Search

CN-121508396-B - Permanent magnet motor local demagnetizing fault tolerance control method and system based on parameter robust model predictive control

CN121508396BCN 121508396 BCN121508396 BCN 121508396BCN-121508396-B

Abstract

The invention provides a permanent magnet motor local demagnetizing fault tolerance control method and system based on parameter robust model predictive control, which are characterized by comprising the steps of performing super-localization processing on a permanent magnet motor model after local demagnetizing fault to obtain a super-local model; the method comprises the steps of designing a generalized proportional-integral observer, estimating and feedforward compensating lumped disturbance caused by demagnetization in real time, utilizing voltage and current data of adjacent sampling periods as real-time updating of super-local model gain, obtaining estimated values of current and disturbance from the generalized proportional-integral observer, and realizing on-line self-setting of input gain. According to the invention, super-localization is carried out on the motor model after faults, unbiased estimation is carried out on the faults introduced by the faults and the higher derivative thereof through the generalized proportional integral demagnetizing disturbance observer, and the input gain in the super-localized model can be calculated in real time by using voltage and current data of two continuous periods through the input gain self-tuning algorithm, so that the operation quality of the motor and the robust performance after the faults are improved.

Inventors

  • ZHOU ZHANQING
  • WANG JIAHAN

Assignees

  • 天津工业大学

Dates

Publication Date
20260508
Application Date
20260114

Claims (6)

  1. 1. A permanent magnet motor local demagnetizing fault tolerance control method based on parameter robust model predictive control is characterized by comprising the following steps: s1, performing super-localization processing on a permanent magnet motor model after local demagnetization failure to obtain a super-local model, wherein the method specifically comprises the following steps: S101, analyzing the empty-load air gap flux density of a prototype of the permanent magnet motor to obtain a motor flux linkage equation under a three-phase A-B-C natural coordinate system after a local demagnetizing fault; S102, taking fractional fault harmonic waves in the air gap magnetic flux density to 7/5 times, reserving 3 times and 5 times of harmonic waves with larger amplitude by integer harmonic waves, converting a motor flux linkage equation in the step S101 to a d-q coordinate system to obtain a motor model in the d-q coordinate system after local demagnetizing faults, and finishing the motor model into a state space equation; S103, performing super-localization processing on the state space equation to obtain a super-local model Wherein A state vector representing a state space equation, An input vector representing a state space equation, The method comprises the steps of representing input gain of a super local model, wherein F represents unknown disturbance of the super local model, a state vector refers to current, and an input vector refers to voltage; S2, designing a generalized proportional integral observer, and estimating and feedforward compensating lumped disturbance caused by demagnetization in real time; constructing a generalized proportional integral observer to estimate and predict current of unknown disturbance F of a super local model so as to realize model-free dead current prediction control under local demagnetization fault; ; in the formula, Wherein Is that Is used for the estimation of the (c), The estimated value of unknown disturbance F of the super local model and the estimated value of 1 times to m-1 times of differentiation of F are respectively obtained, The gain coefficient of the generalized proportional integral observer, superscript "in the formula" "Means the derivative; and S3, utilizing the voltage and current data of adjacent sampling periods as real-time updating of the super local model gain, acquiring estimated values of current and disturbance from a generalized proportional integral observer, and realizing online self-setting of the input gain.
  2. 2. The permanent magnet motor local demagnetization fault tolerance control method based on parameter robust model predictive control according to claim 1, characterized in that gain coefficients of a generalized proportional-integral observer are set according to a pole allocation strategy.
  3. 3. The permanent magnet motor local demagnetization fault tolerance control method based on parameter robust model predictive control according to claim 1, wherein step S3 specifically includes: S301, discretizing the super local model, setting disturbance F to be unchanged in two continuous control periods, and inputting gain to the super local model Performing real-time estimation; s302, considering one beat delay compensation, d-q axis reference voltage Expressed as: ; in the formula, For the reference current at the kth time , Is the state vector estimate at time k +1, Is the disturbance estimated value at the k+1 time, And Respectively obtaining from the generalized proportional-integral observer, wherein T s is the control period of the system, For inputting gain An estimated value at the kth time; S303, inputting the calculated reference voltage into space vector pulse width modulation, constructing a modulation signal through different voltage vector combinations, and driving an inverter power device according to a pre-designed sequence, so that sinusoidal current is obtained in a permanent magnet motor stator.
  4. 4. A permanent magnet motor local demagnetizing fault-tolerant control system based on parameter robust model predictive control is characterized by comprising: the super-local model module comprises a motor flux linkage equation unit, a state space equation unit and a model unit, wherein the motor flux linkage equation unit analyzes empty-load air gap flux density of a permanent magnet motor prototype to obtain a motor flux linkage equation under a three-phase A-B-C natural coordinate system after local demagnetization failure, the state space equation unit takes fractional failure harmonic waves in the air gap flux density to 7/5 times, 3 times and 5 times harmonic waves with larger integer harmonic wave retention amplitude values are obtained, the motor flux linkage equation in the step S101 is converted into a d-q coordinate system to obtain a motor model under the d-q coordinate system after the local demagnetization failure, and the motor flux linkage equation unit is used for performing super-local processing on the state space equation to obtain the super-local model Wherein A state vector representing a state space equation, An input vector representing a state space equation, The method comprises the steps of representing input gain of a super local model, wherein F represents unknown disturbance of the super local model, a state vector refers to current, and an input vector refers to voltage; The observer module is used for designing a generalized proportional-integral observer and estimating and feedforward compensating lumped disturbance caused by demagnetization in real time, and specifically comprises the steps of constructing the generalized proportional-integral observer to estimate and predict current of unknown disturbance F of a super-local model so as to realize model-free dead current prediction control under local demagnetization fault; ; in the formula, Wherein Is that Is used for the estimation of the (c), The estimated value of unknown disturbance F of the super local model and the estimated value of 1 times to m-1 times of differentiation of F are respectively obtained, The gain coefficient of the generalized proportional integral observer, superscript "in the formula" "Means the derivative; And the control module is used for acquiring estimated values of current and disturbance from the generalized proportional-integral observer by using the voltage and current data of adjacent sampling periods as real-time updating of the super local model gain, so as to realize on-line self-setting of the input gain.
  5. 5. The permanent magnet motor local demagnetization fault tolerance control system based on parameter robust model predictive control of claim 4, wherein the gain coefficient of the generalized proportional-integral observer is set according to a pole allocation strategy.
  6. 6. The permanent magnet motor local demagnetization fault tolerance control system based on parameter robust model predictive control of claim 4, wherein the control module comprises: Discretizing the super local model, setting the disturbance F to be unchanged in two continuous control periods, and inputting gain to the super local model Performing real-time estimation; taking into account one beat delay compensation, d-q axis reference voltage Expressed as: ; in the formula, For the reference current at the kth time , Is the state vector estimate at time k +1, Is the disturbance estimated value at the k+1 time, And Respectively obtaining from the generalized proportional-integral observer, wherein T s is the control period of the system, For inputting gain An estimated value at the kth time; The calculated reference voltage is input into space vector pulse width modulation, modulation signals are constructed through different voltage vector combinations, and inverter power devices are driven according to a pre-designed sequence, so that sinusoidal current is obtained in a permanent magnet motor stator.

