transmission off-line testing run and test bench shown are composed of a drive system 1, a support system 2, a lifting system 3, a transmission 4, a load system 5, and a laser speed monitoring system 6.

[0066] S31, the drive system 1 includes: drive motor, motor bracket, connecting flange, etc., which can realize power drive, automatic clamping and omnidirectional movement.

[0067] S32. The support system includes: hydraulic components, a cast iron platform, and elastic vibration isolators. It provides hydraulic clamping and foundation installation functions for systems such as brackets, inputs, and supports. An elastic vibration isolator designed based on the environmental input fundamental frequency is located beneath the base, effectively isolating low-frequency vibrations; an oil and sludge drain is also provided below to effectively drain oil and sludge generated during gearbox operation. The cross-shaped base design facilitates quick and precise gearbox movement and also includes cylinders that provide power and hydraulic components for clamping various connected systems.

[0068] S33. The lifting system primarily transports the gearboxes from the production line to the appropriate testing position, enabling rapid and accurate positioning of the tested object. Once the gearbox running and testing are completed, the carrier system can lift the tested gearbox to the appropriate workstation for rapid transport out of the test bench.

[0069] S34. The laser speed monitoring system consists of a vibration sensor, a laser tachometer, and a cylinder. The vibration sensor is a probe-type accelerometer with an embedded elastic element. The cylinder automatically drives the probe accelerometer and laser tachometer to adhere to the monitored object, eliminating the inaccuracy caused by manual clamping. The offline testing platform proposed in this invention facilitates rapid and accurate identification of gearbox faults.

[0070] The vibration acceleration signal of the transmission is collected by a vibration acceleration sensor installed at predetermined measurement points on a test bench and off-line testing platform. Feature extraction is performed on the collected vibration acceleration signal to obtain time-domain and frequency-domain indices. Based on the transmission structure and the obtained time-domain and frequency-domain indices, the energy proportion characteristic parameters and order characteristic parameters of each gear pair are calculated, resulting in a characteristic parameter matrix. Sensitive characteristic parameters are obtained from the characteristic parameter matrix. Weighting factors are then derived from these sensitive characteristic parameters to obtain the overall coefficient of the transmission. Transmissions with an overall coefficient exceeding a set threshold are defined as faulty transmissions.

[0071] S4. Select characteristic parameters for the four selected transmission test points through testing.

[0072] S41 Figure 1 The transmission routes of each gear in the intermediate gearbox are shown in Table 1.

[0073] Table 1 Gear transmission route of the transmission

[0074] gear Gear transmission route F1 Z1 / Z2*(1 / (Z5 / Z3+1))*Z6 / Z7 F2 Z1 / Z2*Z6 / Z7 R1 Z1 / Z2*Z3 / Z4*Z4 / Z5*Z6 / Z7

[0075] As can be seen from the transmission routes of each gear, the system contains three gears. Acceleration signals in three directions from four positions of 100 machines are collected. Twelve time-domain indicators are extracted from each gearbox: peak value, peak-to-peak value, mean, variance, kurtosis, skewness, waveform indicator, margin indicator, peak value, impulse indicator, kurtosis indicator, and root mean square value; and five frequency-domain indicators: input energy, low-frequency energy ratio, sum of the first three orders of input gear meshing frequency, sum of the first three orders of planetary gear meshing frequency, and sum of the first three orders of output gear meshing frequency.

[0076] S42, due to the research subject ( Figure 1 The gearbox shown contains a hydraulic torque converter. The input speed of the pump gear is the driving speed of the engine on the test bench. The speed is known and relatively stable. The pump gear energy and energy proportion characteristic parameters can be selected by the following method: Perform Fourier transform on the vibration acceleration signal of the gearbox. According to the frequency domain characteristics of the research object, extract the pump gear fundamental frequency and harmonic energy, which account for a large proportion of the energy signal. First, calculate the characteristic frequency and harmonics of the pump gear. Then, calculate the energy value of 1 / 3 bandwidth with the center frequency as the center frequency to obtain the frequency domain pump gear energy and select it as the frequency domain pump energy characteristic parameter. Secondly, according to the bandwidth, further divide the frequency band into low frequency (0-1000Hz) and calculate the low frequency energy proportion and select it as the energy proportion characteristic parameter.

[0077] S43. The speeds of the remaining gears transmitted to the first-stage input gear after being changed by the hydraulic torque converter are not the same. Therefore, the speed of the output gear is monitored by the speed sensor. Based on this speed, the meshing frequency and harmonics of each gear pair are converted to obtain the order characteristics of each gear pair. The first three orders of each gear pair are selected and summed to obtain the order sum of each gear pair. This is used as the order characteristic parameter of each gear pair.

