Motor fault detection method and system

Abstract

The invention discloses a motor fault detection method and system, and the method comprises the steps: obtaining all data of a motor during the current operation through sensor equipment disposed on the motor, arranging the data in sequence, and generating an original feature vector; performing normalization processing on each component in the original feature vector to generate an input feature vector; the input feature vectors are input into N support vector machine detection models, corresponding fault detection results are generated respectively, and the N support vector machine detection models are used for detecting N different types of faults respectively; and according to the corresponding fault detection result, judging whether the motor has a fault, and when the motor is judged to have the fault, controlling the motor to decelerate and power off, and simultaneously sending a fault alarm of a corresponding fault type. The fault detection method is simple in structure, small in parameter quantity and small in calculation amount, the multiple support vector machine detection models are used for detecting each type of fault, pertinence is achieved for each type of fault, and the detection accuracy is high.

Description

Technical Field

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

[0002] The electric tricycle motor converts the battery energy into mechanical energy to propel the vehicle forward. It is the core equipment of the electric tricycle. However, due to factors of the motor itself or some external force majeure factors, the electric tricycle motor may fail during operation. If the motor is still powered on and output power at this time, it will cause permanent damage to the motor. Therefore, real-time fault detection of the motor equipment and rapid response when a fault occurs, gradually slowing down and cutting off the power, and sounding an alarm are the key to the motor's continued normal operation for a long time.

[0003] However, the embedded devices on electric tricycles cannot support the complex calculation of huge parameters of large-scale fault detection models. In addition, there are many types of motor faults, and multi-type detection of multiple types of faults is very difficult and has poor accuracy. Summary of the invention

[0004] In view of the above-mentioned technical deficiencies, the purpose of the present invention is to provide a motor fault detection method and system, aiming to solve the problem that the embedded devices on the electric tricycle cannot support the complex calculation of huge parameters of the large-scale fault detection model. In addition, there are many types of motor failures, and it is very difficult to perform multi-type detection of multiple types of failures and the accuracy is poor.

[0005] In view of the above problems, the present application provides a method and system for detecting motor faults.

[0006] The first aspect disclosed in the present application provides a method for detecting a motor fault, the method comprising the following steps: Step 1: Obtain various data of the motor during its current operation through the sensor device installed on the motor, and arrange the data in order to generate an original feature vector, wherein the data includes current, voltage, vibration acceleration, stator temperature, speed and noise sound pressure level, which correspond to various components in the original feature vector respectively; Step 2: Normalize each component of the original feature vector to generate an input feature vector; Step 3: Input the input feature vector into N support vector machine detection models to generate corresponding fault detection results, where the N support vector machine detection models are used to detect N different types of faults respectively; Step 4: Determine whether the motor is faulty based on the corresponding fault detection result. If it is determined that a fault has occurred, control the motor to slow down and cut off the power, and issue a fault alarm of the corresponding fault type.

[0007] Preferably, the step 1 specifically includes: Step 1.1: Obtain various data of the motor during operation through the sensor equipment installed on the motor, wherein the data includes current, voltage, vibration acceleration, stator temperature, speed and noise sound pressure level; Step 1.2: Represent the data described in step 1.1 as x1, x1, x3, x4, x5 and x6, respectively, where x1 represents current, x1 represents voltage, x3 represents vibration acceleration, x4 represents stator temperature, x5 represents speed, and x6 represents noise sound pressure level; Step 1.3: Arrange x1, x2, x3, x4, x5, and x6 in order to generate the original feature vector X=[x1, x2, x3, x4, x5, x6].

[0008] Preferably, the step 2 specifically includes: Step 2.1: For each component of the original eigenvector X=[x1, x2, x3, x4, x5, x6], take x * i =(x i -μ i ) / σ i The mean standard deviation is normalized by the method of i represents the component of the original eigenvector, x * i Represents x i The normalized value of the mean standard deviation, i=1,2,3,4,5,6, is used to represent the component subscripts in the original eigenvector, μ i represents the average value of the current component in all samples, σ i Represents the standard deviation of the feature in all samples; Step 2.2: Set x * i Rearrange in order to generate the input feature vector X * =[x * 1, x * 2, x * 3, x * 4, x * 5,x * 6], where x * 1 represents the current after the mean standard deviation is normalized, x * 1 represents the voltage after the mean standard deviation is normalized, x * 3 represents the vibration acceleration after the mean standard deviation is normalized, x * 4 represents the stator temperature after normalization of the mean standard deviation, x *5 represents the speed after the mean standard deviation is normalized, x * 6 represents the noise sound pressure level after normalization of the mean standard deviation.

