Fault early warning and detection method for permanent magnet synchronous motor

Abstract

The invention belongs to the technical field of engineering machinery control, and discloses a permanent magnet synchronous motor fault early warning and detection method, which comprises the following steps of: obtaining a system model of a permanent magnet synchronous motor by adopting a mechanism modeling method, and calculating the fault early warning of the permanent magnet synchronous motor through different analysis methods of the system model according to an optimization theory under constraint conditions; different minimum output deviation functions are designed under different external cause conditions, numerical analysis is carried out on the deviation functions to achieve triggering conditions of different faults, and meanwhile the controller transmits control signals to the fault early warning device to achieve detection of different faults. According to the method disclosed by the invention, the defects that a traditional motor fault diagnosis method is not timely, fault point judgment is inaccurate and misjudgment is easily caused by uncertain interference factors are overcome, and the steady-state performance and the transient-state performance of a permanent magnet synchronous motor system are enhanced.

Description

Technical Field

[0001] The invention relates to a permanent magnet synchronous motor fault early warning and detection method, which is applied to an electric drive control and vehicle alarm system of a mining dump truck and belongs to the technical field of engineering machinery control. Background Art

[0002] Permanent magnet synchronous motors have the advantages of good speed regulation performance, simple structure, and high efficiency. They have been widely used in the fields of new energy vehicles and engineering machinery. However, due to the influence of external interference and uncertain interference factors during its operation, various faults occur in the system, thus affecting the efficiency of engineering operation. Therefore, designing a more efficient fault warning method is crucial to the stable operation of the motor. At present, the fault warning of permanent magnet motors is mostly based on the system model, designing different functions to analyze different fault points. However, the permanent magnet motor system is a nonlinear system. The traditional modeling method ignores the influence of external interference and uncertain interference factors of the system, resulting in the failure of the designed method to fully and accurately detect the fault point. Summary of the invention

[0003] In order to solve the technical problems existing in the existing permanent magnet motor fault warning technology, the present invention provides a permanent magnet synchronous motor fault warning and detection method, which utilizes the displacement / speed sensor and current sensor data of the permanent magnet synchronous motor system, adopts a mechanism method to establish a system model under uncertain interference factors, and calculates the state of the motor system output under sensor, stator phase short circuit, stator turn-to-turn short circuit, and demagnetization faults, performs numerical analysis on different states, and realizes early warning of different faults.

[0004] A permanent magnet synchronous motor fault warning and detection method adopts a mechanism modeling method to obtain a system model of the permanent magnet synchronous motor. When the permanent magnet synchronous motor fails, the controller analyzes sensor data and determines whether the permanent magnet synchronous motor has an internal fault or an external fault through sensor fault diagnosis based on a matrix full-rank robust observer. When it is determined to be an internal fault, the stator turn-to-turn short circuit fault diagnosis based on a fuzzy neural network, the stator phase-to-phase fault diagnosis based on a model predictive control, and the demagnetization fault diagnosis based on a minimum marginal magnetic density function are used to respectively determine the stator turn-to-turn short circuit fault, the stator phase-to-phase short circuit fault and the demagnetization fault of the permanent magnet synchronous motor. Finally, an alarm is given to the fault and the fault point is displayed.

[0005] Furthermore, sensor fault diagnosis based on robust observer with full matrix rank establishes a model by considering system modeling error and uncertainty interference factors, and designs a robust observer to make the system reach a stable state, and uses system deviation to determine whether the system has an external sensor failure.

[0006] Furthermore, without considering factors such as modeling dynamics and external uncertainty interference, when the system fails, there is Whether the system has a sensor failure is determined as follows

[0007]

[0008] Furthermore, the stator inter-turn short-circuit fault diagnosis based on fuzzy neural network utilizes a multi-layer forward fuzzy neural network to perform predictive analysis on the system model, and corrects the weights in combination with the fuzzy control idea, performs fitting analysis on the difference between the system output and the expected output, and judges the system's inter-turn short-circuit fault by analyzing the frequency and phase of the fitting function.