[0078] S44. Combining the above three steps, 17 characteristic parameters are determined (corresponding to 12 time-domain indicators and 5 frequency-domain indicators for each gearbox), resulting in a characteristic parameter matrix for 12 channels of 100 prototypes:

[0079]

[0080] Where i represents the sample label, i.e., the number of gearboxes, which is 100 in this case; m represents the number of channels for the vibration acceleration signal of the gearbox, which is 12 channels in this case; n represents the number of feature parameters contained in each channel, which is 17 feature parameters in this case; X imn The result is the characteristic parameter of all gearboxes.

[0081] S5, Alarm Threshold Extraction.

[0082] S51, Sensitive Factor Extraction.

[0083] S511. Select one qualified machine (non-faulty gearbox) from 100 machines, where all characteristic parameters are relatively good. Using this qualified machine as a benchmark, extract its 17 characteristic parameters and form a first diagonal matrix. The inverse of the first diagonal matrix is:

[0084]

[0085] Among them, X mn The characteristic parameter results are for a qualified transmission;

[0086] S512. Multiply the inverse of the first diagonal matrix from step S511 with the characteristic parameter matrix of the gearbox from step S44 to obtain the sensitivity coefficient matrix of the 12 channels of the gearbox:

[0087]

[0088] S513. Calculate the mean of the 17 sensitivity coefficients in the sensitivity coefficient matrix of 100 gearboxes:

[0089]

[0090] Because there are 12 channels, the average value is taken from 1200 parameters;

[0091] The results are as follows Figure 5As shown, the sensitivity coefficients are sorted from largest to smallest, and nine feature parameters with coefficients greater than 1.6 are selected as sensitive feature parameters: peak value, variance, root mean square value, sum of the first three orders of the planetary gears, sum of the first three orders of the output gears, kurtosis, low-frequency energy ratio, pump energy, and sum of the first three orders of the first-stage input gears; and the sensitive feature parameters of each of the other channels are calculated in turn.

[0092] S52, Definition of weighting factors.

[0093] S521. Select several typical fault prototypes from 100 machines. Taking one fault prototype as an example, extract its nine sensitive feature parameters and form a second diagonal matrix:

[0094]

[0095] Where n represents the number of feature parameters contained in each channel, i.e., the result of 9 feature parameters in the 12 channels of this faulty gearbox.

[0096] S522. Take the reciprocal of each feature parameter in the feature parameter matrix of the gearbox and multiply it with the second diagonal matrix in step S521 to obtain the weight coefficient matrix of each channel:

[0097]

[0098] S523. Calculate the mean of the nine weighting coefficients for 100 prototypes:

[0099]

[0100] Among them, Q n Y represents the mean of the nth weighting coefficient of 100 gearboxes. mn X represents the characteristic parameter results of the faulty machine. imn Results of various characteristic parameters for 100 gearboxes;

[0101] Normalize the nine weight coefficients to obtain the feature parameter weight factor for each feature parameter:

[0102]

[0103] Among them, Z n The feature parameter weighting factors represent the nine feature parameters;

[0104] The characteristic parameter weighting factor for each characteristic parameter is calculated sequentially for other typical fault prototypes. Then, the characteristic parameter weighting factors for each characteristic parameter are averaged to obtain the final comprehensive characteristic parameter weighting factor B. n The result is as follows Figure 6 As shown.

[0105] S53, Comprehensive Coefficient.

[0106] The overall coefficient for each channel of the gearbox is defined as follows:

[0107]

[0108] Among them, B n X is the comprehensive feature parameter weighting factor for the nth feature parameter. imn For characteristic parameters, F m This is the composite coefficient for the m-th channel;

[0109] The final distribution of the comprehensive coefficients is as follows: Figure 6 As shown, gearboxes with a comprehensive coefficient result greater than a set threshold (set to 3 in this embodiment) are defined as faulty gearboxes. Five faulty machines were identified here.

[0110] This invention solves the problems of highly sensitive feature parameter screening and weight factor definition, ensuring the accuracy of transmission alarm threshold definition and realizing automatic transmission fault alarm.

[0111] A weight factor matrix is ​​constructed based on the sensitive characteristic parameters and weighting factors of the transmission, thereby obtaining a normalized weight factor matrix. Using the normalized weight factor matrix of the fault types of the identified faulty transmissions as a benchmark, the PSI values ​​of other faulty transmissions are calculated, and faults are classified based on the PSI values ​​of the faulty transmissions. This invention establishes a transmission fault identification process and realizes quantitative analysis of fault alarms.

[0112] S6. Fault source clustering analysis based on PSI.

