Multi-fault diagnosis method and system for motor driving system

Abstract

The invention relates to the technical field of fault diagnosis, and discloses a multi-fault diagnosis method and system for a motor driving system. According to the method, a traction motor driving system discretization mathematical model, an extended Kalman filter prediction estimation model and an extended Kalman filter correction state model are constructed, fault detection variables are defined, and accurate positioning of faults is achieved. Fault features are obtained through order analysis, a multi-fault diagnosis model of the motor driving system is constructed, and rapid diagnosis of different faults can be realized.

Description

Technical Field

[0001] The present invention relates to the technical field of fault diagnosis, and in particular to a method and system for diagnosing multiple faults of a motor drive system. Background Art

[0002] The motor drive system is a core technology in the field of modern automation and is widely used in many fields such as industrial control, electricity, and transportation. However, with long-term use or the influence of the external environment, various faults may occur in the motor itself, affecting the performance and reliability of the system. Common motor faults include winding short circuit, open circuit, ground fault, demagnetization fault, overheating, etc. These faults may cause the motor to reduce efficiency, overheat or fail to start. In severe cases, the motor may be damaged and affect the operation of the entire system. In addition, the three-phase current sensor in the drive system is mainly used to monitor the current signal in real time to reflect the operating status of the motor. When the sensor fails, it will cause the current signal to be distorted or the measurement to be inaccurate, which will affect the motor operation control and cause system abnormalities or damage.

[0003] Fault diagnosis of motor drive systems mainly uses signal-based methods and observer-based methods. In existing research, signal-based methods mostly focus on detecting faults by analyzing signal characteristics such as current, voltage or vibration. However, such methods have limitations when facing complex or multiple faults, and usually cannot fully reflect the complex physical processes inside the motor, making it difficult to accurately determine the root cause of the fault. Observer-based methods often rely on accurate models of the motor. In practical applications, the nonlinearity, parameter changes and environmental factors of the motor may cause the model to not match the actual situation, affecting the accuracy of fault diagnosis. Such methods have poor adaptability in dealing with dynamic conditions, especially in the case of multiple types of faults, and are difficult to deal with complex and multiple fault modes.

[0004] Therefore, there is an urgent need to provide a method that combines observers with signal analysis to perform multi-fault diagnosis of motor drive systems. Summary of the invention

[0005] The present invention aims to disclose a motor drive system multi-fault diagnosis method and system to solve the problems existing in the prior art.

[0006] In order to achieve the above objectives, in a first aspect, the present invention provides a motor drive system multi-fault diagnosis method and system, comprising:

[0007] S1: construct a discrete mathematical model of the traction motor drive system, and construct an extended Kalman filter prediction estimation model and an extended Kalman filter correction state model according to the discrete mathematical model; obtain a stator current estimation value in the α and β coordinate systems at time k based on the extended Kalman filter prediction estimation model, the extended Kalman filter correction state model, input variables and the state estimation value at time k-1;

[0008] S2: defining a fault detection variable according to the stator current estimation value, and performing fault location based on the fault detection variable;

[0009] S3: performing order analysis on the stator current signals collected by the sensor under various fault conditions by using equal-angle resampling according to the fault location results, and determining the maximum analysis order and resampling time interval of the stator current signals under various fault conditions;

[0010] S4: linear interpolation is used to sample the stator current signal under various faults to equal angle positions, and the current resampled signal in the angular domain is obtained based on the resampling time interval; the angular domain frequency order spectrum is calculated using the DFT method, and the fault current angular domain frequency of the stator current resampled signal under various faults is obtained based on the maximum analysis order extraction, and an angular domain frequency fault feature value set is constructed; the fault current amplitude of the stator current resampled signal under various faults is obtained, and an amplitude fault feature value set is constructed;

[0011] S5: Setting a fault feature flag based on the angular domain frequency fault feature value set and the amplitude fault feature value set, and constructing a multi-fault diagnosis model for the motor drive system based on the fault feature flag to determine the phase and specific fault type of the fault.

[0012] In a second aspect, the present application also provides a control system for a motor drive system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the processor implements the steps corresponding to the method described in the first aspect.

