Subway traction motor bearing fault identification method, device and product

Abstract

The application discloses a kind of subway traction motor bearing fault identification method, device and product, it is related to the technical field of fault identification.The vibration acceleration signal of subway traction motor is obtained;The vibration acceleration signal is divided into multiple sections, and multiple sub-signal sections are obtained;Multiple time-frequency distribution matrices are obtained by time-frequency transformation to multiple sub-signal sections;Based on the energy density value of the matrix element corresponding to each frequency, the energy amplitude spectrum entropy corresponding to each frequency in each time-frequency distribution matrix is calculated;Based on the energy amplitude spectrum entropy corresponding to all frequencies in each time-frequency distribution matrix, the energy amplitude spectrum entropy sequence corresponding to multiple time-frequency distribution matrices is determined;Energy amplitude spectrum entropy sequence is used as the input of pre-trained fault identification model for fault detection, and the fault identification result of subway traction motor is obtained.The subway traction motor bearing fault identification method, device and product disclosed in the application can improve the accuracy of subway traction motor bearing fault detection.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of fault identification, and particularly relates to a metro traction motor bearing fault identification method, device and product. BACKGROUND

[0002] The metro traction motor bearing is a core component of the train power system, and its running state directly affects the train safety and operation efficiency. Since the metro traction motor bearing is one of the core components most prone to damage in the train power system, a large part of train failures are caused by traction motor bearing failures. Therefore, metro traction motor bearing fault diagnosis is an important work in train operation and maintenance.

[0003] In the traditional fault diagnosis method, the signal kurtosis analysis is widely used due to its simple calculation and no need for complex modeling. The method realizes abnormal detection by quantifying the sharpness of the impact component in the vibration signal. However, this method is very sensitive to abnormal data caused by accidental factors. In the healthy state of the traction motor bearing, the kurtosis value is also likely to exceed the threshold due to noise disturbance, thereby easily causing misjudgment.

[0004] Therefore, how to provide an effective scheme to improve the accuracy of metro traction motor bearing fault detection has become a difficult problem to be solved in the prior art. SUMMARY

[0005] The purpose of the application is to provide a metro traction motor bearing fault identification method, device and product to solve the above problems existing in the prior art.

[0006] In order to achieve the above purpose, the application adopts the following technical scheme:

[0007] In a first aspect, the application provides a metro traction motor bearing fault identification method, comprising:

[0008] obtaining a vibration acceleration signal of the metro traction motor;

[0009] dividing the vibration acceleration signal into multiple segments with a specified time length to obtain multiple sub-signal segments;

[0010] performing time-frequency transformation on the multiple sub-signal segments to obtain multiple time-frequency distribution matrices corresponding to the multiple sub-signal segments one by one;

[0011] Based on the energy density value of each frequency corresponding to the matrix element in each time-frequency distribution matrix, the energy amplitude spectrum entropy corresponding to each frequency in each time-frequency distribution matrix is calculated, which is used to represent the energy distribution frequency domain complexity of the matrix element corresponding to the same frequency. The greater the energy amplitude spectrum entropy, the greater the energy distribution frequency domain complexity of the matrix element corresponding to the same frequency. The smaller the energy amplitude spectrum entropy, the smaller the energy distribution frequency domain complexity of the matrix element corresponding to the same frequency.

[0012] Based on the energy amplitude spectrum entropy corresponding to all frequencies in each time-frequency distribution matrix, the energy amplitude spectrum entropy sequence corresponding to the plurality of time-frequency distribution matrices is determined.

[0013] The energy amplitude spectrum entropy sequence is used as the input of the pre-trained fault recognition model for fault detection, and the fault recognition result of the subway traction motor is obtained.

[0014] The fault recognition model is trained by taking the energy amplitude spectrum entropy sequence of the plurality of sample time-frequency distribution matrices corresponding to the plurality of sample sub-signal segments divided by the historical sample vibration acceleration signal of the sample subway traction motor as the sample input, and taking the fault classification result of the sample subway traction motor when the historical sample vibration acceleration signal is detected as the output.