[0113] Based on the characteristics of the fault types, the faults of this gearbox and test bench can be divided into three categories: pump faults, input gear pair faults, and output gear pair faults.

[0114] S61: Weighting factor matrix.

[0115] The weight factor matrix C is formed by multiplying each element of the sensitivity coefficient matrix (composed of the nine sensitive feature parameters) with the weight factor of the comprehensive feature parameters.

[0116]

[0117] Among them, B n X is the comprehensive feature parameter weighting factor for the nth feature parameter. mn X represents the characteristic parameter result of a qualified machine. imn The result is the characteristic parameter of 100 machines;

[0118] S61, Normalized weight factor matrix.

[0119] The column values ​​of the weight factor matrix are normalized to obtain the normalized weight factor matrix for each column. Taking the five identified faulty transmissions, the PSI is used to calculate the three different fault types to which these five faulty transmissions belong. Based on the normalized weight factors of these three fault types, the PSI results of the three benchmarks are calculated for other faulty transmissions with a comprehensive coefficient greater than 3. PSI results less than 0.2 are grouped into one category. If the PSI results of all three categories are greater than 0.2, the prototype is automatically classified as a new fault sample and its fault type is analyzed and defined. Subsequent faulty prototypes will use the new fault sample library as a benchmark to calculate the PSI, thereby realizing fault source identification.

[0120] The sample contained three fault types: output gear pair 2 fault, pump 88 fault, and input gear pair 100 fault. The other two fault samples, 1 and 74, both belonged to pump faults. Using PSI and <0.2 fault source clustering methods, the fault types were accurately identified, as shown in the results. Figure 7 As shown.

[0121] The application of the PSI fault source identification method in this invention solves the problem of fault source identification in a transmission fault prototype and realizes automatic classification of fault samples.

[0122] Example 2:

[0123] Based on the gearbox fault identification method described in Embodiment 1, this embodiment provides a gearbox fault identification system, such as... Figure 8 As shown, it includes:

[0124] The fault simulation analysis module is used to perform deformation analysis on the gearbox using the established dynamic simulation model and determine the gearbox measurement points;

[0125] The data acquisition module is used to acquire the vibration acceleration signal of the gearbox. The vibration acceleration signal of the gearbox is measured by a vibration acceleration sensor, which is installed on the established gearbox off-line testing run and test bench according to the determined gearbox measuring points.

[0126] The fault alarm module is used to extract features from the collected vibration acceleration signals of the gearbox, obtain time-domain and frequency-domain indicators, and calculate the energy proportion characteristic parameters and order characteristic parameters of each gear pair of the gearbox based on the gearbox structure and the obtained time-domain and frequency-domain indicators, thereby obtaining the characteristic parameter matrix of the gearbox; obtain the sensitive characteristic parameters of the gearbox based on the characteristic parameter matrix of the gearbox; obtain the characteristic parameter weighting factor based on the sensitive characteristic parameters of the gearbox, thereby obtaining the comprehensive coefficient of the gearbox, and define the gearbox with a comprehensive coefficient exceeding a set threshold as a faulty gearbox;

[0127] The PSI fault source analysis module is used to construct a weight factor matrix based on the sensitive characteristic parameters and characteristic parameter weight factors of the transmission, and then obtain a normalized weight factor matrix. Based on the normalized weight factor matrix of the fault types of the identified faulty transmissions, the PSI values ​​of other faulty transmissions are calculated, and the faults are classified according to the PSI values ​​of the faulty transmissions.