[0013] The present invention has the following beneficial effects:

[0014] The multi-fault diagnosis method for the motor drive system provided in the present application adopts a multi-fault diagnosis strategy for the motor drive system by combining the extended Kalman observer with order analysis. The extended Kalman observer estimates the system state in real time, quickly discovers and locates faults, and copes with the complexity of the motor drive system. The frequency spectrum of the current signal during the operation of the motor is decomposed through order analysis, and the characteristic frequencies related to the fault are extracted. The signal changes caused by the motor body or load changes under complex working conditions are identified, and the typical order characteristics in the spectrum are revealed. The combination of the two can reduce the dependence on mathematical models, while adapting to the non-stationary nonlinear characteristics of actual signals, and can accurately and quickly diagnose multiple types of faults in the motor drive system, ensuring the safety and reliability of the motor drive system during operation.

[0015] The present invention will be further described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The drawings constituting a part of this application are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0017] Figure 1 is a flow chart of a multi-fault diagnosis method for a motor drive system according to a preferred embodiment of the present invention;

[0018] Figure 2 is a control block diagram of a motor drive system according to a preferred embodiment of the present invention;

[0019] Figure 3 The gain of the current sensor before and after the A phase current failure of the motor drive system according to the preferred embodiment of the present invention occurs. q Change waveform diagram;

[0020] Figure 4 This is a diagnostic indication diagram when a gain fault occurs in the A-phase current sensor of the motor drive system of the preferred embodiment of the present invention;

[0021] Figure 5 The current sensor offset fault of the phase A current sensor of the motor drive system according to the preferred embodiment of the present invention occurs before and after the current sensor offset fault occurs. q Change waveform diagram;

[0022] Figure 6 This is a diagnostic indication diagram when an A-phase current sensor of a motor drive system according to a preferred embodiment of the present invention has an offset fault;

[0023] Figure 7 It is a diagnostic indication diagram when the A-phase current sensor of the motor drive system of the preferred embodiment of the present invention is stuck. DETAILED DESCRIPTION

[0024] The technical solution of the present invention is described clearly and completely below. Obviously, the described embodiments are only a 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.

[0025] Unless otherwise defined, the technical terms or scientific terms used in the present invention shall have the usual meanings understood by persons with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. Similarly, words such as "one" or "one" do not indicate quantity restrictions, but indicate the existence of at least one. Words such as "connect" or "connected" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the object being described changes, the relative positional relationship also changes accordingly.

[0026] It is worth noting that the motor drive system multi-fault diagnosis method and system can be applied to rail transit vehicle operation scenarios, which is only used as an example and not limited here.

[0027] like Figure 1 As shown, this embodiment provides a motor drive system multi-fault diagnosis method and system, including:

[0028] S1: construct a discrete mathematical model of the traction motor drive system, and construct an extended Kalman filter prediction estimation model and an extended Kalman filter correction state model according to the discrete mathematical model; obtain a stator current estimation value in the α and β coordinate systems at time k based on the extended Kalman filter prediction estimation model, the extended Kalman filter correction state model, input variables and the state estimation value at time k-1;

[0029] S2: defining a fault detection variable according to the stator current estimation value, and performing fault location based on the fault detection variable;

[0030] S3: performing order analysis on the stator current signals collected by the sensor under various fault conditions by using equal-angle resampling according to the fault location results, and determining the maximum analysis order and resampling time interval of the stator current signals under various fault conditions;

[0031] S4: linear interpolation is used to sample the stator current signal under various faults to equal angle positions, and the current resampled signal in the angular domain is obtained based on the resampling time interval; the angular domain frequency order spectrum is calculated using the DFT method, and the fault current angular domain frequency of the stator current resampled signal under various faults is obtained based on the maximum analysis order extraction, and an angular domain frequency fault feature value set is constructed; the fault current amplitude of the stator current resampled signal under various faults is obtained, and an amplitude fault feature value set is constructed;

[0032] S5: Setting a fault feature flag based on the angular domain frequency fault feature value set and the amplitude fault feature value set, and constructing a multi-fault diagnosis model for the motor drive system based on the fault feature flag to determine the phase and specific fault type of the fault.