[0015] Based on the above disclosure, the application provides a new scheme for accurately detecting the fault of the subway traction motor bearing, that is, obtaining the vibration acceleration signal of the subway traction motor; dividing the vibration acceleration signal into multiple segments of a specified time length to obtain multiple sub-signal segments; performing time-frequency transformation on the multiple sub-signal segments to obtain multiple time-frequency distribution matrices corresponding to the multiple sub-signal segments one by one; based on the energy density value of the matrix element corresponding to each frequency in each time-frequency distribution matrix, calculating the energy amplitude spectrum entropy corresponding to each frequency in each time-frequency distribution matrix, the energy amplitude spectrum entropy is used to represent the energy distribution frequency domain complexity of the matrix element corresponding to the same frequency, the greater the energy amplitude spectrum entropy, the greater the energy distribution frequency domain complexity of the matrix element corresponding to the same frequency, and the smaller the energy amplitude spectrum entropy, the smaller the energy distribution frequency domain complexity of the matrix element corresponding to the same frequency; based on the energy amplitude spectrum entropy corresponding to all frequencies in each time-frequency distribution matrix, determining the energy amplitude spectrum entropy sequence corresponding to the multiple time-frequency distribution matrices; taking the energy amplitude spectrum entropy sequence as the input of the pre-trained fault recognition model for fault detection to obtain the fault recognition result of the subway traction motor. In this way, through time-frequency transformation and calculation of energy amplitude spectrum entropy, since the complexity of signal frequency component change of normal and fault bearings is different, the energy amplitude spectrum entropy is also different, which can be used for fault recognition of the subway traction motor bearing, and the influence of abnormal data caused by accidental factors on the complexity of signal frequency component change is small, the influence of abnormal data caused by accidental factors can be reduced, thereby reducing the misjudgment of the bearing state and improving the accuracy of the subway traction motor bearing fault detection.

[0016] In one possible design, the time-frequency transformation on the multiple sub-signal segments to obtain multiple time-frequency distribution matrices corresponding to the multiple sub-signal segments one by one includes:

[0017] The time-frequency transformation on the multiple sub-signal segments is performed by short-time Fourier transform to obtain multiple time-frequency distribution matrices corresponding to the multiple sub-signal segments one by one.

[0018] In one possible design, the calculation of the energy amplitude spectrum entropy corresponding to each frequency in each time-frequency distribution matrix based on the energy density value of the matrix element corresponding to each frequency in each time-frequency distribution matrix includes:

[0019] The energy amplitude spectrum entropy corresponding to each frequency in each time-frequency distribution matrix is calculated based on the energy density value of the matrix element corresponding to each frequency in each time-frequency distribution matrix according to the following formula (1):

[0020]

[0021] Wherein, H represents the energy amplitude spectrum entropy corresponding to a certain frequency in the time-frequency distribution matrix, N represents the total number of matrix elements corresponding to a certain frequency, X(i) represents the normalized value of the energy density value corresponding to the i-th matrix element corresponding to a certain frequency, i is a positive integer less than or equal to N.

[0022] In one possible design, the determination of the energy amplitude spectrum entropy sequence corresponding to the plurality of time-frequency distribution matrices based on the energy amplitude spectrum entropy corresponding to all frequencies in each time-frequency distribution matrix comprises:

[0023] The minimum value of the energy amplitude spectrum entropy corresponding to all frequencies in each time-frequency distribution matrix is selected to obtain the energy amplitude spectrum entropy sequence corresponding to the plurality of time-frequency distribution matrices.

[0024] In one possible design, the fault detection by taking the energy amplitude spectrum entropy sequence as the input of the pre-trained fault identification model comprises:

[0025] The energy amplitude spectrum entropy sequence is normalized and then taken as the input of the pre-trained fault identification model for fault detection.

[0026] In one possible design, after the vibration acceleration signal of the metro traction motor is acquired, the method further comprises:

[0027] The vibration acceleration signal is filtered.

[0028] In one possible design, the fault identification model is a convolutional neural network model or a recurrent neural network model.

[0029] In a second aspect, the present application provides a metro traction motor bearing fault identification device, comprising:

[0030] An acquisition unit is configured to acquire a vibration acceleration signal of a metro traction motor.

[0031] A division unit is configured to divide the vibration acceleration signal into a plurality of segments with a specified time length to obtain a plurality of sub-signal segments.

[0032] A time-frequency transformation unit is configured to perform time-frequency transformation on the plurality of sub-signal segments to obtain a plurality of time-frequency distribution matrices corresponding one-to-one to the plurality of sub-signal segments.

