CN114400940A - Demagnetization fault diagnosis method for permanent magnet driving motor for electric automobile and electric automobile - Google Patents

Demagnetization fault diagnosis method for permanent magnet driving motor for electric automobile and electric automobile Download PDF

Info

Publication number
CN114400940A
CN114400940A CN202111547348.9A CN202111547348A CN114400940A CN 114400940 A CN114400940 A CN 114400940A CN 202111547348 A CN202111547348 A CN 202111547348A CN 114400940 A CN114400940 A CN 114400940A
Authority
CN
China
Prior art keywords
network
permanent magnet
dimensional wavelet
fault diagnosis
driving motor
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202111547348.9A
Other languages
Chinese (zh)
Other versions
CN114400940B (en
Inventor
张晓飞
黄凤琴
谢金平
黄守道
周俊鸿
龙卓
唐瑶
唐镜博
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hunan University
Original Assignee
Hunan University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Hunan University filed Critical Hunan University
Priority to CN202111547348.9A priority Critical patent/CN114400940B/en
Publication of CN114400940A publication Critical patent/CN114400940A/en
Application granted granted Critical
Publication of CN114400940B publication Critical patent/CN114400940B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02P—CONTROL OR REGULATION OF ELECTRIC MOTORS, ELECTRIC GENERATORS OR DYNAMO-ELECTRIC CONVERTERS; CONTROLLING TRANSFORMERS, REACTORS OR CHOKE COILS
    • H02P21/00—Arrangements or methods for the control of electric machines by vector control, e.g. by control of field orientation
    • H02P21/0003—Control strategies in general, e.g. linear type, e.g. P, PI, PID, using robust control
    • H02P21/0014—Control strategies in general, e.g. linear type, e.g. P, PI, PID, using robust control using neural networks
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/005—Testing of electric installations on transport means
    • G01R31/006—Testing of electric installations on transport means on road vehicles, e.g. automobiles or trucks
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G01R31/34—Testing dynamo-electric machines
    • G01R31/343—Testing dynamo-electric machines in operation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/047—Probabilistic or stochastic networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02P—CONTROL OR REGULATION OF ELECTRIC MOTORS, ELECTRIC GENERATORS OR DYNAMO-ELECTRIC CONVERTERS; CONTROLLING TRANSFORMERS, REACTORS OR CHOKE COILS
    • H02P21/00—Arrangements or methods for the control of electric machines by vector control, e.g. by control of field orientation
    • H02P21/14—Estimation or adaptation of machine parameters, e.g. flux, current or voltage
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02P—CONTROL OR REGULATION OF ELECTRIC MOTORS, ELECTRIC GENERATORS OR DYNAMO-ELECTRIC CONVERTERS; CONTROLLING TRANSFORMERS, REACTORS OR CHOKE COILS
    • H02P25/00—Arrangements or methods for the control of AC motors characterised by the kind of AC motor or by structural details
    • H02P25/02—Arrangements or methods for the control of AC motors characterised by the kind of AC motor or by structural details characterised by the kind of motor
    • H02P25/022—Synchronous motors
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T10/00—Road transport of goods or passengers
    • Y02T10/60—Other road transportation technologies with climate change mitigation effect
    • Y02T10/64—Electric machine technologies in electromobility

Abstract

The invention discloses a demagnetization fault diagnosis method for a permanent magnet driving motor for an electric automobile and the electric automobile, wherein the demagnetization fault diagnosis method comprises the steps of collecting a magnetic flux leakage signal of the permanent magnet driving motor for the electric automobile; converting the magnetic leakage signal into a two-dimensional wavelet time-frequency diagram through wavelet transformation; extracting self-coding features of the two-dimensional wavelet time-frequency graph to obtain a feature vector 1, and extracting maximum stable extremum region features of the two-dimensional wavelet time-frequency graph to obtain a feature vector 2; and inputting the characteristic vector 1 and the characteristic vector 2 into a machine learning model trained in advance to obtain a demagnetization fault diagnosis result of the permanent magnet driving motor for the electric automobile. The invention can extract more effective fault high-dimensional characteristics, convert demagnetization fault diagnosis of the permanent magnet driving motor for the electric automobile into more intuitive image processing, and can improve the comprehensiveness and accuracy of the fault high-dimensional characteristics and improve the accuracy of the demagnetization fault diagnosis of the permanent magnet driving motor for the electric automobile by fusing the two types of characteristics.