[0033] The above-mentioned multi-fault diagnosis method of the motor drive system combines the extended Kalman observer with order analysis to carry out a multi-fault diagnosis strategy for the motor drive system. The extended Kalman observer estimates the system state in real time, quickly discovers and locates faults, and effectively copes with the complexity of the motor drive system. The frequency spectrum of the current signal during the operation of the motor is decomposed through order analysis, and the characteristic frequencies related to the fault are effectively extracted. The signal changes caused by the motor body or load changes under complex working conditions are identified, and the typical order characteristics in the spectrum are revealed. The combination of the two can reduce the dependence on mathematical models, while adapting to the non-stationary nonlinear characteristics of actual signals, and can accurately and quickly diagnose multiple types of faults in the motor drive system, ensuring the safety and reliability of the motor drive system during operation.

[0034] This embodiment is carried out under the virtual simulation platform Simulink environment. The simulation platform consists of a voltage signal generation module, a rectifier module, a motor module, an extended Kalman observer module, an oscilloscope module, and a control module. The simulation platform is a commonly used existing technology in this field, and no further details are given here. The parameters used in the simulation experiment are shown in Table 1;

[0035] Table 1 Simulation experiment parameters

[0036] Parameters parameter DC bus voltage 600V Traction motor stator resistance 1.469Ω Traction motor q-axis inductance 0.06263H Traction motor d-axis inductance 0.02820H Traction motor moment of inertia <![CDATA[0.003kg·m 2 ]]> Traction motor flux 0.44381Wb Number of pole pairs of traction motor 2

[0037] Optionally, the S1 includes:

[0038] S11: Construct a discrete mathematical model of the traction motor drive system, the formula is:

[0039]

[0040] In the formula, i α,k 、i β,k are the stator currents in the α and β coordinate systems at time k, respectively, where the α and β coordinate systems are the two-phase stationary coordinate systems in the vector transformation, iα,k-1 、i β,k-1 are the stator currents in the α and β coordinate systems at time k-1, and u α,k-1 and u β,k-1 are the stator voltages in the α and β coordinate systems at time k-1, R is the stator resistance, ω e is the motor rotor speed, L d , L q They are the d-axis and q-axis stator inductances, T s is the sampling period, E α,k-1 、E β,k-1 They are the back electromotive force in the α and β coordinate systems at time k-1, respectively, satisfying the following relationship:

[0041]

[0042] In the formula, ψ f,k-1 is the permanent magnet flux at time k-1, θ e is the electrical angle, i d,k-1 is the d-axis stator current at time k-1, i q,k is the q-axis stator current at time k, i q,k-1 is the q-axis stator current at time k-1;.

[0043] Rewriting formula (1) into a state space equation, the formula is:

[0044] x k =Ax k-1 +Bu k-1 +E f f k-1 +ω k (3)

[0045] y k =Cx k +υ k (4)

[0046] In the formula, represents the state variable at time k-1, is the state variable at time k, that is, x k-1 The difference of is the input variable at time k-1, f k-1 =ψ f,k-1 is the fault variable at time k-1, y k is the output variable at time k; ω k is the process noise at time k, υ k is the observation noise at time k; A, B, E f , C are system matrix, control matrix, fault matrix, and output matrix respectively, and we have:

[0047]

[0048] in

[0049] S12: constructing an extended Kalman filter prediction estimation model and an extended Kalman filter correction state model respectively according to the constructed discretized mathematical model; obtaining the stator current estimation value in the α and β coordinate systems at the time k according to the input variables and the state estimation value at the time k-1;

[0050] The state value at time k is predicted based on the input variables and the estimated state value at time k-1. The formula is:

[0051]

[0052] In the formula, is the state value at time k predicted at time k-1, is the estimated value of the state at time k-1.

[0053] Calculate the prior error covariance matrix at time k, the formula is:

[0054]

[0055] In the formula, represents the prior error covariance matrix at time k; represents the posterior error covariance matrix at time k-1, Among them G k-1 , are the extended Kalman filter gain matrix and the posterior error covariance matrix at time k-1, respectively, which are calculated by formulas (7) and (9); Ψ is the Jacobian matrix of the observation model, φ represents the Jacobian matrix of the state space equation, Q k is the covariance matrix of the process noise at time k,

[0056] Formulas (5) to (6) are the constructed extended Kalman filter prediction estimation model.