[0033] The entropy calculation unit is configured to calculate an energy amplitude spectrum entropy corresponding to each frequency in each time-frequency distribution matrix based on an energy density value of a matrix element corresponding to the frequency, the energy amplitude spectrum entropy being used to represent a frequency domain complexity of energy distribution of the matrix element corresponding to the same frequency, and the greater the energy amplitude spectrum entropy, the greater the frequency domain complexity of energy distribution of the matrix element corresponding to the same frequency, and the smaller the energy amplitude spectrum entropy, the smaller the frequency domain complexity of energy distribution of the matrix element corresponding to the same frequency;

[0034] The determination unit is configured to determine an energy amplitude spectrum entropy sequence corresponding to the plurality of time-frequency distribution matrices based on the energy amplitude spectrum entropy corresponding to all frequencies in each time-frequency distribution matrix.

[0035] The fault detection unit is configured to perform fault detection by taking the energy amplitude spectrum entropy sequence as an input of a pre-trained fault recognition model, and obtain a fault recognition result of the metro traction motor.

[0036] The fault recognition model is trained by taking, as sample input, an energy amplitude spectrum entropy sequence of a plurality of sample time-frequency distribution matrices corresponding to a plurality of sample sub-signal segments divided from a historical sample vibration acceleration signal of a sample metro traction motor, and taking, as output, a fault classification result of the sample metro traction motor when the historical sample vibration acceleration signal is detected.

[0037] In a third aspect, the present application provides an electronic device, comprising a memory, a processor and a transceiver connected in sequence and in communication, wherein the memory is configured to store a computer program, the transceiver is configured to transceive messages, and the processor is configured to read the computer program and execute the metro traction motor bearing fault recognition method according to the first aspect or any possible design of the first aspect.

[0038] In a fourth aspect, the present application provides a computer readable storage medium, wherein instructions are stored on the computer readable storage medium, and when the instructions are run on a computer, the metro traction motor bearing fault recognition method according to the first aspect or any possible design of the first aspect is executed.

[0039] In a fifth aspect, the present application provides a computer program product comprising instructions, which, when executed on a computer, cause the computer to execute the metro traction motor bearing fault recognition method according to the first aspect or any possible design of the first aspect.

[0040] Advantages:

[0041] The subway traction motor bearing fault identification method, device and product provided by the application, through time-frequency conversion and calculation of energy amplitude spectrum entropy, since the complexity of signal frequency component change of the bearing in normal and fault states is different, the energy amplitude spectrum entropy is also different, thereby can be used for fault identification of the subway traction motor bearing, meanwhile, the abnormal data caused by accidental factors has less influence on the complexity of signal frequency component change, the influence of the abnormal data caused by accidental factors can be reduced, thereby reducing the misjudgment of the bearing state, improving the accuracy of the subway traction motor bearing fault detection, facilitating practical application and promotion. BRIEF DESCRIPTION OF DRAWINGS

[0042] Figure 1 The flowchart of the subway traction motor bearing fault identification method provided by the embodiment of the application is shown in the figure.

[0043] Figure 2 The block diagram schematic view of the subway traction motor bearing fault identification device provided by the embodiment of the application is shown in the figure.

[0044] Figure 3 The block diagram schematic view of the electronic device provided by the embodiment of the application is shown in the figure. DETAILED DESCRIPTION

[0045] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the present application will be briefly introduced with reference to the drawings and the description of the embodiments or the prior art. Obviously, the following description of the drawings is only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained without creative labor under the premise of the drawings. It should be noted that the description of these embodiments is used to help understand the present application, but does not constitute a limitation on the present application.

[0046] It should be understood that although the terms first, second, etc. can be used herein to describe various units, these units should not be limited by these terms. These terms are only used to distinguish one unit from another unit. For example, a first unit can be called a second unit, and similarly a second unit can be called a first unit, without departing from the scope of the example embodiments of the present application.

[0047] It should be understood that, for the term "and / or" that can appear in the present text, it is only a description of the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B, which can represent: the existence of A alone, the existence of B alone, and the existence of A and B at the same time; for the term " / and" that can appear in the present text, it is another description of the association relationship of another associated object, which means that there can be two relationships, for example, A / and B, which can represent: the existence of A alone, and the existence of A and B; in addition, for the character " / " that can appear in the present text, it generally represents that the associated objects before and after are an "or" relationship.