[0009] Furthermore, by utilizing the ability of neural networks to approximate arbitrary nonlinear functions, an approximate fitting function y=a*sin(wx+φ) is adopted, where the amplitude a>0, and the frequency w and phase φ are used to separate faults, diagnose the strength of the inter-turn short-circuit fault, and determine the stator winding where the fault occurs.

[0010] Furthermore, the stator interphase fault diagnosis based on model predictive control discretizes the system model to obtain the predicted output of the system within a given time, designs the criterion function under the mutual influence of the phases, obtains the optimal output sequence that stabilizes the system, and analyzes the weight coefficients of the criterion function to determine the phase-to-phase short-circuit fault of the permanent magnet synchronous motor.

[0011] Furthermore, after discretizing the system model, a tracking differentiator TD is added to the system output and the expected input signal. The expression of TD is as follows:

[0012]

[0013] Wherein, x(k) represents the input signal; x1(k) represents the tracking signal of the input signal; x2(k) represents the differential signal of the input signal; r represents the tracking speed factor; h represents the filter factor; and fhan is a nonlinear function.

[0014] Furthermore, the criterion function is

[0015] J=||I(k+N)-I′(k+N)|| 2 +||λ-g(u,i a ,i b ,i c )|| 2 +||μ-h(u,i a ,i b ,i c ,i′ a ,i′ b ,i′ c )|| 2 ,

[0016] In the formula, g(i a ,i b ,i c ) represents the current difference caused by the same bridge phase, h(i a ,i b ,i c ,i′ a ,i′ b ,i′ c ) represents the current difference caused by the phases of different bridges; the criterion function is calculated with respect to the voltage u d , d q By taking the partial derivative of and setting it to zero, we can obtain the optimal control input sequence that stabilizes the system;

[0017] when u=u 最优 In the criterion function, if λ=0, μ≠0, it means that an inter-phase short circuit fault occurs in different bridges; if λ≠0, μ=0, it means that an inter-phase short circuit fault occurs in the same bridge; if λ=μ=0, it means that the system operates normally.

[0018] Furthermore, the demagnetization fault diagnosis based on the minimum marginal magnetic density function is carried out on the system model by Taylor expansion analysis. The function of the current change when the motor is running normally and faulty is used for analysis. The minimum marginal magnetic density function of the motor current is designed, and the high-order powers of the rotor angle are used to determine whether the system has a demagnetization fault.

[0019] Furthermore, the minimum edge magnetic density function is

[0020] e=I 故障 (θ)-I 正常 (θ) = g(θ,θ 2 ,θ 3 ,θ 4 ,θ 5 ,)+h(θ 6 )-f 正常 (θ)+ε,

[0021] The minimum edge magnetic density function is obtained to obtain the value range of the minimum edge magnetic density function θ 边缘 , if |θ|≤θ 边缘 Indicates that the system is running normally, if |θ|>θ 边缘 Indicates that a demagnetization fault has occurred in the system.

[0022] Furthermore, using θ 边缘 Observe the change in current amplitude. If the current amplitude increases, a uniform demagnetization fault occurs. Determine whether a non-uniform demagnetization fault occurs based on whether the third or higher harmonics appear in the current.

[0023] Compared with the prior art, the present invention has the following beneficial effects:

[0024] Compared with the traditional mechanism modeling method, which easily ignores the influence of uncertain interference factors on the system, the present invention fully considers such problems, and analyzes the output of the system in combination with different design methods. The system model can be obtained by using the input and output data of the system, which overcomes the problem that the accurate model of the permanent magnet synchronous motor is difficult to establish in engineering practice and the modeling error has unmodeled dynamic problems. The design method is simple and easy to apply in practice, and the system has stronger fault diagnosis ability. The present invention solves the disadvantages of the traditional motor fault diagnosis method that is not timely, the fault point is not accurately judged, and it is easy to be misjudged by uncertain interference factors, and enhances the steady-state performance and transient performance of the permanent magnet synchronous motor system.