Description

Demagnetization fault diagnosis method for permanent magnet driving motor for electric automobile and electric automobile
Technical Field
The invention relates to an electric vehicle fault diagnosis technology, in particular to a demagnetization fault diagnosis method for a permanent magnet drive motor for an electric vehicle and the electric vehicle.
Background
Since the 2001 starts the major science and technology specialization of electric automobiles, the new energy automobile industry in China has gone through the development process of 20 years, and China has become the production country and the consumption country of the largest new energy automobile in the world, wherein the scale of the pure electric automobile accounts for more than 50% of the world, and the first global level is internationally advanced. Along with the continuous improvement of new energy automobile intellectuality, integration, its inner structure is also complicated day by day. In recent two years, safety accidents of new energy automobiles present an increasing trend, the industry has gradually changed from mileage anxiety to safety anxiety, and safety problems have become one of the core problems to be solved in the development of new energy automobiles.
The heart of the power of an electric vehicle is a motor. The permanent magnet driving motor adopts the permanent magnet to generate a motor magnetic field, has simple structure and high efficiency and control precision and is widely applied to the fields of electric automobiles, aerospace, wind power generation and the like. The power of domestic and international electric vehicles is mostly provided by permanent magnet synchronous motors. The magnetic steel sheet of the permanent magnet driving motor is mostly made of neodymium iron boron permanent magnet materials, and the Curie temperature of the magnetic steel sheet is low. Therefore, overload of the motor, damage of a heat dissipation system and the like can cause magnetic loss of the permanent magnet, and demagnetization of the permanent magnet is easily caused. The demagnetization fault can aggravate torque ripple and motor loss, seriously reduce the performance of the automobile and cause property loss and casualties in serious cases. Therefore, the demagnetization fault diagnosis of the permanent magnet driving motor for the electric automobile has important practical significance.
The existing research on demagnetization fault diagnosis of the permanent magnet driving motor for the electric automobile is few, and particularly, the existing diagnosis method aiming at practical application is less. Related research is carried out, and only the demagnetization fault diagnosis of the permanent magnet motor without considering the application field can be referred. The existing demagnetization fault diagnosis method of the permanent magnet motor has achieved a series of achievements, and both the data-driven diagnosis method and the model-based diagnosis method have achieved some achievements. For example, data-based diagnostic methods mostly focus on fault diagnosis on intelligent algorithms or neural network algorithms. However, most of the diagnosis methods based on data driving require a large amount of data to train the classifier, and the model-based methods require accurate fault models to be established. In addition, the existing method mainly focuses on fault diagnosis under one-dimensional signals, and does not consider abundant fault high-dimensional characteristics. In data-based demagnetization fault diagnosis, the depth of a diagnosis network under a convolutional neural network algorithm is generally deep, and the calculation cost is high. However, in an actual electric vehicle application environment, a fault signal is easily affected by noise in a complex environment, resulting in difficulty in extracting fault features. Meanwhile, because the permanent magnet driving motor for the electric automobile is in a normal state for a long time, actual fault signals in different states are difficult to obtain. In addition, both complex working conditions and complex environments are tests faced by electric vehicles in practical application, fault signals obtained under the two conditions are also complex, and the diagnosis effect of the traditional single and single-layer diagnosis algorithm is poor. Therefore, how to realize demagnetization fault diagnosis of the permanent magnet driving motor for the electric automobile becomes a key technical problem to be solved urgently.
Disclosure of Invention
The technical problems to be solved by the invention are as follows: the invention can extract more effective fault high-dimensional characteristics, convert the demagnetization fault diagnosis of the permanent magnet driving motor for the electric automobile into more intuitive image processing, improve the comprehensiveness and the accuracy of the fault high-dimensional characteristics through the fusion of the two types of characteristics, and improve the accuracy of the demagnetization fault diagnosis of the permanent magnet driving motor for the electric automobile.
In order to solve the technical problems, the invention adopts the technical scheme that:
a demagnetization fault diagnosis method for a permanent magnet driving motor for an electric automobile comprises the following steps:
1) collecting a magnetic leakage signal of a permanent magnet driving motor for the electric automobile;
2) converting the magnetic leakage signal into a two-dimensional wavelet time-frequency diagram through wavelet transformation;
3) extracting self-coding features of the two-dimensional wavelet time-frequency graph to obtain a feature vector 1, and extracting maximum stable extremum region features of the two-dimensional wavelet time-frequency graph to obtain a feature vector 2;
4) and inputting the characteristic vector 1 and the characteristic vector 2 into a machine learning model trained in advance to obtain a demagnetization fault diagnosis result of the permanent magnet driving motor for the electric automobile.
Optionally, when the leakage magnetic signal is converted into a two-dimensional wavelet time-frequency graph through wavelet transform in step 2), a functional expression of the wavelet transform is as follows:
Figure BDA0003416108640000021
in the above formula, WT (α, τ) is the result of wavelet transform, α is the scale, τ is the amount of translation, f (t) is the one-dimensional leakage magnetic signal,
Figure BDA0003416108640000022
is the function of the parent wave, t is time.
Optionally, the step 3) of extracting the self-coding feature of the two-dimensional wavelet time-frequency graph to obtain the feature vector 1 includes: converting the two-dimensional wavelet time-frequency graph into a gray-scale graph, inputting the gray-scale graph into a pre-trained double-layer self-encoder network, and extracting global features of the gray-scale graph obtained after the two-dimensional wavelet time-frequency graph is converted through the double-layer self-encoder network to serve as feature vectors 1; the double-layer self-encoder comprises two stacked self-encoders, the self-encoders comprise two network layers including an encoding layer and a decoding layer, the first self-encoder is used for extracting features to obtain shallow features, and the second self-encoder is used for extracting features of the shallow features to obtain depth features and using the depth features as finally obtained global features.