[0048] Embodiments

[0049] As Figure 1 shown, the subway traction motor bearing fault identification method provided in the first aspect of the embodiment can be executed by a computer device with certain computing resources, such as a cloud server, an edge computer configured with a GPU, a personal computer (PC, which refers to a multi-purpose computer suitable for personal use in size, price and performance; desktop computers, notebook computers to small notebook computers and tablet computers, and ultrabooks, etc. all belong to personal computers), a smart phone, a personal digital assistant (PDA), or a wearable device, etc. electronic device. As Figure 1 shown, the subway traction motor bearing fault identification method can include the following steps S101-S106.

[0050] Step S101. Obtain the vibration acceleration signal of the subway traction motor.

[0051] In the embodiment of the present application, an acceleration sensor can be arranged on or near the shell of the subway traction motor, and the vibration acceleration signal of the subway traction motor during operation can be detected through the acceleration sensor.

[0052] In one or more embodiments, after obtaining the vibration acceleration signal of the subway traction motor, the obtained vibration acceleration signal can be filtered.

[0053] Step S102. Divide the vibration acceleration signal into multiple segments of a specified time length to obtain multiple sub-signal segments.

[0054] The specified time length can be set according to the actual situation, for example, the specified time length can be 0.4s, 0.8s or 1s, etc.

[0055] Step S103. Perform time-frequency transformation on the multiple sub-signal segments to obtain multiple time-frequency distribution matrices corresponding to the multiple sub-signal segments.

[0056] In the embodiments of the present application, the short-time Fourier transform (SIFT) can be used to perform time-frequency transformation on the multiple sub-signal segments to obtain multiple time-frequency distribution matrices corresponding to the multiple sub-signal segments. It can be understood that in some other embodiments, other methods such as wavelet transform (WT) or Hilbert-Huang Transform can also be used to perform time-frequency transformation on the multiple sub-signal segments.

[0057] The time-frequency transformation can convert the sub-signal segment into a time-frequency distribution matrix of the signal, which can be expressed as where M represents the number of frequencies (the number of frequency points), C represents the number of matrix elements corresponding to the same frequency, r m,c represents the cth matrix element in the mth frequency. C = N / L, N represents the number of signals in the sub-signal segment, and L represents the step length of the window function along the time axis.

[0058] Step S104. Based on the energy density value of the matrix element corresponding to each frequency in each time-frequency distribution matrix, the energy amplitude spectrum entropy corresponding to each frequency in each time-frequency distribution matrix is calculated.

[0059] The energy amplitude spectrum entropy is used to represent the energy distribution frequency domain complexity of the matrix element corresponding to the same frequency. The greater the energy amplitude spectrum entropy, the greater the energy distribution frequency domain complexity of the matrix element corresponding to the same frequency. The smaller the energy amplitude spectrum entropy, the smaller the energy distribution frequency domain complexity of the matrix element corresponding to the same frequency.

[0060] The more complex the signal frequency components, that is, the more the signal energy is distributed in numerous frequencies, the greater the energy amplitude spectrum entropy. Especially when the signal energy is uniformly distributed in the entire frequency band, the energy amplitude spectrum entropy takes the maximum value 1. The simpler the signal frequency components, that is, the more the signal energy is concentrated in a few frequency components, the smaller the energy amplitude spectrum entropy. Especially when the signal energy is concentrated in a certain frequency, the energy amplitude spectrum entropy takes the minimum value 0.

[0061] In the embodiments of the present application, based on the energy density value of the matrix element corresponding to each frequency in each time-frequency distribution matrix, the energy amplitude spectrum entropy corresponding to each frequency in each time-frequency distribution matrix can be calculated according to the following formula (1);

[0062]

[0063] Wherein, H represents the energy amplitude spectrum entropy corresponding to a certain frequency in the time-frequency distribution matrix, N represents the total number of matrix elements corresponding to a certain frequency, X(i) represents the normalized value of the energy density value corresponding to the i-th matrix element corresponding to a certain frequency, i is a positive integer less than or equal to N.

[0064] Step S105. Based on the energy amplitude spectrum entropy corresponding to all frequencies in each time-frequency distribution matrix, the energy amplitude spectrum entropy sequence corresponding to the plurality of time-frequency distribution matrices is determined.

[0065] In the embodiments of the present application, the minimum value of the energy amplitude spectrum entropy corresponding to all frequencies in each time-frequency distribution matrix can be selected, and the selected energy amplitude spectrum entropy is combined in the order of the time-frequency distribution matrix to obtain the energy amplitude spectrum entropy sequence corresponding to the plurality of time-frequency distribution matrices.