[0128] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for identifying transmission faults, characterized in that, include: The established dynamic simulation model was used to perform deformation analysis on the gearbox and determine the gearbox measurement points; The vibration acceleration signal of the gearbox is collected. The vibration acceleration signal of the gearbox is measured by a vibration acceleration sensor. The vibration acceleration sensor is installed on the established gearbox off-line testing run and test bench according to the determined gearbox measuring points. Feature extraction is performed on the collected vibration acceleration signal of the gearbox to obtain time-domain and frequency-domain indices. Based on the gearbox structure and the obtained time-domain and frequency-domain indices, the energy proportion characteristic parameters of the gearbox and the order characteristic parameters of each gear pair of the gearbox are calculated to obtain the characteristic parameter matrix of the gearbox. The sensitive characteristic parameters of the gearbox are obtained based on the characteristic parameter matrix of the gearbox. The characteristic parameter weighting factor is obtained based on the sensitive characteristic parameters of the gearbox to obtain the comprehensive coefficient of the gearbox. The gearbox with a comprehensive coefficient exceeding a set threshold is defined as a faulty gearbox. A weight factor matrix is ​​constructed based on the sensitive characteristic parameters and weight factors of the transmission, thus obtaining a normalized weight factor matrix. Using the normalized weight factor matrix of the fault types of the identified faulty transmissions as a benchmark, the PSI values ​​of other faulty transmissions are calculated. Fault classification is then performed based on the PSI values ​​of the faulty transmissions, including: Construct a weight factor matrix based on the sensitive feature parameters and feature parameter weight factors of the gearbox: （10） Where i represents the sample label, i.e. the number of gearboxes; m represents the number of channels of the vibration acceleration signal of the gearbox; and n represents the number of feature parameters contained in each channel. The result is the characteristic parameter of all gearboxes; The characteristic parameter results are for a qualified transmission; The final comprehensive feature parameter weighting factor; The columns of the weight factor matrix are normalized to obtain a normalized weight factor matrix. Using the normalized weight factor matrix of the fault types of the identified faulty transmissions as a benchmark, the PSI values ​​of other faulty transmissions are calculated. Transmissions with PSI values ​​less than the set PSI benchmark value are grouped into one category. If the PSI values ​​of all fault types are greater than the set PSI benchmark value, they are classified as new fault samples and their fault types are analyzed and defined. Subsequent fault samples will use the new fault sample library as a benchmark to calculate the PSI, thereby realizing fault source identification. The method for determining the transmission measurement points includes: The gear pitch error and circular runout error of the gearbox under normal operating conditions are used as excitation sources and input into the established dynamic simulation model for deformation analysis to obtain the first vibration response of the gearbox housing. The gear pitch error and circular runout error of the gearbox under fault conditions are used as excitation sources and input into the established dynamic simulation model for deformation analysis to obtain the second vibration response of the gearbox housing. The points in the first and second vibration responses where the vibration response accelerations meet the set conditions are taken as the gearbox measurement points.

2. The transmission fault identification method according to claim 1, characterized in that, The time-domain metrics include: peak value, peak-to-peak value, mean, variance, kurtosis, skewness, waveform metrics, margin metrics, peak value, impulse metrics, kurtosis metrics, and root mean square value; the frequency-domain metrics include: input energy, low-frequency energy ratio, the sum of the first three orders of input gear meshing frequency, the sum of the first three orders of planetary gear meshing frequency, and the sum of the first three orders of output gear meshing frequency.

3. The transmission fault identification method according to claim 1, characterized in that, The energy ratio characteristic parameters of the transmission are obtained through the following method: The vibration acceleration signal of the gearbox is subjected to Fourier transform. Based on the frequency domain characteristics of the research object, the fundamental frequency and harmonic energy of the pump gear that account for a set value of the energy signal are extracted. First, the characteristic frequency and harmonics of the pump gear are calculated. Then, the energy value of 1 / 3 bandwidth with the characteristic frequency of the pump gear as the center frequency is calculated to obtain the frequency domain pump gear energy and select it as the frequency domain pump energy characteristic parameter. Second, the frequency band is further divided into low frequency according to the bandwidth and the proportion of low frequency energy is calculated and selected as the energy proportion characteristic parameter. Among them, low frequency refers to the frequency of 0-1000Hz.

4. The transmission fault identification method according to claim 3, characterized in that, The order characteristic parameters of each gear pair in the transmission are obtained through the following method: Using the speed of the output gear as a reference, the meshing frequency and harmonics of each gear pair are converted to obtain the order characteristics of each gear pair. The first three orders of each gear pair are selected and summed to obtain the order sum of each gear pair, which is used as the order characteristic parameter of each gear pair.

5. The transmission fault identification method according to claim 4, characterized in that, The characteristic parameter matrix of the gearbox is: （1）。 6. The transmission fault identification method according to claim 5, characterized in that, The sensitive characteristic parameters of the transmission are obtained from the characteristic parameter matrix of the transmission, including: If the characteristic parameters corresponding to the time-domain and frequency-domain indices of a certain non-faulty transmission are used to form the first diagonal matrix, then the inverse of the first diagonal matrix is: （2） Multiplying the inverse of the first diagonal matrix by the characteristic parameter matrix of the gearbox yields the sensitivity coefficient matrix of the gearbox: （3） Calculate the mean of the sensitivity coefficients in the sensitivity coefficient matrix of a set number of gearboxes, sort them from largest to smallest, and select t parameters whose sensitivity coefficients satisfy the set conditions as the sensitive feature parameters of the gearbox.