Optionally, the extracting the maximum stable extremum region feature of the two-dimensional wavelet time-frequency map in step 3) to obtain the feature vector 2 includes: sequentially carrying out binarization processing on the two-dimensional wavelet time-frequency graph by using a group of binarization threshold values and calculating a variable V (i) according to the following formula:
Figure BDA0003416108640000023
in the above formula, QiExpressing the area of the ith connected region in the binary region obtained by the current binary threshold, wherein delta represents the preset variable quantity of the binary threshold, and Qi+ΔRepresenting the area Q of the ith connected region in the binary region obtained after the current binary threshold value is increased by the preset binary threshold value variation deltai-ΔRepresenting the area of the ith connected region in the binary region obtained after the current binary threshold value is reduced by a preset binary threshold value variation delta; and if the variable V (i) is smaller than the set value, taking the binary region obtained by the current binarization threshold value as the obtained maximum stable extremum region characteristic and converting the binary region into a vector as a characteristic vector 2.
Optionally, the machine learning model in the step 4) includes a softmax classifier 1, a softmax classifier 2 and a comprehensive classifier, the softmax classifier 1 is configured to obtain a classification probability value 1 according to the feature vector 1, the softmax classifier 2 is configured to obtain a classification probability value 2 according to the feature vector 2, and the comprehensive classifier is configured to obtain a final demagnetization fault diagnosis result of the driving motor for the electric permanent magnet vehicle according to the classification probability value 1 and the classification probability value 2.
Optionally, the comprehensive classifier is a support vector machine classifier.
Optionally, the step 4) is preceded by a step of training the machine learning model by using the data samples, and the step of performing data sample expansion by using the generated countermeasure network when the machine learning model is trained by using the data samples includes:
s1) inputting a group of random numbers and corresponding sample labels into a generator for generating the countermeasure network, generating a virtual two-dimensional wavelet time-frequency graph with labels, and inputting the real two-dimensional wavelet time-frequency graph with labels and the virtual two-dimensional wavelet time-frequency graph into a discriminator for generating the countermeasure network;
s2) respectively calculating the network loss and the network accuracy of a generator in the generation countermeasure network and the network loss and the network accuracy of a discriminator, if the network loss and the network accuracy of one or both of the generator and the discriminator meet the requirements, judging that the training of the countermeasure network is finished, and jumping to the step S3); otherwise, adjusting the parameters of the generator and the discriminator for generating the confrontation network, and jumping to the step S1) to continue training to generate the confrontation network;
s3) generating a labeled virtual two-dimensional wavelet time-frequency graph as the two-dimensional wavelet time-frequency graph in the data sample by using the trained generator for generating the countermeasure network.
Optionally, the function expression of the network loss and the network accuracy of the generator in step S2) is:
Figure BDA0003416108640000031
Figure BDA0003416108640000032
in the above formula, LgTo loss of the network of the generator, SgTo the network accuracy of the generator, mean is the averaging calculation,
Figure BDA0003416108640000033
the probability that the virtual two-dimensional wavelet time-frequency graph is identified as a real two-dimensional wavelet time-frequency graph in the discriminator is obtained; the functional expression of the network loss and the network accuracy of the discriminator in step S2) is:
Figure BDA0003416108640000034
Figure BDA0003416108640000035
in the above formula, LDFor discriminator network loss, SDFor the purpose of the network accuracy of the arbiter,
Figure BDA0003416108640000036
is the probability of the real two-dimensional wavelet time-frequency diagram at the output of the discriminator network,
Figure BDA0003416108640000037
is the probability that the real two-dimensional wavelet time-frequency graph is identified as the virtual two-dimensional wavelet time-frequency graph by the discriminator.
In addition, the invention also provides an electric automobile which comprises an electric automobile body which is provided with a control unit and adopts the permanent magnet driving motor, wherein the control unit comprises a microprocessor and a memory which are connected with each other, the microprocessor is connected with a magnetic leakage signal sensor for collecting a magnetic leakage signal, and the microprocessor is programmed or configured to execute the steps of the demagnetization fault diagnosis method of the permanent magnet driving motor for the electric automobile.
In addition, the present invention also provides a computer readable storage medium having stored therein a computer program executed by a computer apparatus to implement the demagnetization fault diagnosis method of a permanent magnet drive motor for an electric vehicle
Compared with the prior art, the invention has the following advantages:
1. the method comprises the steps of collecting magnetic leakage signals of the permanent magnet driving motor for the electric automobile, converting the magnetic leakage signals into a two-dimensional wavelet time-frequency graph through wavelet transformation, showing more effective high-dimensional fault characteristics, and converting demagnetization fault diagnosis of the permanent magnet driving motor for the electric automobile into more intuitive image processing.
2. According to the invention, the self-coding feature extraction is carried out on the two-dimensional wavelet time-frequency graph to obtain the feature vector 1, the maximum stable extremum region feature extraction is carried out on the two-dimensional wavelet time-frequency graph to obtain the feature vector 2, the comprehensiveness and the accuracy of the high-dimensional fault feature can be improved through the fusion of the two types of features, and the accuracy of the demagnetization fault diagnosis of the permanent magnet driving motor for the electric automobile can be improved.
Drawings
FIG. 1 is a schematic diagram of a basic flow of a method according to an embodiment of the present invention.
Fig. 2 shows a two-dimensional wavelet time-frequency diagram of a normal motor and a demagnetization fault of different types.
FIG. 3 is a schematic diagram of extracting self-coding features of a two-dimensional wavelet time-frequency graph to obtain a feature vector 1 according to the embodiment of the present invention.
FIG. 4 is a schematic diagram of extracting self-coding features of a two-dimensional wavelet time-frequency graph to obtain a feature vector 2 according to the embodiment of the present invention.