[0025] A permanent magnet synchronous motor fault warning and detection method designed by the present invention can perform a fault alarm after a motor fault occurs, and perform a warning analysis before a fault occurs, and can use the system's online or offline data for warning analysis, including the permanent magnet motor's displacement / speed, current, voltage, back electromotive force, and torque parameter data. The method of the present invention fully considers the influence of the system's uncertain factors, avoids the drawbacks of incomplete modeling of traditional methods, enhances the authenticity of the data, and improves the stability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 A schematic diagram of a permanent magnet synchronous motor according to the present invention;

[0027] Figure 2 The permanent magnet synchronous motor principle diagram of the present invention;

[0028] Figure 3 The present invention is a block diagram of a permanent magnet synchronous motor control based on a robust observer;

[0029] Figure 4 It is a structural diagram of a three-layer feedforward neural network of the present invention;

[0030] Figure 5 The control block diagram of the permanent magnet synchronous motor based on fuzzy neural network of the present invention;

[0031] Figure 6 The control block diagram of the permanent magnet synchronous motor based on model predictive control of the present invention;

[0032] Figure 7 It is the minimum edge magnetic density function block diagram of the present invention;

[0033] Figure 8 This is the fault tree alarm analysis of the present invention. DETAILED DESCRIPTION

[0034] In order to make the technical solution of the present invention clearer, the present invention is further described below in conjunction with the accompanying drawings.

[0035] The working principle of the permanent magnet synchronous motor fault warning and detection method is as follows: the sensor collects system data, the controller models the collected data, and fits and analyzes it with the expected output data of the motor under normal working conditions to obtain the error fitting curve. Finally, the controller issues a warning based on the triggering conditions of different faults.

[0036] Implementation method: Permanent magnet synchronous motor fault warning and detection method, while using sensor fault diagnosis based on matrix full rank robust observer, stator turn-to-turn short circuit fault diagnosis based on fuzzy neural network, stator phase-to-phase fault diagnosis based on model predictive control, and demagnetization fault diagnosis based on minimum marginal magnetic density function, to calculate the output data of the motor in normal operation and different faults in real time, and automatically design criterion functions under different conditions to determine the triggering conditions of different faults. Specifically: fault diagnosis based on robust observer of matrix singular value decomposition, on the basis of the model of uncertain interference factors, calculates the appropriate observer input matrix, and then calculates the system current change, and judges whether the sensor fault occurs by the size of the observer value; combines the multi-layer forward neural network to predict the model output of the system, uses the fuzzy controller to achieve stable control of the system, and at the same time fits and analyzes the deviation between the actual output and the expected output of the system to judge the turn-to-turn short-circuit fault; combines the model predictive control idea to predict the system model, designs the criterion function of the phase-to-phase influence to make the system stable and obtain the optimal input sequence of the system, and judges the phase-to-phase short-circuit fault by the size of the weight; based on the demagnetization fault diagnosis of the minimum marginal magnetic density function, the system is expanded to analyze the system, and the high-order terms of the current and the high-order harmonics are considered to judge the demagnetization fault of the system.

[0037] Permanent magnet synchronous motor model establishment

[0038] like Figure 1 The figure shows a simplified structural diagram of a permanent magnet synchronous motor, which is mainly composed of a stator 1, a rotor 2, a permanent magnet 3, and a stator coil 4.

[0039] like Figure 2 The figure shows the working principle diagram of the permanent magnet synchronous motor. The permanent magnet DC motor system is essentially a nonlinear system. At present, most modeling methods are to treat the system as a linear time-invariant system. For this reason, the mechanism modeling method is used to obtain the stationary reference coordinate system model (i.e., system model) of the permanent magnet synchronous motor system:

[0040]

[0041] In the formula, u d ,u q , I d , I q , E d , E qare the voltage, current, and back electromotive force of the stator on the d-axis and q-axis, respectively. s is the stator phase resistance, L dq is the matrix of inductance on the d-axis and q-axis, and the expression is

[0042]

[0043] Where, L d is the inductance when the stator and rotor windings are axially symmetrical, L q is the inductance when the stator and rotor winding axes are orthogonal, and θ is the angle of the rotor.