[0066] It can be understood that in other embodiments, the median value or average value of the energy amplitude spectrum entropy corresponding to all frequencies in each time-frequency distribution matrix can also be selected and combined in time sequence to obtain the energy amplitude spectrum entropy sequence corresponding to the plurality of time-frequency distribution matrices.

[0067] Step S106. The energy amplitude spectrum entropy sequence is used as the input of the pre-trained fault recognition model for fault detection to obtain the fault recognition result of the subway traction motor.

[0068] Specifically, the energy amplitude spectrum entropy sequence can be normalized and used as the input of the pre-trained fault recognition model for fault detection to obtain the fault recognition result of the subway traction motor.

[0069] Wherein, the fault recognition model can be trained by taking the energy amplitude spectrum entropy sequence of the plurality of sample time-frequency distribution matrices corresponding to the plurality of sample sub-signal segments divided by the historical sample vibration acceleration signal of the sample subway traction motor as the sample input, and taking the fault classification result of the sample subway traction motor when detecting the historical sample vibration acceleration signal as the output.

[0070] The fault recognition model can be but is not limited to a convolutional neural network (CNN) model or a recurrent neural network (RNN) model, etc., which is not specifically limited in the embodiments of the present application.

[0071] In summary, the subway traction motor bearing fault identification method provided by the application obtains the vibration acceleration signal of the subway traction motor; divides the vibration acceleration signal into multiple segments of a specified time length to obtain multiple sub-signal segments; performs time-frequency transformation on the multiple sub-signal segments to obtain multiple time-frequency distribution matrices corresponding to the multiple sub-signal segments one by one; based on the energy density value of the matrix element corresponding to each frequency in each time-frequency distribution matrix, the energy amplitude spectrum entropy corresponding to each frequency in each time-frequency distribution matrix is calculated, the energy amplitude spectrum entropy is used to represent the energy distribution frequency domain complexity of the matrix element corresponding to the same frequency, the greater the energy amplitude spectrum entropy, the greater the energy distribution frequency domain complexity of the matrix element corresponding to the same frequency, and the smaller the energy amplitude spectrum entropy, the smaller the energy distribution frequency domain complexity of the matrix element corresponding to the same frequency; based on the energy amplitude spectrum entropy corresponding to all frequencies in each time-frequency distribution matrix, the energy amplitude spectrum entropy sequence corresponding to the multiple time-frequency distribution matrices is determined; the energy amplitude spectrum entropy sequence is used as the input of the pre-trained fault identification model for fault detection to obtain the fault identification result of the subway traction motor. In this way, through time-frequency transformation and calculation of energy amplitude spectrum entropy, since the complexity of signal frequency component change of normal and fault bearings is different, the energy amplitude spectrum entropy is also different, which can be used for fault identification of subway traction motor bearings, and the influence of abnormal data caused by accidental factors on the complexity of signal frequency component change is small, the influence of abnormal data caused by accidental factors can be reduced, thereby reducing the misjudgment of the bearing state, improving the accuracy of subway traction motor bearing fault detection, and facilitating practical application and promotion.

[0072] Please refer to Figure 2 The second aspect of the embodiment of the application provides a subway traction motor bearing fault identification device, which comprises:

[0073] The acquisition unit is configured to acquire a vibration acceleration signal of a subway traction motor.

[0074] The division unit is configured to divide the vibration acceleration signal into multiple segments of a specified time length to obtain multiple sub-signal segments.

[0075] The time-frequency transformation unit is configured to perform time-frequency transformation on the multiple sub-signal segments to obtain multiple time-frequency distribution matrices corresponding to the multiple sub-signal segments one by one.

[0076] The entropy calculation unit is configured to calculate an energy amplitude spectrum entropy corresponding to each frequency in each time-frequency distribution matrix based on an energy density value of a matrix element corresponding to the frequency, the energy amplitude spectrum entropy being used to represent a frequency domain complexity of energy distribution of the matrix element corresponding to the same frequency, and the greater the energy amplitude spectrum entropy, the greater the frequency domain complexity of energy distribution of the matrix element corresponding to the same frequency, and the smaller the energy amplitude spectrum entropy, the smaller the frequency domain complexity of energy distribution of the matrix element corresponding to the same frequency;

[0077] The determination unit is configured to determine an energy amplitude spectrum entropy sequence corresponding to the plurality of time-frequency distribution matrices based on the energy amplitude spectrum entropy corresponding to all frequencies in each time-frequency distribution matrix.