7. The transmission fault identification method according to claim 6, characterized in that, Based on the sensitive characteristic parameters of the transmission, the characteristic parameter weighting factors are obtained, and then the comprehensive coefficient of the transmission is obtained. Transmissions with a comprehensive coefficient exceeding a set threshold are defined as faulty transmissions, including: Taking a faulty transmission as an example, the sensitive characteristic parameters of the faulty transmission are arranged into a second diagonal matrix: （5） in, The characteristic parameter results for the faulty transmission; Take the reciprocal of each feature parameter in the gearbox's feature parameter matrix and multiply it by the second diagonal matrix to obtain the weight coefficient matrix for each channel: （6） Calculate the mean of the t weighted coefficients for i prototypes: （7） in, This represents the mean of the weight coefficients of the nth feature parameter. The result represents the elements in the weight coefficient matrix. Since each gearbox contains m channels, the mean is calculated as m. The sum of i elements; The weight coefficients of the gearbox are calculated and normalized based on the weight coefficient matrix of each channel to obtain the feature parameter weight factor for each feature parameter: Normalize the n weight coefficients to obtain the feature parameter weight factor for each feature parameter: （8） in, The weight factor represents the normalized feature parameter weighting factor for normalization. The normalized characteristic parameter weighting factor is calculated for each characteristic parameter of other typical fault prototypes in turn. Then, the normalized characteristic parameter weighting factors of each characteristic parameter are averaged to obtain the final comprehensive characteristic parameter weighting factor. ; The comprehensive feature parameter weighting factor of each channel's feature parameters is multiplied by the corresponding feature parameter and summed to obtain the comprehensive coefficient of each channel of the gearbox: （9） in, This is the composite coefficient for the m-th channel; A transmission whose overall coefficient exceeds a set threshold is defined as a faulty transmission.

8. A transmission fault identification system, characterized in that, include: The fault simulation analysis module is used to perform deformation analysis on the gearbox using the established dynamic simulation model and determine the gearbox measurement points; The data acquisition module is used to acquire the vibration acceleration signal of the gearbox. The vibration acceleration signal of the gearbox is measured by a vibration acceleration sensor, which is installed on the established gearbox off-line testing run and test bench according to the determined gearbox measuring points. The fault alarm module is used to extract features from the collected vibration acceleration signals of the gearbox, obtain time-domain and frequency-domain indicators, and calculate the energy proportion characteristic parameters and order characteristic parameters of each gear pair of the gearbox based on the gearbox structure and the obtained time-domain and frequency-domain indicators, thereby obtaining the characteristic parameter matrix of the gearbox; obtain the sensitive characteristic parameters of the gearbox based on the characteristic parameter matrix of the gearbox; obtain the characteristic parameter weighting factor based on the sensitive characteristic parameters of the gearbox, thereby obtaining the comprehensive coefficient of the gearbox, and define the gearbox with a comprehensive coefficient exceeding a set threshold as a faulty gearbox; The PSI fault source analysis module is used to construct a weight factor matrix based on the transmission's sensitive characteristic parameters and characteristic parameter weighting factors, thereby obtaining a normalized weight factor matrix. Using the normalized weight factor matrix of the identified faulty transmission types as a benchmark, it calculates the PSI values ​​of other faulty transmissions and classifies faults based on the PSI values ​​of the faulty transmissions, including: Construct a weight factor matrix based on the sensitive feature parameters and feature parameter weight factors of the gearbox: （10） Where i represents the sample label, i.e. the number of gearboxes; m represents the number of channels of the vibration acceleration signal of the gearbox; and n represents the number of feature parameters contained in each channel. The result is the characteristic parameter of all gearboxes; The characteristic parameter results are for a qualified transmission; The final comprehensive feature parameter weighting factor; The columns of the weight factor matrix are normalized to obtain a normalized weight factor matrix. Using the normalized weight factor matrix of the fault types of the identified faulty transmissions as a benchmark, the PSI values ​​of other faulty transmissions are calculated. Transmissions with PSI values ​​less than the set PSI benchmark value are grouped into one category. If the PSI values ​​of all fault types are greater than the set PSI benchmark value, they are classified as new fault samples and their fault types are analyzed and defined. Subsequent fault samples will use the new fault sample library as a benchmark to calculate the PSI, thereby realizing fault source identification. The method for determining the transmission measurement points includes: The gear pitch error and circular runout error of the gearbox under normal operating conditions are used as excitation sources and input into the established dynamic simulation model for deformation analysis to obtain the first vibration response of the gearbox housing. The gear pitch error and circular runout error of the gearbox under fault conditions are used as excitation sources and input into the established dynamic simulation model for deformation analysis to obtain the second vibration response of the gearbox housing. The points in the first and second vibration responses where the vibration response accelerations meet the set conditions are taken as the gearbox measurement points.

Images