[0057] According to the prior error covariance matrix at time k, the extended Kalman filter gain is calculated as follows:

[0058]

[0059] In the formula, G k represents the extended Kalman filter gain matrix at time k, R k is the covariance matrix of the observation noise at time k,

[0060] The extended Kalman filter gain matrix at time k is used to correct and update the state value at time k predicted at time k-1 to obtain the posterior estimated state at time k. The formula is:

[0061]

[0062] In the formula, Represents the posterior estimated state at time k.

[0063] Update the prior error covariance matrix at time k to obtain the posterior error covariance matrix at time k. The formula is:

[0064]

[0065] In the formula, represents the posterior error covariance matrix at time k, which is used for estimation at the next moment, and I represents the identity matrix.

[0066] Specifically, When k = 1, the initial posterior error covariance matrix is ​​as follows:

[0067]

[0068] χ is a large positive number, reflecting higher uncertainty.

[0069] Formulas (7) to (9) are the constructed extended Kalman filter correction state model.

[0070] Correct the updated state variables That is, the estimated value of the stator current in the α and β coordinate systems obtained at time k by the extended Kalman filter prediction estimation model and the correction state model. are the estimated values ​​of stator current in α and β coordinate systems at time k respectively.

[0071] The S2 includes:

[0072] S21: Convert the estimated value of the stator current in the α and β coordinate systems at time k to the estimated value of the stator current in the ABC coordinate system. The formula is:

[0073]

[0074] In the formula, They are the estimated values ​​of stator current of phase A, phase B and phase C in the ABC coordinate system at time k, where the ABC coordinate system is the three-phase stationary coordinate system in the quantity transformation.

[0075] According to the estimated value of the stator current in the ABC coordinate system, the A-phase and B-phase stator current fault detection variables are defined as follows:

[0076]

[0077] In the formula, J A , J B are the fault detection variables of the stator current of phase A and phase B respectively, N m The total number of samples for one rotation of the motor; i A,j 、i B,j are the jth sampling values ​​of the A-phase and B-phase stator currents in one motor rotation cycle, i A,j 、i B,j The corresponding estimated values ​​are obtained from formulas (8) and (10).

[0078] Set the stator current fault detection threshold ε for phase A and phase B respectively A , ε B , the formula is:

[0079] J A >ε A (13)

[0080] J B >ε B (14)

[0081] In the formula, ε A , ε B are the stator current fault detection thresholds for phase A and phase B respectively. A When the condition of formula (13) is satisfied, it is considered that phase A has a fault, otherwise phase A has no fault. B When the condition of formula (14) is met, it is considered that phase B has a fault, otherwise, phase B has no fault.

[0082] Optionally, the S3 includes:

[0083] The stator current signal collected by the sensor under the h-th type fault is analyzed by equal angle resampling, and the maximum analysis order of the stator current signal under the h-th type fault is calculated. The formula is:

[0084]

[0085] Where N h-max represents the maximum analysis order of the stator current signal under the h-th type fault, f h-fault is the motor frequency under the hth type fault, f s is the motor fundamental frequency, [·] indicates rounding down, h=0,1,2,…,H, H is the total number of fault types, and h=0 indicates no fault / normal operation.

[0086] Calculate the resampling frequency of the stator current signal under the h-th type fault, the formula is:

[0087]

[0088] In the formula, f h-re represents the resampling frequency of the stator current signal under the hth type fault, v ref Indicates the motor reference speed, v h Indicates the actual motor speed corresponding to the stator current signal under the hth type fault.

[0089] Calculate the resampling time interval of the stator current signal under the h-th type fault, the formula is:

[0090]

[0091] Where, t h-re It represents the resampling time interval of the stator current signal under the hth type fault.

[0092] Take h=0,1,2,…,H to obtain the maximum analysis order N of the stator current signal under various faults 0-max ,N 1-max ,…,N h-max ,…,N H-max , resampling frequency f 0-re ,f 1-re ,…,f h-re ,…,f H-re and the resampling time interval t 0-re ,t 1-re ,…,t h-re ,…,t H-re .