[0009] Preferably, the step 3 specifically includes: Step 3.1: Input feature vector X * =[x * 1, x * 2, x * 3, x * 4, x * 5, x * 6] Enter SVM (m) Fault detection is performed, where SVM (m) Represents the mth support vector machine detection model, m=1,2,3,...,N, used to represent the serial number of the support vector machine detection model, SVM (m) Used to detect the mth type of fault; Step 3.2: SVM (m) Construct high-dimensional hyperplane XW respectively (m) T +b (m) =0, where W (m) It is SVM (m) The weight vector of X is the input feature vector, b (m) It is SVM (m) The bias term of Step 3.3: Use y (m) =sign(XW (m) T +b (m) ) represents SVM (m) The fault detection result is shown in Figure 2, where sign represents the sign function. For any real number a, if a>0, then sign(a)=1; if a=0, then sign(a)=0; if a<0, then sign(a)=-1.

[0010] Preferably, the step 4 specifically includes: Step 4.1: If y (m) = -1, it means SVM (m) It is detected that the mth type of fault has occurred. If y (m) =1 or 0, it means SVM (m) It is detected that the mth type of fault has not occurred at present; Step 4.2: If one or more support vector machine detection models detect that a corresponding type of fault has occurred, it is determined that the motor is currently faulty; Step 4.3: When it is determined that the motor is currently faulty, the motor is controlled to gradually decelerate and the power supply is cut off, and at the same time, a corresponding fault alarm is issued for the mth type of fault currently occurring.

[0011] Preferably, the support vector machine detection model further includes a training step before performing fault detection, and the training step includes: Collect 300 groups of data when the motor is in normal operation, and arrange the data in each group in order and perform mean standard deviation normalization to generate a motor positive sample data set. For N different types of faults, collect 300 groups of data when the motor has corresponding faults, and arrange the data in each group in order and perform mean standard deviation normalization to generate N motor negative sample data sets corresponding to N different types of faults, wherein the data include current, voltage, vibration acceleration, stator temperature, speed and noise sound pressure level, which correspond to each component in the original feature vector respectively; The SVM is trained by using the motor positive sample data set and the mth motor negative sample data set corresponding to the mth type of fault (m) The training goal is to maximize the distance from the high-dimensional hyperplane to the nearest sample point in the positive and negative sample data sets, by minimizing 1 / 2||W (m) ||2 2 To achieve this, while satisfying the constraint y (k) (θ (k) W (m) T +b (m) )≥1, where ||W (m) ||2 means SVM (m) The second norm of the weight vector, θ (k) is the kth group of data with sequence number k in the positive sample data set and the negative sample data set, y (k) is the corresponding sample label. If the kth group of data belongs to the positive sample data set, then the sample label y (k) is 1, the kth group of data belongs to the negative sample data set, then the sample label y (k) is -1.

[0012] A second aspect disclosed in the present application provides a motor fault detection system, the system comprising: An acquisition module, which is used to obtain various data of the motor during its current operation through a sensor device installed on the motor, and arrange the data in order to generate an original feature vector, wherein the data includes current, voltage, vibration acceleration, stator temperature, rotation speed and noise sound pressure level, which respectively correspond to various components in the original feature vector; A normalization module, wherein the normalization module is used to perform normalization processing on each component of the original feature vector to generate an input feature vector; A detection module, wherein the detection module is used to input the input feature vector into N support vector machine detection models to generate corresponding fault detection results respectively, wherein the N support vector machine detection models are used to detect N different types of faults respectively; The control module is used to determine whether the motor has a fault according to the corresponding fault detection result. When it is determined that a fault has occurred, the motor is controlled to slow down and cut off the power, and a fault alarm corresponding to the fault type is issued.

[0013] The beneficial effects of the present invention are: (1) The traditional machine learning method is used for fault detection, which has a simple structure, few parameters, and small computational complexity, and can be run on the embedded device on the electric tricycle.

[0014] (2) Multiple support vector machine detection models are used to detect each type of fault separately, which is targeted at each type of fault. Multi-type detection is achieved in a multiple binary classification manner, and the detection accuracy for each type of fault is high. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0016] Figure 1 The figure is a schematic diagram of the overall process of a motor fault detection method.