Fig. 5 is a schematic structural diagram of the softmax classifier in the embodiment of the present invention.
Fig. 6 is a schematic structural diagram of functional modules of an electric vehicle body according to an embodiment of the invention.
Detailed Description
The first embodiment is as follows:
as shown in fig. 1, the method for diagnosing demagnetization fault of permanent magnet drive motor for electric vehicle in this embodiment includes:
1) collecting a magnetic leakage signal of a permanent magnet driving motor for the electric automobile;
2) converting the magnetic leakage signal into a two-dimensional wavelet time-frequency diagram through wavelet transformation;
3) extracting self-coding features of the two-dimensional wavelet time-frequency graph to obtain a feature vector 1, and extracting maximum stable extremum region features of the two-dimensional wavelet time-frequency graph to obtain a feature vector 2;
4) and inputting the characteristic vector 1 and the characteristic vector 2 into a machine learning model trained in advance to obtain a demagnetization fault diagnosis result of the permanent magnet driving motor for the electric automobile.
Fig. 2 shows a two-dimensional wavelet time-frequency diagram generated by measuring a magnetic flux leakage signal on the surface of a motor by using a non-contact ac magnetic sensor under 1000r/min for a normal motor and different types of demagnetization faults (demagnetization fault 1 and demagnetization fault 2, demagnetization fault 1 is 30%, and demagnetization fault 2 is 100%), which is known from fig. 2.
In this embodiment, when the magnetic leakage signal is converted into a two-dimensional wavelet time-frequency diagram through wavelet transform in step 2), a functional expression of the wavelet transform is as follows:
Figure BDA0003416108640000051
in the above formula, WT (α, τ) is the result of wavelet transform, α is the scale, τ is the amount of translation, f (t) is the one-dimensional leakage magnetic signal,
Figure BDA0003416108640000052
is the function of the parent wave, t is time.
In this embodiment, the extracting the self-coding feature of the two-dimensional wavelet time-frequency graph in step 3) to obtain the feature vector 1 includes: converting the two-dimensional wavelet time-frequency graph into a gray-scale graph, inputting the gray-scale graph into a pre-trained double-layer self-encoder network, and extracting global features of the gray-scale graph obtained after the two-dimensional wavelet time-frequency graph is converted through the double-layer self-encoder network to serve as feature vectors 1; the double-layer self-encoder comprises two stacked self-encoders, the self-encoders comprise two network layers including an encoding layer and a decoding layer, the first self-encoder is used for extracting features to obtain shallow features, and the second self-encoder is used for extracting features of the shallow features to obtain depth features and using the depth features as finally obtained global features. Fig. 3 is a schematic diagram of a two-dimensional wavelet time-frequency graph being subjected to self-coding feature extraction to obtain a feature vector 1. The double-layer self-encoder network is an existing neural network, and the required settings comprise maximum convolution times, L2 network weight regularization parameters, sparse regularization controller parameters, sparse regularization item parameters and telescopic data setting.
When the maximum stable extremum region feature in the image is extracted, the image sample is subjected to binarization processing by using a series of gray level threshold values, then a corresponding binary region is obtained at each threshold value, and finally, a region which can keep a stable shape in a wider gray level threshold value range is the maximum stable extremum region feature. In this embodiment, the extracting the maximum stable extremum region feature of the two-dimensional wavelet time-frequency map in step 3) to obtain the feature vector 2 includes: sequentially carrying out binarization processing on the two-dimensional wavelet time-frequency graph by using a group of binarization threshold values and calculating a variable V (i) according to the following formula:
Figure BDA0003416108640000053
in the above formula, QiExpressing the area of the ith connected region in the binary region obtained by the current binary threshold, wherein delta represents the preset variable quantity of the binary threshold, and Qi+ΔRepresenting the area Q of the ith connected region in the binary region obtained after the current binary threshold value is increased by the preset binary threshold value variation deltai-ΔRepresenting the area of the ith connected region in the binary region obtained after the current binary threshold value is reduced by a preset binary threshold value variation delta; and if the variable V (i) is smaller than the set value, taking the binary region obtained by the current binarization threshold value as the obtained maximum stable extremum region characteristic and converting the binary region into a vector as a characteristic vector 2. Fig. 4 is a schematic diagram of extracting self-coding features of a two-dimensional wavelet time-frequency graph to obtain a feature vector 2.
Referring to fig. 1, the machine learning model in step 4) of this embodiment includes a softmax classifier 1, a softmax classifier 2, and a comprehensive classifier, where the softmax classifier 1 is configured to obtain a classification probability value 1 according to a feature vector 1, the softmax classifier 2 is configured to obtain a classification probability value 2 according to a feature vector 2, and the comprehensive classifier is configured to obtain a final demagnetization fault diagnosis result of the permanent magnet drive motor for the electric vehicle according to the classification probability value 1 and the classification probability value 2. Because the traditional softmax classifier may not have good effect in some diagnostic data, for this reason, the softmax classifier is improved in this embodiment, namely, the softmax classifier is improved by adding n layers (n is 1-m, m is a positive integer) of comprehensive classifiers after the traditional softmax classifier (softmax classifier 1, softmax classifier 2), and classification is performed by reusing the classification probability value output by softmax, so that the diagnostic effect under complex conditions and complex environments is improved.
The softmax classifier 1 and the softmax classifier 2 are both softmax neural networks, and the structures thereof are shown in fig. 5. The softmax classifier 1 and the softmax classifier 2 calculate classification probability values based on a traditional softmax neural network, the sum of the classification probability values is 1, and the calculation function expression is as follows:
Figure BDA0003416108640000061
in the above formula, softmax denotes a softmax classifier, yiRepresenting the sofmax neural network classifier input and n representing the dimension of the softmax classifier. For multi-classification, the closeness of the actual output to the desired output needs to be determined by a cross-entropy loss function whose functional expression is:
Figure BDA0003416108640000062
in the above formula, LossiRepresenting the cross entropy loss function, tiRepresenting true values, softmax representing softmax classifier, yiRepresenting the sofmax neural network classifier input.
It should be noted that the comprehensive classifier can adopt a required classifier, such as a softmax classifier with one or more layers or other types of classifiers, as required. For example, as an alternative embodiment, the comprehensive classifier is a support vector machine classifier (SVM).
In this embodiment, the step 4) further includes a step of training a machine learning model by using the data samples, and since the softmax neural network and the training of the support vector machine classifier (SVM) are well-known methods, details thereof are not described herein.
In addition, this embodiment still provides an electric automobile, including the electric automobile body that has the control unit and adopt permanent magnet drive motor, the control unit includes interconnect's microprocessor and memory, microprocessor is connected with the magnetic leakage signal sensor that is used for gathering the magnetic leakage signal, and this microprocessor is programmed or is configured in order to carry out the step of aforementioned permanent magnet drive motor demagnetization fault diagnosis method for electric automobile. Fig. 6 is a schematic structural diagram of functional modules of an electric vehicle body according to an embodiment of the invention. Referring to fig. 6, in the whole system of the electric vehicle, the whole vehicle monitoring system issues a control instruction to the vehicle driving control system through can communication, and the vehicle driving control system controls the driving motor so that the electric vehicle operates. For fault diagnosis, the integral monitoring system integrates fault diagnosis result display, and all algorithms and calculation processes of fault diagnosis are in an automobile driving control system. And measuring a magnetic flux leakage signal on the surface of the driving motor by adopting a magnetic flux sensor, transmitting the signal to an automobile driving control system, realizing the fault diagnosis of the motor by the automobile driving control system based on the method of the embodiment, and displaying and outputting the diagnosis result in the whole automobile monitoring system.
Furthermore, the present embodiment also provides a computer-readable storage medium in which a computer program executed by a computer apparatus to implement the foregoing demagnetization failure diagnosis method for a permanent magnet drive motor for an electric vehicle is stored.
Example two:
this embodiment is a further improvement of the first embodiment. In order to implement demagnetization fault diagnosis of a permanent magnet driving motor for an electric vehicle under a small sample, the method for training a machine learning model by using a data sample in the embodiment includes the steps of generating a countermeasure network to expand the data sample:
s1) inputting a group of random numbers and corresponding sample labels into a generator for generating the countermeasure network, generating a virtual two-dimensional wavelet time-frequency graph with labels, and inputting the real two-dimensional wavelet time-frequency graph with labels and the virtual two-dimensional wavelet time-frequency graph into a discriminator for generating the countermeasure network;
s2) calculating the network loss and the network accuracy of the discriminator, if the network loss and the network accuracy of the discriminator meet the requirements (the network loss is less than a set value, and the network accuracy is more than the set value), judging that the training of the generation countermeasure network is finished, and skipping to the step S3); otherwise, adjusting the parameters of the generator and the discriminator for generating the confrontation network, and jumping to the step S1) to continue training to generate the confrontation network;
s3) generating a labeled virtual two-dimensional wavelet time-frequency graph as the two-dimensional wavelet time-frequency graph in the data sample by using the trained generator for generating the countermeasure network.
The conditional generation countermeasure network is a deep neural network that is capable of generating data having the same characteristics as the real input data. A conditional generation countermeasure network is composed of two networks: a generator and a discriminator. The goal of the generator is to generate the data that the arbiter recognizes as true. To maximize the probability that the image generated by the generator is recognized by the discriminator as a true image, the negative log-likelihood function is minimized. The countermeasure network is generated by using the condition to generate a virtual two-dimensional wavelet time-frequency diagram, and the problem of few fault signals in the practical application of the electric automobile is solved.
The requirement of the discriminator is determined by a loss function and a network accuracy function, and the functional expression of the network loss and the network accuracy of the discriminator in step S2) of the embodiment is as follows:
Figure BDA0003416108640000071
Figure BDA0003416108640000072
in the above formula, LDFor discriminator network loss, SDFor the purpose of the network accuracy of the arbiter,
Figure BDA0003416108640000073
is the probability of the real two-dimensional wavelet time-frequency diagram at the output of the discriminator network,
Figure BDA0003416108640000074
is the probability that the real two-dimensional wavelet time-frequency graph is identified as the virtual two-dimensional wavelet time-frequency graph by the discriminator.
In addition, this embodiment still provides an electric automobile, including the electric automobile body that has the control unit and adopt permanent magnet drive motor, the control unit includes interconnect's microprocessor and memory, microprocessor is connected with the magnetic leakage signal sensor that is used for gathering the magnetic leakage signal, and this microprocessor is programmed or is configured in order to carry out the step of aforementioned permanent magnet drive motor demagnetization fault diagnosis method for electric automobile.
Furthermore, the present embodiment also provides a computer-readable storage medium in which a computer program executed by a computer apparatus to implement the foregoing demagnetization failure diagnosis method for a permanent magnet drive motor for an electric vehicle is stored.
Example three:
the embodiment is basically the same as the second embodiment, and the main difference is that the iterative determination conditions in step S2) are different, in this embodiment, S2) the network loss and the network accuracy of the generator are calculated, and if the network loss and the network accuracy of the generator meet the requirements (the network loss is less than the set value, and the network accuracy is greater than the set value), it is determined that the generation of the antagonistic network is completed, and step S3 is skipped; otherwise, adjusting the parameters of the generator and the discriminator for generating the confrontation network, and jumping to the step S1) to continue training to generate the confrontation network.