[0093] Optionally, the S4 includes:

[0094] S41: Use linear interpolation to convert the stator current signal i of phase x (x∈{A, B, C}) under the h-th type fault x,h Sampling to equal angle positions, the current resampling signal in the angular domain is obtained, and the formula is:

[0095]

[0096] In the formula, is the output current signal of the stator current sensor of phase x (x∈{A, B, C}) after resampling under the h-th type fault, k θ , k θ+1 Respectively represent the θth and θ+1th angular positions, They are respectively the kth θ , k θ+1 The resampled x-phase stator current signal corresponding to the angular position, are respectively the stator current signal i under the h-th type fault x,hIn the kth θ , k θ+1 The sampling moment corresponding to the angular position.

[0097] S42: Using DFT method to analyze the current resampling signal under the hth type fault Frequency components in the angular domain, calculate the angular frequency n h The order spectrum, as the extracted h-th fault current angular domain frequency, is given by:

[0098]

[0099] In the formula, f h-p represents the extracted hth fault current angular domain frequency, DFT(·) represents the discrete Fourier transform operation on the signal in brackets, n h represents the order of the hth fault current angular frequency, and satisfies 0 <n h ≤N h-max A positive integer.

[0100] S43: Take h=0,1,…,H, and extract the angular domain frequency f of the stator current resampling signal under various fault conditions 0-p ,f 1-p ,…,f h-p ,…,f H-p .

[0101] The angular domain frequency fluctuation factor Δf of the x-phase stator current resampling signal under the h-th type fault is introduced h-p , construct the angular domain frequency fault feature F h-p =f h-p ±Δf h-p , where the angular frequency fluctuation factor satisfies is the limit value of the angular frequency fluctuation factor, which is a very small constant. Take h=0,1,…,H, and the angular frequency fault characteristic is F 0-p ,F 1-p ,…,F h-p ,…,F H-p Considering that the angular frequency fault characteristics are equal under different types of faults, a set of angular frequency fault characteristic values ​​is constructed. l=0,1,…,L, L is the total number of angular domain frequency fault characteristic values, L<H.

[0102] S44: Obtain the amplitude of the stator current resampled signal under various fault conditions Amp h (·) represents the amplitude of the stator current resampled signal extracted under the h-th type fault.

[0103] Introducing the amplitude fluctuation factor of the x-phase stator current resampling signal under the h-th type fault Constructing Amplitude Fault Signatures The amplitude fluctuation factor satisfies λ Amp is the limit value of the amplitude fluctuation factor, which is a very small constant. Take h=0,1,…,H, and the amplitude fault feature is F 0-Amp ,F 1-Amp ,…,F h-Amp ,…,F H-Amp Considering that the amplitude fault characteristics are equal under different types of faults, a set of amplitude fault characteristic values ​​is constructed. z=0,1,…,Z, Z is the total number of amplitude fault feature values, Z<H.

[0104] Optionally, S5 includes:

[0105] S51: Set the two fault characteristic flags of the stator current resampling signal angular domain frequency and amplitude respectively, the formulas are:

[0106]

[0107] In the formula, Indicates the fault characteristic flag of the angular domain frequency of the stator current resampling signal under various faults, δ Amp Indicates the fault characteristic flag of the stator current resampling signal amplitude under various faults.

[0108] S52: Construct a multi-fault diagnosis model for the motor drive system. The formula is:

[0109]

[0110] Wherein, ξ represents the fault state, ξ=0 represents the normal operation state, and ξ=1, 2, …, H represent the 1st, 2nd, …H types of faults respectively.

[0111] After collecting the current stator current signal and determining whether the A-phase or B-phase stator current is faulty according to formula (13) or (14), two features, namely the angular domain frequency and amplitude of the A-phase or B-phase stator current resampled signal, are extracted respectively. According to formula (22), the phase where the fault occurs and the specific fault type can be determined.

[0112] Specifically, in this embodiment, a fault is injected into the current sensor at 0.2s. In the diagnostic indication diagram, 1 indicates that a fault is detected, and 0 indicates that no fault is detected. The control block diagram of the motor drive system of this embodiment is shown in FIG. Figure 2 As shown in the figure; before and after the A-phase current sensor gain fault of the motor drive system occurs q The changing waveform is as follows Figure 3 As shown; the diagnostic indication diagram when the motor drive system A phase current sensor has a gain fault is as follows Figure 4As shown in the figure; before and after the A-phase current sensor offset fault of the motor drive system occurs q The changing waveform is as follows Figure 5 As shown; the diagnostic indication diagram when the A phase current sensor of the motor drive system has an offset fault is as follows Figure 6 As shown; the diagnostic indication diagram when the A phase current sensor of the motor drive system is stuck is as follows Figure 7 shown.