[0078] The fault detection unit is configured to perform fault detection by taking the energy amplitude spectrum entropy sequence as an input of a pre-trained fault recognition model to obtain a fault recognition result of the metro traction motor.

[0079] The fault recognition model is trained by taking an energy amplitude spectrum entropy sequence of a plurality of sample time-frequency distribution matrices corresponding to a plurality of sample sub-signal segments divided from a historical sample vibration acceleration signal of a sample metro traction motor as a sample input and taking a fault classification result of the sample metro traction motor when the historical sample vibration acceleration signal is detected as an output.

[0080] The working process, working details and technical effects of the metro traction motor bearing fault recognition device provided in the second aspect of the embodiment can be referred to the first aspect of the embodiment, and will not be described here.

[0081] As shown in Figure 3 The third aspect of the embodiment of the present application provides an electronic device, which comprises a memory, a processor and a transceiver connected in sequence and in communication, wherein the memory is configured to store a computer program, the transceiver is configured to receive and send messages, and the processor is configured to read the computer program and execute the metro traction motor bearing fault recognition method as described in the first aspect of the embodiment.

[0082] For example, the memory can include, but is not limited to, random access memory (RAM), read-only memory (ROM), flash memory, first-in-first-out memory (FIFO), first-in-last-out memory (FILO), and the like; the processor can be, but is not limited to, a microprocessor of STM32F105 series, an ARM (Advanced RISC Machines) processor, an X86 architecture processor, or an integrated NPU (neural-network processing units) processor; and the transceiver can be, but is not limited to, a WiFi (Wireless Fidelity) transceiver, a Bluetooth transceiver, a General Packet Radio Service (GPRS) transceiver, a ZigBee transceiver, a 3G transceiver, a 4G transceiver, a 5G transceiver, and the like.

[0083] The fourth aspect of the embodiment provides a computer readable storage medium storing instructions of the metro traction motor bearing fault identification method of the first aspect of the embodiment, that is, the computer readable storage medium stores instructions, and when the instructions are run on a computer, the metro traction motor bearing fault identification method of the first aspect is executed. The computer readable storage medium is a carrier for storing data, and can include, but is not limited to, a floppy disk, an optical disk, a hard disk, a flash memory, a USB flash disk, a memory stick, and the like. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices.

[0084] The fifth aspect of the embodiment provides a computer program product containing instructions, which, when executed on a computer, cause the computer to execute the metro traction motor bearing fault identification method of the first aspect of the embodiment. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices.

[0085] It should be understood that in the following description, specific details are provided to facilitate a complete understanding of the example embodiments. However, a person of ordinary skill in the art should understand that the example embodiments can be implemented without these specific details. For example, systems can be shown in block diagrams to avoid obscuring the examples with unnecessary details. In other instances, well-known processes, structures, and techniques can not be shown in unnecessary detail to avoid obscuring the example embodiments.

[0086] Finally, it should be noted that the above description is only the preferred embodiment of the present application, and is not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for identifying a fault of a bearing of a metro traction motor, characterized in that, The method comprises the following steps: obtaining a vibration acceleration signal of a metro traction motor; dividing the vibration acceleration signal into multiple segments of a specified time length to obtain multiple segmental sub-signal segments; performing time-frequency transformation on the multiple segmental sub-signal segments to obtain multiple time-frequency distribution matrices corresponding to the multiple segmental sub-signal segments one by one; based on the energy density values of the matrix elements corresponding to each frequency in each time-frequency distribution matrix, calculating the energy amplitude spectrum entropy corresponding to each frequency in each time-frequency distribution matrix according to the following formula (1); （1） wherein H represents the energy amplitude spectrum entropy corresponding to a certain frequency in the time-frequency distribution matrix, N represents the total number of matrix elements corresponding to a certain frequency, X(i) represents the normalized value of the energy density value corresponding to the i-th matrix element corresponding to a certain frequency, i is a positive integer less than or equal to N, and the energy amplitude spectrum entropy is used to represent the energy distribution frequency domain complexity of the matrix elements corresponding to the same frequency. The greater the energy amplitude spectrum entropy, the greater the energy distribution frequency domain complexity of the matrix elements corresponding to the same frequency. The smaller the energy amplitude spectrum entropy, the smaller the energy distribution frequency domain complexity of the matrix elements corresponding to the same frequency; selecting the minimum value of the energy amplitude spectrum entropy corresponding to all frequencies in each time-frequency distribution matrix to obtain an energy amplitude spectrum entropy sequence corresponding to the multiple time-frequency distribution matrices; using the energy amplitude spectrum entropy sequence as the input of a pre-trained fault recognition model for fault detection to obtain the fault recognition result of the metro traction motor; wherein the fault recognition model is trained by taking the energy amplitude spectrum entropy sequence of the multiple sample time-frequency distribution matrices corresponding to the multiple segmental sample sub-signal segments divided from the historical sample vibration acceleration signal of the sample metro traction motor as the sample input, and taking the fault classification result of the sample metro traction motor when the historical sample vibration acceleration signal is detected as the output.