Description

Permanent magnet motor local demagnetizing fault tolerance control method and system based on parameter robust model predictive control Technical Field The invention belongs to the field of permanent magnet motor fault-tolerant control, and particularly relates to a permanent magnet motor local demagnetizing fault-tolerant control method and system based on parameter robust model predictive control. Background After the rotor permanent magnet of the permanent magnet synchronous motor works for a long time, the magnetic performance of the rotor permanent magnet can be influenced by various factors such as working temperature, an external magnetic field, service time and the like, wherein the influence of temperature is particularly remarkable. When the working temperature exceeds the curie temperature, the permanent magnet has the phenomenon that magnetism is irreversibly disappeared, and finally, the local demagnetization fault or the uniform demagnetization fault is caused. Compared with the uniform demagnetization fault, the damage of the local demagnetization fault is more prominent, the harmonic disturbance in the flux linkage can be caused, and meanwhile, the perturbation of all parameters of the motor can be caused, so that the robustness and the control precision of the system are inevitably reduced, and even the system is possibly unstable or damaged under serious conditions. The classical demagnetization fault-tolerant control algorithm used in the prior art is mostly aimed at uniform demagnetization unfolding research, and the attention of local demagnetization faults is less. In addition, the existing algorithm cannot abandon dependence on motor parameters, so that the problem of parameter perturbation caused by local demagnetization faults cannot be effectively solved, and the robustness of the parameters is difficult to guarantee. Disclosure of Invention The invention provides a permanent magnet motor local demagnetizing fault tolerance control method and system based on parameter robust model predictive control, and the running quality of a motor after local demagnetizing fault is optimized based on a robust dead beat predictive current control strategy of a super local model. In order to achieve the above purpose, the technical scheme of the invention is realized as follows: a permanent magnet motor local demagnetizing fault tolerance control method based on parameter robust model predictive control comprises the following steps: s1, performing super-localization processing on a permanent magnet motor model after local demagnetization faults to obtain a super-local model; s2, designing a generalized proportional integral observer, and estimating and feedforward compensating lumped disturbance caused by demagnetization in real time; and S3, utilizing the voltage and current data of adjacent sampling periods as real-time updating of the super local model gain, acquiring estimated values of current and disturbance from a generalized proportional integral observer, and realizing online self-setting of the input gain. Further, the step S1 specifically includes: S101, analyzing the empty-load air gap flux density of a prototype of the permanent magnet motor to obtain a motor flux linkage equation under a three-phase A-B-C natural coordinate system after a local demagnetizing fault; S102, taking fractional fault harmonic waves in the air gap magnetic flux density to 7/5 times, reserving 3 times and 5 times of harmonic waves with larger amplitude by integer harmonic waves, converting a motor flux linkage equation in the step S101 to a d-q coordinate system to obtain a motor model in the d-q coordinate system after local demagnetizing faults, and finishing the motor model into a state space equation; S103, performing super-localization processing on the state space equation to obtain a super-local model WhereinA state vector representing a state space equation,An input vector representing a state space equation,Representing the input gain of the super local model, F represents the unknown disturbance of the super local model. Further, the step S2 specifically includes: constructing a generalized proportional integral observer to estimate F and predict current so as to realize model-free dead beat current prediction control under the local demagnetizing fault; ; in the formula, WhereinIs thatIs used for the estimation of the (c),The unknown disturbance F of the super local model and the estimated value of m times of differentiation of the unknown disturbance F are respectively obtained,Is the gain factor of the generalized proportional integral observer. Preferably, the gain coefficient of the generalized proportional-integral observer is set according to a pole allocation strategy. Further, the step S3 specifically includes: S301, discretizing the super local model, setting disturbance F to be unchanged in two continuous control periods, and inputting gain to the super local model Performing real-time