[0113] In summary, the present application provides a method for multi-fault diagnosis of a motor drive system, which takes the rail transit motor drive system as the research object, and analyzes the application of the extended Kalman filter and order analysis method in the multi-fault diagnosis of the rail transit motor drive system. By constructing a discrete mathematical model of the traction motor drive system, an extended Kalman filter prediction estimation model, and an extended Kalman filter correction state model, fault detection variables are defined to achieve accurate positioning of the fault. By obtaining fault characteristics through order analysis and constructing a multi-fault diagnosis model for the motor drive system, rapid diagnosis of different faults can be achieved.

[0114] Corresponding to the above method embodiment, this embodiment provides a motor drive system multi-fault diagnosis system, including a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the above method when executing the computer program. The motor drive system multi-fault diagnosis system can implement various embodiments of the above motor drive system multi-fault diagnosis method, and can achieve the same beneficial effects, which will not be described in detail here.

[0115] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A multi-fault diagnosis method for a motor drive system, characterized in that: include: S1: construct a discrete mathematical model of the traction motor drive system, and construct an extended Kalman filter prediction estimation model and an extended Kalman filter correction state model according to the discrete mathematical model; obtain a stator current estimation value in the α and β coordinate systems at time k based on the extended Kalman filter prediction estimation model, the extended Kalman filter correction state model, input variables and the state estimation value at time k-1; S2: defining a fault detection variable according to the stator current estimation value, and performing fault location based on the fault detection variable; S3: performing order analysis on the stator current signals collected by the sensor under various fault conditions by using equal-angle resampling according to the fault location results, and determining the maximum analysis order and resampling time interval of the stator current signals under various fault conditions; S4: linear interpolation is used to sample the stator current signals under various faults to equal angle positions, and the current resampled signals in the angular domain are obtained based on the resampling time interval; the angular domain frequency order spectrum is calculated using the DFT method, and the fault current angular domain frequency of the stator current resampled signals under various faults is obtained based on the maximum analysis order extraction, and a set of angular domain frequency fault feature values ​​is constructed; Obtain the fault current amplitude of the stator current resampling signal under various fault conditions and construct a set of amplitude fault feature values; S5: Setting a fault feature flag based on the angular domain frequency fault feature value set and the amplitude fault feature value set, and constructing a multi-fault diagnosis model for the motor drive system based on the fault feature flag to determine the phase and specific fault type of the fault.