The requirement of the generator is determined by the network loss and the network accuracy of the generator, and the functional expression of the network loss and the network accuracy of the generator in step S2) of the present embodiment is:
Figure BDA0003416108640000081
Figure BDA0003416108640000082
in the above formula, LgTo loss of the network of the generator, SgTo the network accuracy of the generator, mean is the averaging calculation,
Figure BDA0003416108640000083
is the probability that the virtual two-dimensional wavelet time-frequency graph is identified as a real two-dimensional wavelet time-frequency graph in the discriminator.
In addition, this embodiment still provides an electric automobile, including the electric automobile body that has the control unit and adopt permanent magnet drive motor, the control unit includes interconnect's microprocessor and memory, microprocessor is connected with the magnetic leakage signal sensor that is used for gathering the magnetic leakage signal, and this microprocessor is programmed or is configured in order to carry out the step of aforementioned permanent magnet drive motor demagnetization fault diagnosis method for electric automobile.
Furthermore, the present embodiment also provides a computer-readable storage medium in which a computer program executed by a computer apparatus to implement the foregoing demagnetization failure diagnosis method for a permanent magnet drive motor for an electric vehicle is stored.
Example four:
this embodiment is substantially the same as the second embodiment, and the main difference is that the determination conditions of the iteration in step S2) are different, and in this embodiment, S2) is the or logic operation of the first embodiment and the second embodiment, that is: respectively calculating the network loss and the network accuracy of a generator in the generated countermeasure network and the network loss and the network accuracy of a discriminator, if the network loss and the network accuracy of one of the generator and the discriminator meet the requirements, judging that the training of the countermeasure network is finished, and jumping to the step S3); otherwise, adjusting the parameters of the generator and the discriminator for generating the countermeasure network, and jumping to step S1) to continue training to generate the countermeasure network, which can also realize the training of the countermeasure network.
In addition, this embodiment still provides an electric automobile, including the electric automobile body that has the control unit and adopt permanent magnet drive motor, the control unit includes interconnect's microprocessor and memory, microprocessor is connected with the magnetic leakage signal sensor that is used for gathering the magnetic leakage signal, and this microprocessor is programmed or is configured in order to carry out the step of aforementioned permanent magnet drive motor demagnetization fault diagnosis method for electric automobile.
Furthermore, the present embodiment also provides a computer-readable storage medium in which a computer program executed by a computer apparatus to implement the foregoing demagnetization failure diagnosis method for a permanent magnet drive motor for an electric vehicle is stored.
Example five:
this embodiment is substantially the same as the second embodiment, and the main difference is that the determination conditions of the iteration in step S2) are different, and S2) in this embodiment is the and logic operation of the first and second embodiments, that is: respectively calculating the network loss and the network accuracy of a generator in the generated countermeasure network and the network loss and the network accuracy of a discriminator, if the network loss and the network accuracy of all the generators and the discriminators meet the requirements, judging that the training of the countermeasure network is finished, and jumping to the step S3); otherwise, adjusting the parameters of the generator and the discriminator for generating the countermeasure network, and jumping to step S1) to continue training to generate the countermeasure network, which can also realize the training of the countermeasure network.
In addition, this embodiment still provides an electric automobile, including the electric automobile body that has the control unit and adopt permanent magnet drive motor, the control unit includes interconnect's microprocessor and memory, microprocessor is connected with the magnetic leakage signal sensor that is used for gathering the magnetic leakage signal, and this microprocessor is programmed or is configured in order to carry out the step of aforementioned permanent magnet drive motor demagnetization fault diagnosis method for electric automobile.
Furthermore, the present embodiment also provides a computer-readable storage medium in which a computer program executed by a computer apparatus to implement the foregoing demagnetization failure diagnosis method for a permanent magnet drive motor for an electric vehicle is stored.
Example six:
the embodiment is basically the same as the second embodiment, and the main difference is that the determination conditions of the iteration in the step S2) are different, in this embodiment S2), based on the number of iterations, if the number of iterations is equal to the set value, it is determined that the training of the generation countermeasure network is completed, and the step S3) is skipped; otherwise, adjusting the parameters of the generator and the discriminator for generating the countermeasure network, and jumping to step S1) to continue training to generate the countermeasure network, which can also realize the training of the countermeasure network.
In addition, this embodiment still provides an electric automobile, including the electric automobile body that has the control unit and adopt permanent magnet drive motor, the control unit includes interconnect's microprocessor and memory, microprocessor is connected with the magnetic leakage signal sensor that is used for gathering the magnetic leakage signal, and this microprocessor is programmed or is configured in order to carry out the step of aforementioned permanent magnet drive motor demagnetization fault diagnosis method for electric automobile.
Furthermore, the present embodiment also provides a computer-readable storage medium in which a computer program executed by a computer apparatus to implement the foregoing demagnetization failure diagnosis method for a permanent magnet drive motor for an electric vehicle is stored.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-readable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein. The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks. These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks. These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The above description is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments, and all technical solutions belonging to the idea of the present invention belong to the protection scope of the present invention. It should be noted that modifications and embellishments within the scope of the invention may occur to those skilled in the art without departing from the principle of the invention, and are considered to be within the scope of the invention.