[0017] Figure 2 The figure is a schematic diagram of the overall structure of a motor fault detection system. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] like Figure 1 As shown, an embodiment of the present application provides a method for detecting a motor fault, the method comprising the following steps: Step 1: Obtain various data of the motor during its current operation through the sensor device installed on the motor, and arrange the data in order to generate an original feature vector, wherein the data includes current, voltage, vibration acceleration, stator temperature, speed and noise sound pressure level, which correspond to various components in the original feature vector respectively; Step 2: Normalize each component of the original feature vector to generate an input feature vector; Step 3: Input the input feature vector into N support vector machine detection models to generate corresponding fault detection results, where the N support vector machine detection models are used to detect N different types of faults respectively; Step 4: Determine whether the motor is faulty based on the corresponding fault detection result. If it is determined that a fault has occurred, control the motor to slow down and cut off the power, and issue a fault alarm of the corresponding fault type.

[0020] Furthermore, the step 1 specifically includes: Step 1.1: Obtain various data of the motor during operation through the sensor equipment installed on the motor, wherein the data includes current, voltage, vibration acceleration, stator temperature, speed and noise sound pressure level; Step 1.2: Represent the data described in step 1.1 as x1, x1, x3, x4, x5 and x6, respectively, where x1 represents current, x1 represents voltage, x3 represents vibration acceleration, x4 represents stator temperature, x5 represents speed, and x6 represents noise sound pressure level; Step 1.3: Arrange x1, x2, x3, x4, x5, and x6 in order to generate the original feature vector X=[x1, x2, x3, x4, x5, x6].

[0021] Furthermore, the step 2 specifically includes: Step 2.1: For each component of the original eigenvector X=[x1, x2, x3, x4, x5, x6], take x * i =(x i -μ i ) / σ i The mean standard deviation is normalized by the method of i represents the component of the original eigenvector, x * i Represents x i The normalized value of the mean standard deviation, i=1,2,3,4,5,6, is used to represent the component subscripts in the original eigenvector, μ i represents the average value of the current component in all samples, σ i Represents the standard deviation of the feature in all samples; Step 2.2: Set x * i Rearrange in order to generate the input feature vector X * =[x * 1, x * 2, x * 3, x * 4, x * 5,x * 6], where x * 1 represents the current after the mean standard deviation is normalized, x * 1 represents the voltage after the mean standard deviation is normalized, x * 3 represents the vibration acceleration after the mean standard deviation is normalized, x * 4 represents the stator temperature after normalization of the mean standard deviation, x * 5 represents the speed after the mean standard deviation is normalized, x * 6 represents the noise sound pressure level after normalization of the mean standard deviation.

[0022] Furthermore, the step 3 specifically includes: Step 3.1: Input feature vector X * =[x * 1, x * 2, x * 3, x * 4, x * 5, x * 6] Enter SVM (m) Fault detection is performed, where SVM (m) Represents the mth support vector machine detection model, m=1,2,3,...,N, used to represent the serial number of the support vector machine detection model, SVM (m) Used to detect the mth type of fault; Step 3.2: SVM (m) Construct high-dimensional hyperplane XW respectively (m) T +b (m) =0, where W (m) It is SVM (m) The weight vector of X is the input feature vector, b (m) It is SVM (m) The bias term of Step 3.3: Use y (m) =sign(XW (m) T +b (m) ) represents SVM (m)The fault detection result is: where sign represents the sign function. For any real number a, if a>0, then sign(a)=1; if a=0, then sign(a)=0; if a<0, then sign(a)=-1; In brief, multiple support vector machine detection models are used to detect each type of fault respectively, and each type of fault is detected in a targeted manner, and multi-type detection is achieved in a multiple binary classification manner.

[0023] Furthermore, the step 4 specifically includes: Step 4.1: If y (m) = -1, it means SVM (m) It is detected that the mth type of fault has occurred. If y (m) =1 or 0, it means SVM (m) It is detected that the mth type of fault has not occurred at present; Step 4.2: If one or more support vector machine detection models detect that a corresponding type of fault has occurred, it is determined that the motor is currently faulty; Step 4.3: When it is determined that the motor is currently faulty, the motor is controlled to gradually decelerate and the power supply is cut off, and at the same time, a corresponding fault alarm is issued for the mth type of fault currently occurring.

[0024] Furthermore, the support vector machine detection model further includes a training step before performing fault detection, and the training step includes: Collect 300 groups of data when the motor is in normal operation, and arrange the data in each group in order and perform mean standard deviation normalization to generate a motor positive sample data set. For N different types of faults, collect 300 groups of data when the motor has corresponding faults, and arrange the data in each group in order and perform mean standard deviation normalization to generate N motor negative sample data sets corresponding to N different types of faults, wherein the data include current, voltage, vibration acceleration, stator temperature, speed and noise sound pressure level, which correspond to each component in the original feature vector respectively; The SVM is trained by using the motor positive sample data set and the mth motor negative sample data set corresponding to the mth type of fault (m) The training goal is to maximize the distance from the high-dimensional hyperplane to the nearest sample point in the positive and negative sample data sets, by minimizing 1 / 2||W (m) ||2 2 To achieve this, while satisfying the constraint y (k) (θ (k) W (m) T +b (m) )≥1, where ||W (m) ||2 means SVM (m)The second norm of the weight vector, θ (k) is the kth group of data with sequence number k in the positive sample data set and the negative sample data set, y (k) is the corresponding sample label. If the kth group of data belongs to the positive sample data set, then the sample label y (k) is 1, the kth group of data belongs to the negative sample data set, then the sample label y (k) is -1; In short, in order to achieve a balance between positive and negative samples, the sample size in the motor positive sample data set and the motor negative sample data set is 300 groups. In addition, the training objective is converted into a convex quadratic programming problem, and the two types of data in the motor positive sample data set and the motor negative sample data set will compete with each other to find the optimal high-dimensional hyperplane.

[0025] In summary, the motor fault detection method provided by the embodiment of the present application has the following technical effects: (1) The traditional machine learning method is used for fault detection, which has a simple structure, few parameters, and small computational complexity, and can be run on the embedded device on the electric tricycle.

[0026] (2) Multiple support vector machine detection models are used to detect each type of fault separately, which is targeted at each type of fault. Multi-type detection is achieved in a multiple binary classification manner, and the detection accuracy for each type of fault is high.

[0027] Based on the same inventive concept as the motor fault detection method in the aforementioned embodiment, Figure 2 As shown, the present application provides a motor fault detection system, the system comprising: An acquisition module, which is used to obtain various data of the motor during its current operation through a sensor device installed on the motor, and arrange the data in order to generate an original feature vector, wherein the data includes current, voltage, vibration acceleration, stator temperature, rotation speed and noise sound pressure level, which respectively correspond to various components in the original feature vector; A normalization module, wherein the normalization module is used to perform normalization processing on each component of the original feature vector to generate an input feature vector; A detection module, wherein the detection module is used to input the input feature vector into N support vector machine detection models to generate corresponding fault detection results respectively, wherein the N support vector machine detection models are used to detect N different types of faults respectively; The control module is used to determine whether the motor has a fault according to the corresponding fault detection result. When it is determined that a fault has occurred, the motor is controlled to slow down and cut off the power, and a fault alarm corresponding to the fault type is issued.

[0028] Through the above-mentioned detailed description of a motor fault detection method in this specification, those skilled in the art can clearly understand a motor fault detection system in this embodiment. Since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0029] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0030] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for detecting motor failure, characterized in that: The method comprises the following steps: Step 1: Obtain various data of the motor during its current operation through the sensor device installed on the motor, and arrange the data in order to generate an original feature vector, wherein the data includes current, voltage, vibration acceleration, stator temperature, speed and noise sound pressure level, which correspond to various components in the original feature vector respectively; Step 2: Normalize each component of the original feature vector to generate an input feature vector; Step 3: Input the input feature vector into N support vector machine detection models to generate corresponding fault detection results, where the N support vector machine detection models are used to detect N different types of faults respectively; Step 4: Determine whether the motor is faulty based on the corresponding fault detection result. If it is determined that a fault has occurred, control the motor to slow down and cut off the power, and issue a fault alarm of the corresponding fault type.

2. A method for detecting motor failure according to claim 1, characterized in that: The step 1 specifically includes: Step 1.1: Obtain various data of the motor during operation through the sensor equipment installed on the motor, wherein the data includes current, voltage, vibration acceleration, stator temperature, speed and noise sound pressure level; Step 1.2: Represent the data described in step 1.1 as x1, x1, x3, x4, x5 and x6, respectively, where x1 represents current, x1 represents voltage, x3 represents vibration acceleration, x4 represents stator temperature, x5 represents speed, and x6 represents noise sound pressure level; Step 1.3: Arrange x1, x2, x3, x4, x5, and x6 in order to generate the original feature vector X=[x1, x2, x3, x4, x5, x6].

3. A method for detecting motor failure according to claim 2, characterized in that: The step 2 specifically includes: Step 2.1: For each component of the original eigenvector X=[x1, x2, x3, x4, x5, x6], take x * i =(x i -μ i ) / σ i The mean standard deviation is normalized by the method of i represents the component of the original eigenvector, x * i Represents x i The normalized value of the mean standard deviation, i=1,2,3,4,5,6, is used to represent the component subscripts in the original eigenvector, μ i represents the average value of the current component in all samples, σ i Represents the standard deviation of the feature in all samples; Step 2.2: Set x * i Rearrange in order to generate the input feature vector X * =[x * 1, x * 2, x * 3, x * 4, x * 5, x * 6], where x * 1 represents the current after the mean standard deviation is normalized, x * 1 represents the voltage after the mean standard deviation is normalized, x * 3 represents the vibration acceleration after the mean standard deviation is normalized, x * 4 represents the stator temperature after normalization of the mean standard deviation, x * 5 represents the speed after the mean standard deviation is normalized, x * 6 represents the noise sound pressure level after normalization of the mean standard deviation.

4. A method for detecting motor failure according to claim 3, characterized in that: The step 3 specifically includes: Step 3.1: Input feature vector X * =[x * 1, x * 2, x * 3, x * 4, x * 5, x * 6] Enter SVM (m) Fault detection is performed, where SVM (m) Represents the mth support vector machine detection model, m=1,2,3,...,N, used to represent the serial number of the support vector machine detection model, SVM (m) Used to detect the mth type of fault; Step 3.2: SVM (m) Construct high-dimensional hyperplane XW respectively (m) T +b (m) =0, where W (m) It is SVM (m) The weight vector of X is the input feature vector, b (m) It is SVM (m) The bias term of Step 3.3: Use y (m) =sign(XW (m) T +b (m) ) represents SVM (m) The fault detection result is shown in Figure 2, where sign represents the sign function. For any real number a, if a>0, then sign(a)=1; if a=0, then sign(a)=0; if a<0, then sign(a)=-1.

5. A method for detecting motor failure according to claim 4, characterized in that: The step 4 specifically includes: Step 4.1: If y (m) = -1, it means SVM (m) It is detected that the mth type of fault has occurred. If y (m) =1 or 0, it means SVM (m) It is detected that the mth type of fault has not occurred at present; Step 4.2: If one or more support vector machine detection models detect that a corresponding type of fault has occurred, it is determined that the motor is currently faulty; Step 4.3: When it is determined that the motor is currently faulty, the motor is controlled to gradually decelerate and the power supply is cut off, and at the same time, a corresponding fault alarm is issued for the mth type of fault currently occurring.

6. A method for detecting motor failure according to claim 4, characterized in that: The support vector machine detection model further includes a training step before performing fault detection, and the training step includes: Collect 300 groups of data when the motor is in normal operation, and arrange the data in each group in order and perform mean standard deviation normalization to generate a motor positive sample data set. For N different types of faults, collect 300 groups of data when the motor has corresponding faults, and arrange the data in each group in order and perform mean standard deviation normalization to generate N motor negative sample data sets corresponding to N different types of faults, wherein the data include current, voltage, vibration acceleration, stator temperature, speed and noise sound pressure level, which correspond to each component in the original feature vector respectively; The SVM is trained by using the motor positive sample data set and the mth motor negative sample data set corresponding to the mth type of fault (m) The training goal is to maximize the distance from the high-dimensional hyperplane to the nearest sample point in the positive and negative sample data sets, by minimizing 1 / 2||W (m) ||2 2 To achieve this, while satisfying the constraint y (k) (θ (k) W (m) T +b (m) )≥1, where ||W (m) ||2 means SVM (m) The second norm of the weight vector, θ (k) is the kth group of data with sequence number k in the positive sample data set and the negative sample data set, y (k) is the corresponding sample label. If the kth group of data belongs to the positive sample data set, then the sample label y (k) is 1, the kth group of data belongs to the negative sample data set, then the sample label y (k) is -1.

7. A motor fault detection system, characterized in that: The system comprises: An acquisition module, which is used to obtain various data of the motor during its current operation through a sensor device installed on the motor, and arrange the data in order to generate an original feature vector, wherein the data includes current, voltage, vibration acceleration, stator temperature, rotation speed and noise sound pressure level, which respectively correspond to various components in the original feature vector; A normalization module, wherein the normalization module is used to perform normalization processing on each component of the original feature vector to generate an input feature vector; A detection module, wherein the detection module is used to input the input feature vector into N support vector machine detection models to generate corresponding fault detection results respectively, wherein the N support vector machine detection models are used to detect N different types of faults respectively; The control module is used to determine whether the motor has a fault according to the corresponding fault detection result. When it is determined that a fault has occurred, the motor is controlled to slow down and cut off the power, and a fault alarm corresponding to the fault type is issued.

Images