2. The motor drive system multi-fault diagnosis method according to claim 1, characterized in that: The S1 includes: S11: Construct a discrete mathematical model of the traction motor drive system to satisfy the following relationship: In the formula, i α,k 、i β,k are the stator currents in the α and β coordinate systems at time k, respectively, where the α and β coordinate systems are the two-phase stationary coordinate systems in the vector transformation, i α,k-1 、i β,k-1 are the stator currents in the α and β coordinate systems at time k-1, and u α,k-1 and u β,k-1 are the stator voltages in the α and β coordinate systems at time k-1, R is the stator resistance, ω e is the motor rotor speed, L d , L q They are the d-axis and q-axis stator inductances, T s is the sampling period, E α,k-1 、E β,k-1 They are the back electromotive force in the α and β coordinate systems at time k-1, respectively, satisfying the following relationship: In the formula, ψ f,k-1 is the permanent magnet flux at time k-1, θ e is the electrical angle, i d,k-1 is the d-axis stator current at time k-1, i q,k is the q-axis stator current at time k, i q,k-1 is the q-axis stator current at time k-1; Rewriting formula (1) into a state space equation, the formula is: x k =Ax k-1 +Bu k-1 +E f f k-1 +ω k (3) y k =Cx k +u k (4) In the formula, represents the state variable at time k-1, is the state variable at time k, that is, x k-1 The difference of is the input variable at time k-1, f k-1 =ψ f,k-1 is the fault variable at time k-1, y k is the output variable at time k; ω k is the process noise at time k, υ k is the observation noise at time k; A, B, E f , C are system matrix, control matrix, fault matrix and output matrix respectively; S12: Predict the state value at time k based on the input variables and the estimated state value at time k-1, satisfying the following relationship: In the formula, is the state value at time k predicted at time k-1, is the estimated value of the state at time k-1; Calculate the prior error covariance matrix at time k, satisfying the following relationship: In the formula, represents the prior error covariance matrix at time k; represents the posterior error covariance matrix at time k-1, Among them G k-1 , are the extended Kalman filter gain matrix and the posterior error covariance matrix at time k-1, Ψ is the Jacobian matrix of the observation model, φ represents the Jacobian matrix of the state space equation, Q k is the covariance matrix of the process noise at time k, Formulas (5) to (6) are used as the constructed extended Kalman filter prediction estimation model; According to the prior error covariance matrix at time k, the extended Kalman filter gain is calculated to satisfy the following relationship: In the formula, G k represents the extended Kalman filter gain matrix at time k, R k is the covariance matrix of the observation noise at time k, The extended Kalman filter gain matrix at time k is used to correct and update the state value at time k predicted at time k-1 to obtain the posterior estimated state at time k, which satisfies the following relationship: In the formula, Represents the posterior estimated state at time k; Update the prior error covariance matrix at time k to obtain the posterior error covariance matrix at time k, which satisfies the following relationship: In the formula, represents the posterior error covariance matrix at time k, which is used for estimation at the next moment, and I represents the identity matrix; Formulas (7) to (9) are used as the constructed extended Kalman filter correction state model; The state variables obtained by correcting the update As the stator current estimation value in the α and β coordinate systems obtained at time k by the extended Kalman filter prediction estimation model and the extended Kalman filter correction state model, are the estimated values ​​of stator current in α and β coordinate systems at time k respectively.

3. The motor drive system multi-fault diagnosis method according to claim 2, characterized in that: The S2 includes: S21: Convert the estimated stator current value in the α and β coordinate systems at time k into the estimated stator current value in the ABC coordinate system, satisfying the following relationship, where the ABC coordinate system is a three-phase stationary coordinate system in the quantity transformation: In the formula, are the estimated values ​​of the stator currents of phase A, phase B, and phase C in the ABC coordinate system at time k, i α,k 、i β,k are the stator currents in the α and β coordinate systems at time k respectively; According to the estimated value of the stator current in the ABC coordinate system, the A-phase and B-phase stator current fault detection variables are defined to satisfy the following relationship: In the formula, J A , J B are the fault detection variables of the stator current of phase A and phase B respectively, N m The total number of samples for one rotation of the motor; i A,j 、i B,j are the jth sampling values ​​of the A-phase and B-phase stator currents in one motor rotation cycle, i A,j 、i B,j The corresponding estimated value; Set the stator current fault detection threshold ε for phase A and phase B respectively A , ε B , satisfying the following relationship: J A >e A (13) J B >e B (14) In the formula, ε A , ε B are the stator current fault detection thresholds for phase A and phase B respectively. When fault location is performed, when J A When the condition of formula (13) is satisfied, it is considered that phase A has a fault, otherwise phase A has no fault. B When the condition of formula (14) is met, it is considered that phase B has a fault, otherwise, phase B has no fault.

4. The motor drive system multi-fault diagnosis method according to claim 1, characterized in that: The S3 includes: The stator current signal collected by the sensor under the h-th type fault is analyzed by equal angle resampling, and the maximum analysis order of the stator current signal under the h-th type fault is calculated to satisfy the following relationship: Where N h-max represents the maximum analysis order of the stator current signal under the h-th type fault, f h-fault is the motor frequency under the hth type fault, f s is the motor fundamental frequency, [·] indicates rounding down, h=0,1,2,…,H, H is the total number of fault types, h=0 indicates no fault / normal operation; Calculate the resampling frequency of the stator current signal under the h-th type fault to satisfy the following relationship: In the formula, f h-re represents the resampling frequency of the stator current signal under the hth type fault, v ref Indicates the motor reference speed, v h Indicates the actual motor speed corresponding to the stator current signal under the hth type fault; Calculate the resampling time interval of the stator current signal under the h-th type fault, satisfying the following relationship: Where, t h-re represents the resampling time interval of the stator current signal under the hth type fault; Take h=0,1,2,…,H to obtain the maximum analysis order N of the stator current signal under various faults 0-max ,N 1-max ,…,N h-max ,…,N H-max , resampling frequency f 0-re ,f 1-re ,…,f h-re ,…,f H-re and the resampling time interval t 0-re ,t 1-re ,…,t h-re ,…,t H-re .

5. The motor drive system multi-fault diagnosis method according to claim 4, characterized in that: The S4 includes: S41: Use linear interpolation to convert the x-phase stator current signal i under the h-th type fault x,h Sampling to equal angle positions, the current resampled signal in the angular domain is obtained, x∈{A, B, C}, which satisfies the following relationship: In the formula, is the output current signal of the x-phase stator current sensor after resampling under the h-th type fault, k θ , k θ+1 Respectively represent the θth and θ+1th angular positions, They are respectively the kth θ , k θ+1 The resampled x-phase stator current signal corresponding to the angular position, are the stator current signal i under the h-th type fault x,h In the kth θ , k θ+1 The sampling time corresponding to the angular position; S42: Using DFT method to analyze the current resampling signal under the hth type fault Frequency components in the angular domain, calculate the angular frequency n h The order spectrum, as the extracted h-th fault current angular domain frequency, satisfies the following relationship: In the formula, f h-p represents the extracted hth fault current angular domain frequency, DFT(·) represents the discrete Fourier transform operation on the signal in brackets, n h represents the order of the hth fault current angular frequency, and satisfies 0 <n h ≤N h-max A positive integer of ; S43: Take h=0,1,…,H, and extract the angular domain frequency of the stator current resampled signal under various fault conditions: f 0-p ,f 1-p ,…,f h-p ,…,f H-p ; The angular domain frequency fluctuation factor Δf of the x-phase stator current resampling signal under the h-th type fault is introduced h-p , construct the angular domain frequency fault feature F h-p =f h-p ±Δf h-p , where the angular frequency fluctuation factor satisfies is the limit value of the angular frequency fluctuation factor, taking h=0,1,…,H, and the angular frequency fault characteristic is F 0-p ,F 1-p ,…,F h-p ,…,F H-p ; Considering that the angular frequency fault characteristics under different types of faults are equal, a set of angular frequency fault characteristic values ​​is constructed. L is the total number of angular domain frequency fault characteristic values, L<H; S44: The amplitude of the stator current resampled signal under various fault conditions is obtained as follows: Amp h (·) represents the amplitude of the stator current resampled signal extracted under the h-th type fault; Introducing the amplitude fluctuation factor of the x-phase stator current resampling signal under the h-th type fault Constructing Amplitude Fault Signatures The amplitude fluctuation factor satisfies λ Amp is the limit value of the amplitude fluctuation factor, taking h=0,1,…,H, and the amplitude fault feature is F 0-Amp ,F 1-Amp ,…,F h-Amp ,…,F H-Amp , considering that the amplitude fault characteristics are equal under different types of faults, construct the amplitude fault characteristic value set z=0,1,…,Z, Z is the total number of amplitude fault feature values, Z<H.

6. The motor drive system multi-fault diagnosis method according to claim 5, characterized in that: The S5 includes: S51: Set the two fault characteristic flags of the angular domain frequency and amplitude of the stator current resampling signal respectively to satisfy the following relationship: In the formula, Indicates the fault characteristic flag of the angular domain frequency of the stator current resampling signal under various faults, δ Amp Indicates the fault characteristic flag bit of the stator current resampling signal amplitude under various faults; S52: Construct a multi-fault diagnosis model for the motor drive system, satisfying the following relationship: Where, ξ represents the fault state, ξ=0 represents the normal operation state, ξ=1, 2, …, H represent the 1st, 2nd, …H type of faults respectively; After collecting the current stator current signal and determining whether the A-phase or B-phase stator current is faulty according to formula (13) or (14), two features, namely the angular domain frequency and amplitude of the A-phase or B-phase stator current resampled signal, are extracted respectively. According to formula (22), the phase where the fault occurs and the specific fault type are determined.

7. A multi-fault diagnosis system for a motor drive system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

Images