Claims (10)

1. The demagnetization fault diagnosis method of the permanent magnet driving motor for the electric automobile is characterized by comprising the following steps of:
1) collecting a magnetic leakage signal of a permanent magnet driving motor for the electric automobile;
2) converting the magnetic leakage signal into a two-dimensional wavelet time-frequency diagram through wavelet transformation;
3) extracting self-coding features of the two-dimensional wavelet time-frequency graph to obtain a feature vector 1, and extracting maximum stable extremum region features of the two-dimensional wavelet time-frequency graph to obtain a feature vector 2;
4) and inputting the characteristic vector 1 and the characteristic vector 2 into a machine learning model trained in advance to obtain a demagnetization fault diagnosis result of the permanent magnet driving motor for the electric automobile.
2. The demagnetization fault diagnosis method of the permanent magnet driving motor for the electric vehicle according to claim 1, wherein when the magnetic leakage signal is transformed into a two-dimensional wavelet time-frequency diagram through wavelet transformation in step 2), the functional expression of the wavelet transformation is as follows:
Figure FDA0003416108630000011
in the above formula, WT (α, τ) is the result of wavelet transform, α is the scale, τ is the amount of translation, f (t) is the one-dimensional leakage magnetic signal,
Figure FDA0003416108630000012
is the function of the parent wave, t is time.
3. The demagnetization fault diagnosis method of the permanent magnet drive motor for the electric vehicle according to claim 1, wherein the step 3) of extracting the self-coding features of the two-dimensional wavelet time-frequency graph to obtain the feature vector 1 comprises the following steps: converting the two-dimensional wavelet time-frequency graph into a gray-scale graph, inputting the gray-scale graph into a pre-trained double-layer self-encoder network, and extracting global features of the gray-scale graph obtained after the two-dimensional wavelet time-frequency graph is converted through the double-layer self-encoder network to serve as feature vectors 1; the double-layer self-encoder comprises two stacked self-encoders, the self-encoders comprise two network layers including an encoding layer and a decoding layer, the first self-encoder is used for extracting features to obtain shallow features, and the second self-encoder is used for extracting features of the shallow features to obtain depth features and using the depth features as finally obtained global features.
4. The demagnetization fault diagnosis method of the permanent magnet drive motor for the electric vehicle according to claim 1, wherein the step 3) of extracting the maximum stable extremum region feature of the two-dimensional wavelet time-frequency map to obtain the feature vector 2 comprises: sequentially carrying out binarization processing on the two-dimensional wavelet time-frequency graph by using a group of binarization threshold values and calculating a variable V (i) according to the following formula:
Figure FDA0003416108630000013
in the above formula, QiExpressing the area of the ith connected region in the binary region obtained by the current binary threshold, wherein delta represents the preset variable quantity of the binary threshold, and Qi+ΔRepresenting the area Q of the ith connected region in the binary region obtained after the current binary threshold value is increased by the preset binary threshold value variation deltai-ΔRepresenting the area of the ith connected region in the binary region obtained after the current binary threshold value is reduced by a preset binary threshold value variation delta; and if the variable V (i) is smaller than the set value, taking the binary region obtained by the current binarization threshold value as the obtained maximum stable extremum region characteristic and converting the binary region into a vector as a characteristic vector 2.
5. The demagnetization fault diagnosis method of the permanent magnet driving motor for the electric vehicle according to claim 1, wherein the machine learning model in the step 4) comprises a softmax classifier 1, a softmax classifier 2 and a comprehensive classifier, the softmax classifier 1 is used for obtaining a classification probability value 1 according to a feature vector 1, the softmax classifier 2 is used for obtaining a classification probability value 2 according to a feature vector 2, and the comprehensive classifier is used for obtaining a final demagnetization fault diagnosis result of the permanent magnet driving motor for the electric vehicle according to the classification probability value 1 and the classification probability value 2.
6. The demagnetization fault diagnosis method of the permanent magnet driving motor for the electric vehicle according to claim 5, wherein the comprehensive classifier is a support vector machine classifier.
7. The demagnetization fault diagnosis method of the permanent magnet driving motor for the electric vehicle according to claim 1, characterized in that step 4) is preceded by a step of training a machine learning model by using data samples, and the step of expanding the data samples by using a generated countermeasure network when the machine learning model is trained by using the data samples comprises the steps of:
s1) inputting a group of random numbers and corresponding sample labels into a generator for generating the countermeasure network, generating a virtual two-dimensional wavelet time-frequency graph with labels, and inputting the real two-dimensional wavelet time-frequency graph with labels and the virtual two-dimensional wavelet time-frequency graph into a discriminator for generating the countermeasure network;
s2) respectively calculating the network loss and the network accuracy of a generator in the generation countermeasure network and the network loss and the network accuracy of a discriminator, if the network loss and the network accuracy of one or both of the generator and the discriminator meet the requirements, judging that the training of the countermeasure network is finished, and jumping to the step S3); otherwise, adjusting the parameters of the generator and the discriminator for generating the confrontation network, and jumping to the step S1) to continue training to generate the confrontation network;
s3) generating a labeled virtual two-dimensional wavelet time-frequency graph as the two-dimensional wavelet time-frequency graph in the data sample by using the trained generator for generating the countermeasure network.
8. The demagnetization fault diagnosis method of the permanent magnet drive motor for the electric vehicle according to claim 7, wherein the functional expression of the network loss and the network accuracy of the generator in the step S2) is as follows:
Figure FDA0003416108630000021
Figure FDA0003416108630000022
in the above formula, LgTo loss of the network of the generator, SgTo the network accuracy of the generator, mean is the averaging calculation,
Figure FDA0003416108630000023
the probability that the virtual two-dimensional wavelet time-frequency graph is identified as a real two-dimensional wavelet time-frequency graph in the discriminator is obtained; the functional expression of the network loss and the network accuracy of the discriminator in step S2) is:
Figure FDA0003416108630000024
Figure FDA0003416108630000025
in the above formula, LDFor discriminator network loss, SDFor the purpose of the network accuracy of the arbiter,
Figure FDA0003416108630000026
is the probability of the real two-dimensional wavelet time-frequency diagram at the output of the discriminator network,
Figure FDA0003416108630000027
is the probability that the real two-dimensional wavelet time-frequency graph is identified as the virtual two-dimensional wavelet time-frequency graph by the discriminator.
9. An electric automobile, including the electric automobile body that has the control unit and adopts permanent magnet driving motor, the control unit includes interconnect's microprocessor and memory, microprocessor is connected with the magnetic leakage signal sensor who is used for gathering the magnetic leakage signal, its characterized in that, this microprocessor is programmed or is configured in order to carry out any one of claim 1 ~ 8 the step of the demagnetization fault diagnosis method of permanent magnet driving motor for electric automobile.
10. A computer-readable storage medium, wherein a computer program executed by a computer device to implement the method for diagnosing a demagnetization failure of a permanent magnet driving motor for an electric vehicle according to any one of claims 1 to 8 is stored in the computer-readable storage medium.
CN202111547348.9A 2021-12-16 2021-12-16 Permanent magnet driving motor demagnetizing fault diagnosis method for electric automobile and electric automobile Active CN114400940B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111547348.9A CN114400940B (en) 2021-12-16 2021-12-16 Permanent magnet driving motor demagnetizing fault diagnosis method for electric automobile and electric automobile