2. The metro traction motor bearing fault identification method of claim 1, wherein, The time-frequency transformation on the multiple segmental sub-signal segments to obtain multiple time-frequency distribution matrices corresponding to the multiple segmental sub-signal segments one by one comprises: performing time-frequency transformation on the multiple segmental sub-signal segments by short-time Fourier transform to obtain multiple time-frequency distribution matrices corresponding to the multiple segmental sub-signal segments one by one.

3. The metro traction motor bearing fault identification method of claim 1, wherein, The fault detection using the energy amplitude spectrum entropy sequence as the input of the pre-trained fault recognition model comprises: normalizing the energy amplitude spectrum entropy sequence and then using the normalized energy amplitude spectrum entropy sequence as the input of the pre-trained fault recognition model for fault detection.

4. The metro traction motor bearing fault identification method of claim 1, wherein, After obtaining the vibration acceleration signal of the metro traction motor, the method further comprises: performing filtering processing on the vibration acceleration signal.

5. The method for metro traction motor bearing fault identification according to claim 1, characterized in that, The fault recognition model is a convolutional neural network model or a recurrent neural network model.

6. A metro traction motor bearing fault identification device, characterized in that, The method comprises the following steps: an obtaining unit configured to obtain a vibration acceleration signal of a metro traction motor; a dividing unit configured to divide the vibration acceleration signal into multiple segments of a specified time length to obtain multiple segmental sub-signal segments; a time-frequency transformation unit configured to perform time-frequency transformation on the multiple segmental sub-signal segments to obtain multiple time-frequency distribution matrices corresponding to the multiple segmental sub-signal segments one by one; An entropy calculation unit is configured to calculate an energy amplitude spectrum entropy corresponding to each frequency in each time-frequency distribution matrix according to the following formula (1) based on the energy density value of the matrix element corresponding to each frequency in the time-frequency distribution matrix: （1） wherein H represents the energy amplitude spectrum entropy corresponding to a certain frequency in the time-frequency distribution matrix, N represents the total number of matrix elements corresponding to a certain frequency, X(i) represents the normalized value of the energy density value corresponding to the i-th matrix element corresponding to a certain frequency, i is a positive integer less than or equal to N, the energy amplitude spectrum entropy is used to represent the energy distribution frequency domain complexity of the matrix elements corresponding to the same frequency, and the greater the energy amplitude spectrum entropy, the greater the energy distribution frequency domain complexity of the matrix elements corresponding to the same frequency, and the smaller the energy amplitude spectrum entropy, the smaller the energy distribution frequency domain complexity of the matrix elements corresponding to the same frequency; A determination unit is configured to select the minimum value of the energy amplitude spectrum entropy corresponding to all frequencies in each time-frequency distribution matrix to obtain an energy amplitude spectrum entropy sequence corresponding to the plurality of time-frequency distribution matrices; A fault detection unit is configured to perform fault detection by taking the energy amplitude spectrum entropy sequence as the input of a pre-trained fault recognition model to obtain a fault recognition result of the subway traction motor. The fault recognition model is trained by taking the energy amplitude spectrum entropy sequence of a plurality of sample time-frequency distribution matrices corresponding to a plurality of sample sub-signal segments divided from a historical sample vibration acceleration signal of a sample subway traction motor as a sample input, and taking the fault classification result of the sample subway traction motor when the historical sample vibration acceleration signal is detected as an output.

7. An electronic device, comprising: The computer program or the instructions realize the subway traction motor bearing fault recognition method of any one of claims 1-5 when executed by a computer.

8. A computer program product comprising computer programs or instructions, characterized in that, The computer program or the instructions realize the subway traction motor bearing fault recognition method of any one of claims 1-5 when executed by a computer.

Images