[0044] The expression of back EMF is as follows:

[0045]

[0046] In the formula, is the rotor speed of the motor, and λ is the flux linkage of the permanent magnet.

[0047] The electromagnetic torque expression is:

[0048]

[0049] Where p is the number of pole pairs of the permanent magnet synchronous motor.

[0050] Sensor Fault Diagnosis Based on Robust Observer with Full Rank Matrix

[0051] The sensor fault diagnosis method based on the robust observer with full matrix rank only analyzes the system model. The principle is to establish a model by considering the system modeling error and uncertainty interference factors, and design a robust observer to make the system reach a stable state, and use the system deviation to determine whether the system has a sensor failure.

[0052] Note: Modeling error refers to the error between the permanent magnet synchronous motor model obtained by the mechanism method and the actual permanent magnet synchronous motor model due to the influence of noise, environment, etc.

[0053] For the linear steady-state system, equation (1) is rewritten into the state space expression form, and the permanent magnet motor model is established by considering the unmodeled dynamic and uncertain interference factors:

[0054]

[0055] Where x, y, and u represent the state, output observable, and control variable of the permanent magnet synchronous motor system, respectively; A, B, and C are matrices of appropriate dimensions; ΔA, ΔB, and ΔC are modeling errors; ΔN is the error caused by external interference; f is the error caused by the actuator; and f0 is the error caused by the sensor.

[0056] Note: Unmodeled dynamics refers to the use of mechanism modeling methods, without considering the data states of permanent magnet synchronous motor speed, torque, etc. caused by mechanical wear between permanent magnet synchronous motors;

[0057] Dynamic modeling refers to the use of mechanism modeling methods to consider the data states of permanent magnet synchronous motor speed, torque, etc. caused by mechanical wear between permanent magnet synchronous motors.

[0058] According to formula (5), the design is as follows Figure 3 The permanent magnet synchronous motor control block diagram based on the robust observer is shown in FIG. , and the observer matrix K can make the following equation hold:

[0059]

[0060] make Then there is

[0061]

[0062] At this time, if there is no modeling error and external uncertainty interference factors and no sensor failure, the gain matrix K is reasonably selected so that (A+KC) is full rank, that is, formula (7) is stable, then

[0063] If we do not consider factors such as modeling dynamics and external uncertainty interference, when a system failure occurs, there is Therefore, the following logic can be used to determine whether the system has a sensor failure.

[0064]

[0065] Stator Turn-to-Turn Short Circuit Fault Diagnosis Based on Fuzzy Neural Network

[0066] The stator turn-to-turn short-circuit fault diagnosis based on fuzzy neural network uses a multi-layer forward fuzzy neural network to predict and analyze the system model, and corrects the weights in combination with fuzzy control ideas. The difference between the system output and the expected output is fitted and analyzed. The influence of the system's turn-to-turn short-circuit fault is determined by analyzing the frequency and phase of the fitting function, and the fault is located in time on which winding of the A, B, and C phase stator windings the fault occurs.

[0067] Based on formula (1), the current and voltage data of the permanent magnet synchronous motor are used to establish the permanent magnet synchronous motor turn-to-turn short circuit fault model using the three-layer forward fuzzy neural network modeling method, as shown in Figure 4 The figure shows the structure of a three-layer forward fuzzy neural network. The activation function of the neuron uses the Sigmoid function, that is, Where a is a variable parameter used to change the gradient of the excitation function and enhance the adaptability of the function, and e is the difference between the voltage and current output of the permanent magnet synchronous motor model established using equation (1) and the voltage and current output of the actual permanent magnet synchronous motor.

[0068] The three-layer forward fuzzy neural network modeling method uses fuzzy control ideas to correct the weights of the neural network to improve the stability of the system. Figure 5 As shown. The difference between the model system and the actual system is fitted and analyzed, and fault detection is performed. At the same time, the ability of the neural network to approximate any nonlinear function is used to approximate the fitting function y=a*sin(wx+φ), the amplitude of the function a>0, and the frequency w and phase φ are used for fault separation to diagnose the strength of the inter-turn short-circuit fault and determine which phase of the stator winding the fault occurs in. The specific value range is:

[0069]

[0070] Among them, w min , w max It determines the impact of the inter-turn short-circuit fault on the system and determines whether it can continue to be used. k is an integer, kπ, It indicates that the inter-turn short-circuit fault occurs in the stator windings of phases A, B, and C respectively.

[0071] Stator Phase Fault Diagnosis Based on Model Predictive Control

[0072] The stator interphase fault diagnosis method based on model predictive control discretizes the system model to obtain the predicted output of the system within a given time, designs the criterion function under the mutual influence of the phases, obtains the optimal output sequence that stabilizes the system, and analyzes the weight coefficients of the criterion function to determine whether the motor has a same-phase or different-phase short-circuit fault.

[0073] Rewrite equation (1) into state space expression form:

[0074]

[0075] Combined with the forward Euler discretization method, that is

[0076]

[0077] Where T = 0.01 is the sampling time.

[0078] Formula (10) is used to discretize formula (9), and the model of the system after N steps is established using the model prediction idea, and I(k+N)=[I d (k+N)I q (k+N)] T The output status of

[0079] Considering that the model prediction idea will increase the system modeling error by discretizing the system, in order to improve the accuracy of the prediction model, a tracking differentiator (TD) is added to the system output and expected input signal. TD can increase the gain of the system error and improve the robustness of the system. At the same time, when a fault occurs, it can effectively reduce the impact of interference on the system and protect the service life of the machine.

[0080] The expression of TD is as follows:

[0081]

[0082] Wherein, x(k) represents the input signal; x1(k) represents the tracking signal of the input signal; x2(k) represents the differential signal of the input signal; r represents the tracking speed factor; h represents the filter factor; fhan is the nonlinear function, and fh is the filter function.

[0083] The structural diagram of phase-to-phase short circuit fault under the model prediction idea is as follows Figure 6 As shown in Figure 2, the criterion function for phase-to-phase short-circuit fault is designed by considering the mutual influence between phases.

[0084] J=||I(k+N)-I′(k+N)|| 2 +||λ-g(u,i a ,i b ,i c )|| 2 +||μ-h(u,i a ,i b ,i c ,i′ a ,i′ b ,i′ c )|| 2 (12)

[0085] In the formula, g(i a ,i b ,i c ) represents the current difference caused by the same bridge phase, h(i a ,i b ,i c ,i′ a ,i′ b ,i′ c ) represents the current difference caused by the phases of different bridges.

[0086] For equation (12), find the voltage u d , d q By taking the partial derivative of and setting it to zero, we can obtain the optimal control input sequence that stabilizes the system.

[0087] when u=u 最优In formula (12), if λ=0 and μ≠0, it means that an inter-phase short circuit fault occurs in different bridges; if λ≠0 and μ=0, it means that an inter-phase short circuit fault occurs in the same bridge; if λ=μ=0, it means that the system operates normally.

[0088] Demagnetization Fault Diagnosis Based on Minimum Marginal Magnetic Density Function

[0089] The demagnetization fault diagnosis method based on the minimum marginal magnetic density function is to carry out Taylor expansion analysis on the system model, and use the function of the current change when the motor is running normally and faulty to analyze, design the minimum marginal magnetic density function of the motor current, and use the high-order power of the rotor angle to determine whether the system has a uniform or uneven demagnetization fault.

[0090] Considering the external influence on the motor performance, ε is defined as the allowable speed measurement error and the current error caused by the electromagnetic interference error. Figure 7 The minimum edge magnetic density function block diagram is shown.

[0091] Field strength is the strength of the magnetic field in a permanent magnet motor and is defined as:

[0092] H=N*I / L(13)

[0093] Where H is the magnetic field strength, N is the number of turns of the magnetic coil, I is the current passing through the magnetic coil, and L is the length of the magnetic circuit.

[0094] Using the Taylor series, the expansion around x = a is

[0095]

[0096] Among them, o(x n ) represents x n Higher-order infinitesimals.

[0097] Combining equations (14) and (15) with equations (1) (2) (3) and (4), we get the polynomial of the system with respect to current:

[0098] I 故障 (θ) = g(θ,θ 2 ,θ 3 ,θ 4 ,θ 5 ,)+h(θ 6 )(16)

[0099] In formula (16), h(θ 6 ) is a high-order additional term, defined as the current change caused by the system demagnetization fault.

[0100] The minimum magnetic density function is established as follows:

[0101] e=I故障 (θ)-I 正常 (θ) = g(θ,θ 2 ,θ 3 ,θ 4 ,θ 5 ,)+h(θ 6 )-f 正常 (θ)+ε (17)

[0102] The minimum edge magnetic density function value range θ can be obtained from equation (17): 边缘 , if |θ|≤θ 边缘 Indicates that the system is running normally, if |θ|>θ 边缘 Indicates that a demagnetization fault has occurred in the system.

[0103] At the same time, using θ 边缘 Observe the change in current amplitude. If the current amplitude increases, a uniform demagnetization fault occurs. Determine whether a non-uniform demagnetization fault occurs based on whether the third or higher harmonics appear in the current.

[0104] Fault location and early warning method for permanent magnet synchronous motor based on fault tree structure

[0105] like Figure 8 As shown, when a permanent magnet synchronous motor fails, the controller analyzes the sensor data and determines whether the permanent magnet synchronous motor has an internal fault or an external sensor fault through sensor fault diagnosis based on a robust observer with full matrix rank; when an internal fault is determined, the stator turn-to-turn short circuit fault diagnosis based on a fuzzy neural network, the stator phase-to-phase fault diagnosis based on model predictive control, and the demagnetization fault diagnosis based on the minimum marginal magnetic density function are used to respectively determine the stator turn-to-turn short circuit fault, the stator phase-to-phase short circuit fault, and the demagnetization fault of the permanent magnet synchronous motor; finally, an alarm is issued for the fault and the fault point is displayed.

Claims

1. A permanent magnet synchronous motor fault warning and detection method, characterized in that: The system model of the permanent magnet synchronous motor is obtained by adopting the mechanism modeling method. When the permanent magnet synchronous motor fails, the controller analyzes the sensor data and determines whether the permanent magnet synchronous motor has an internal fault or an external fault through sensor fault diagnosis based on a robust observer with full matrix rank. When it is determined to be an internal fault, the stator turn-to-turn short-circuit fault diagnosis based on fuzzy neural network, the stator phase-to-phase fault diagnosis based on model predictive control, and the demagnetization fault diagnosis based on the minimum marginal magnetic density function are used to respectively determine the stator turn-to-turn short-circuit fault, stator phase-to-phase short-circuit fault and demagnetization fault of the permanent magnet synchronous motor. Finally, the fault is alarmed and the fault point is displayed.

2. A permanent magnet synchronous motor fault early warning and detection method according to claim 1, characterized in that: Sensor fault diagnosis based on robust observer with full matrix rank establishes a model by considering system modeling error and uncertainty interference factors, and designs a robust observer to make the system reach a stable state, and uses system deviation to determine whether the system has an external sensor failure.

3. A permanent magnet synchronous motor fault early warning and detection method according to claim 2, characterized in that: The specific judgment method is: without considering factors such as modeling dynamics and external uncertainty interference, when the system fails, there is Whether the system has a sensor failure is determined as follows 4. A permanent magnet synchronous motor fault early warning and detection method according to claim 1, characterized in that: The stator turn-to-turn short-circuit fault diagnosis based on fuzzy neural network uses a multi-layer forward fuzzy neural network to predict and analyze the system model, and corrects the weights in combination with fuzzy control ideas. The difference between the system output and the expected output is fitted and analyzed, and the turn-to-turn short-circuit fault of the system is judged by analyzing the frequency and phase of the fitting function.

5. A permanent magnet synchronous motor fault early warning and detection method according to claim 4, characterized in that: By utilizing the ability of neural networks to approximate arbitrary nonlinear functions, an approximate fitting function y=a*sin(wx+φ) is adopted, where the amplitude a>0, and the frequency w and phase φ are used to separate faults, diagnose the strength of the turn-to-turn short-circuit fault, and determine the stator winding where the fault occurs.

6. A permanent magnet synchronous motor fault early warning and detection method according to claim 1, characterized in that: The stator interphase fault diagnosis based on model predictive control discretizes the system model to obtain the predicted output of the system within a given time, designs the criterion function under the mutual influence of the phases, obtains the optimal output sequence that stabilizes the system, and analyzes the weight coefficient of the criterion function to determine the phase-to-phase short-circuit fault of the permanent magnet synchronous motor.

7. A permanent magnet synchronous motor fault early warning and detection method according to claim 6, characterized in that: After discretizing the system model, a tracking differentiator TD is added to the system output and the expected input signal. The expression of TD is as follows: Wherein, x(k) represents the input signal; x1(k) represents the tracking signal of the input signal; x2(k) represents the differential signal of the input signal; r represents the tracking speed factor; h represents the filter factor; and fhan is a nonlinear function.

8. A permanent magnet synchronous motor fault early warning and detection method according to claim 6, characterized in that: The criterion function is J=||I(k+N)-I′(k+N)‖ 2 +‖λ-g(u,i a ,i b ,i c )|| 2 +||μ-h(u,i a ,i b ,i c ,i′ a ,i′ b ,i′ c )|| 2 In the formula, g(i a ,i b ,i c ) represents the current difference caused by the same bridge phase, h(i a ,i b ,i c ,i′ a ,i′ b ,i′ c ) represents the current difference caused by the different bridge phases; the criterion function is calculated with respect to the voltage u d , d q By taking the partial derivative of and setting it to zero, we can obtain the optimal control input sequence that stabilizes the system; when u=u 最优 In the criterion function, if λ=0, μ≠0, it means that an inter-phase short circuit fault occurs in different bridges; if λ≠0, μ=0, it means that an inter-phase short circuit fault occurs in the same bridge; if λ=μ=0, it means that the system operates normally.

9. A permanent magnet synchronous motor fault early warning and detection method according to claim 1, characterized in that: The demagnetization fault diagnosis based on the minimum marginal magnetic density function performs Taylor expansion analysis on the system model, uses the function of current change when the motor is running normally and faulty to analyze, designs the minimum marginal magnetic density function of the motor current, and uses the high-order power of the rotor angle to determine whether the system has a demagnetization fault.

10. A permanent magnet synchronous motor fault early warning and detection method according to claim 9, characterized in that: The minimum edge magnetic density function is e=I 故障 (θ)-I 正常 (θ) = g(θ,θ 2 ,θ 3 ,θ 4 ,θ 5 ,)+h(θ 6 )-f 正常 (θ)+ε, The minimum edge magnetic density function is obtained to obtain the value range of the minimum edge magnetic density function θ 边缘 , if θ|≤θ 边缘 Indicates that the system is running normally. If θ|>θ 边缘 Indicates that a demagnetization fault has occurred in the system.

11. A permanent magnet synchronous motor fault early warning and detection method according to claim 9, characterized in that: Using θ 边缘 Observe the change in current amplitude. If the current amplitude increases, a uniform demagnetization fault occurs. Determine whether a non-uniform demagnetization fault occurs based on whether the third or higher harmonics appear in the current.

Images