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111547348.9A CN114400940B (en) 2021-12-16 2021-12-16 Permanent magnet driving motor demagnetizing fault diagnosis method for electric automobile and electric automobile

Publications (2)

Publication Number Publication Date
CN114400940A true CN114400940A (en) 2022-04-26
CN114400940B CN114400940B (en) 2023-06-13

Family

ID=81226097

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111547348.9A Active CN114400940B (en) 2021-12-16 2021-12-16 Permanent magnet driving motor demagnetizing fault diagnosis method for electric automobile and electric automobile

Country Status (1)

Country Link
CN (1) CN114400940B (en)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP2518456A1 (en) * 2011-04-29 2012-10-31 ABB Technology AG Method for monitoring demagnetization
CN110823574A (en) * 2019-09-30 2020-02-21 安徽富煌科技股份有限公司 Fault diagnosis method based on semi-supervised learning deep countermeasure network
CN112052796A (en) * 2020-09-07 2020-12-08 电子科技大学 Permanent magnet synchronous motor fault diagnosis method based on deep learning
WO2021117303A1 (en) * 2019-12-09 2021-06-17 株式会社明電舎 Demagnetization diagnosis device for motor and demagnetization diagnosis method for motor control device
CN113526282A (en) * 2021-06-28 2021-10-22 江苏威尔曼科技有限公司 Method, device, medium and equipment for diagnosing medium and long-term aging faults of elevator

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP2518456A1 (en) * 2011-04-29 2012-10-31 ABB Technology AG Method for monitoring demagnetization
CN110823574A (en) * 2019-09-30 2020-02-21 安徽富煌科技股份有限公司 Fault diagnosis method based on semi-supervised learning deep countermeasure network
WO2021117303A1 (en) * 2019-12-09 2021-06-17 株式会社明電舎 Demagnetization diagnosis device for motor and demagnetization diagnosis method for motor control device
CN112052796A (en) * 2020-09-07 2020-12-08 电子科技大学 Permanent magnet synchronous motor fault diagnosis method based on deep learning
CN113526282A (en) * 2021-06-28 2021-10-22 江苏威尔曼科技有限公司 Method, device, medium and equipment for diagnosing medium and long-term aging faults of elevator

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
宋俊材: "双定子无铁芯永磁同步直线电机退磁故障识别分类研究", 《中国优秀博士学位论文全文数据库 工程科技II辑》, no. 07 *

Also Published As

Publication number Publication date
CN114400940B (en) 2023-06-13

Similar Documents

Publication Publication Date Title
Zhang et al. The multiclass fault diagnosis of wind turbine bearing based on multisource signal fusion and deep learning generative model
CN110022111B (en) Full-working-condition efficiency optimization method for permanent magnet motor in vehicle electric drive system
CN115359300B (en) Intelligent processing and manufacturing system and method for wind driven generator component
CN113869208A (en) Rolling bearing fault diagnosis method based on SA-ACWGAN-GP
CN112926728B (en) Small sample turn-to-turn short circuit fault diagnosis method for permanent magnet synchronous motor
CN116683092B (en) Disassembling system and method for new energy automobile battery
CN115294550A (en) Automatic driving automobile road scene understanding method based on multi-task learning
CN110705694A (en) Electric larceny monitoring method facing edge data center based on feature extraction
CN114400940A (en) Demagnetization fault diagnosis method for permanent magnet driving motor for electric automobile and electric automobile
CN116911201B (en) Analog motor energy efficiency detection system based on deep generation countermeasure network algorithm
CN117269754A (en) IPSM rotor demagnetizing and eccentric fault diagnosis method based on convolutional neural network operation
Ren et al. Identification of shaft centerline orbit for wind power units based on Hopfield neural network improved by simulated annealing
CN104020769A (en) Robot overall path planning method based on charge system search
CN114282580A (en) Visual image-based permanent magnet driving motor demagnetization fault diagnosis model construction method and fault diagnosis method and system
CN109190701B (en) Image classification method based on mirror image invariance convolutional neural network
Chen et al. A novel fault diagnosis method for high-speed railway turnout based on dcae-logistic regression
Xu et al. Fault diagnosis on imbalanced data using an adaptive cost-sensitive multiscale attention network
CN107274356A (en) A kind of adaptive gray level image strengthening system
CN114545279A (en) Lithium battery health state estimation method based on neural network ordinary differential equation
Mohanasundaram et al. Neuro-fuzzy controller for high performance induction motor drives
CN113177599A (en) Enhanced sample generation method based on GAN
CN113419172A (en) New energy automobile three-phase asynchronous motor fault identification based on GCN and VMD-ED
CN117388716B (en) Battery pack fault diagnosis method, system and storage medium based on time sequence data
Shouqing et al. Prediction Method of High Frequency Electromagnetic Noise of Micro Electric Vehicle based on Wavelet Time-Frequency Analysis
Wang et al. An Intelligent Identification Method of Vehicle Driving Cycle Based on LVQ Model

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant