Search

CN-122001254-A - Double-subspace virtual vector-based model-free predictive repetitive control method and system for double-three-phase permanent magnet synchronous motor

CN122001254ACN 122001254 ACN122001254 ACN 122001254ACN-122001254-A

Abstract

The invention relates to a model-free predictive repetitive control method and a model-free predictive repetitive control system for a double-three-phase permanent magnet synchronous motor based on double subspace virtual vectors, wherein vectors of the motor are decomposed into mutually orthogonal fundamental wave subspaces and harmonic subspaces based on acquired operation parameters, and independent super-local models are built in2 subspaces; the method comprises the steps of constructing a linear expansion state observer in a fundamental wave subspace, constructing a repeated expansion state observer based on repeated control in a harmonic subspace, respectively synthesizing an independent virtual voltage vector set and a decoupling virtual voltage vector set in2 subspaces, calculating reference voltage vectors of the 2 subspaces, selecting an optimal virtual voltage vector and an optimal decoupling virtual voltage vector, respectively calculating an optimal duty ratio, synthesizing control signals applied to each bridge arm of an inverter, and driving the double three-phase permanent magnet synchronous motor. The invention realizes model-free control of the double subspaces, controls the model parameters of the algorithm motor system, and shows better robustness when the parameters are mismatched.

Inventors

  • YANG GUANGHUI
  • SHEN YONG

Assignees

  • 浙江工业大学

Dates

Publication Date
20260508
Application Date
20260410

Claims (10)

  1. 1. A model-free predictive repetitive control method for a double three-phase permanent magnet synchronous motor based on double subspace virtual vectors is characterized in that the method is based on acquired operation parameters of the double three-phase permanent magnet synchronous motor, vectors of the motor are decomposed into mutually orthogonal fundamental wave subspaces and harmonic subspaces through space vector decoupling transformation, independent super local models are built in the fundamental wave subspaces and the harmonic subspaces respectively, linear expansion state observers are built in the fundamental wave subspaces and used for observing direct current disturbance in the fundamental wave subspaces, repetitive expansion state observers based on repetitive control are built in the harmonic subspaces and used for observing periodic disturbance in the harmonic subspaces, and independent virtual voltage vector sets and decoupling virtual voltage vector sets are respectively synthesized in the fundamental wave subspaces and the harmonic subspaces according to output voltage vectors corresponding to the switching states of an inverter; respectively calculating reference voltage vectors of a fundamental wave subspace and a harmonic wave subspace, respectively determining the sector where the optimal virtual voltage vector and the optimal decoupling virtual voltage vector are located according to the phase angles of the reference voltage vectors, and accordingly selecting the optimal virtual voltage vector and the optimal decoupling virtual voltage vector; Based on a preset cost function, calculating an optimal duty ratio for an optimal virtual voltage vector and an optimal decoupling virtual voltage vector selected for the fundamental subspace and the harmonic subspace respectively; and synthesizing control signals finally applied to each bridge arm of the inverter according to the optimal vector and the corresponding optimal duty ratio, and driving the double three-phase permanent magnet synchronous motor.
  2. 2. The model-free predictive repeat control method of a double three-phase permanent magnet synchronous motor based on double subspace virtual vectors of claim 1, wherein the space vector decoupling transformation is to perform VSD transformation on motor vectors to obtain corresponding fundamental subspace variables and harmonic subspace variables, and then the fundamental subspace variables and the harmonic subspace variables are subjected to rotary transformation and inverse transformation respectively to obtain motor vectors under corresponding rotary coordinate systems.
  3. 3. The model-free predictive repeat control method of the double three-phase permanent magnet synchronous motor based on the double subspace virtual vector of claim 1 is characterized in that a super local model of a fundamental wave subspace is associated with current, voltage and lumped disturbance items, and the control gain of the fundamental wave voltage is determined based on inductance parameters of the fundamental wave subspace; The super local model of the harmonic subspace is associated with the current, the voltage and the periodic lumped disturbance term, and the control gain of the harmonic voltage is determined based on the leakage inductance parameter of the harmonic subspace.
  4. 4. The model-free predictive repeat control method of a double three-phase permanent magnet synchronous motor based on double subspace virtual vectors of claim 1 is characterized in that a repeat expansion state observer is combined with a repeat controller to realize periodic disturbance estimation, a transfer function of the repeat control is associated with a pure time delay term and a period for suppressing a preset angular frequency harmonic component, and the preset angular frequency harmonic component is an integer multiple of fundamental current angular frequency.
  5. 5. The model-free predictive repeat control method for the double three-phase permanent magnet synchronous motor based on the double subspace virtual vectors is characterized in that a constructed virtual voltage vector set is obtained by synthesizing a first class vector and a second class vector in a base wave subspace, the first class vector and the second class vector are the first two classes with magnitudes ordered from large to small, a constructed decoupling virtual voltage vector set is obtained by synthesizing a third class vector and a fourth class vector in a harmonic subspace, and the third class vector and the fourth class vector are the last two classes with magnitudes ordered from large to small.
  6. 6. The model-free predictive repetitive control method for a double three-phase permanent magnet synchronous motor based on double subspace virtual vectors of claim 1 is characterized in that reference voltage vectors of a fundamental subspace and a harmonic subspace are calculated based on an observation value of a super local model, a corresponding linear extended state observer or a repetitive extended state observer and a reference current.
  7. 7. The model-free predictive repetitive control method of the double three-phase permanent magnet synchronous motor based on the double subspace virtual vector of claim 6, wherein one-step delay compensation is introduced when calculating a reference voltage vector, and the reference current value is calculated based on future time; And when determining the sector of the associated area where the optimal vector is located, judging according to the phase angle of the reference voltage vector.
  8. 8. The model-free predictive repeat control method of the double three-phase permanent magnet synchronous motor based on the double subspace virtual vectors of claim 1 is characterized in that the cost functions of a fundamental wave subspace and a harmonic subspace are respectively constructed based on the difference between the reference voltage vector of the corresponding subspace and the voltage value corresponding to the selected optimal vector, and the optimal duty ratio is obtained by minimizing the cost function of the corresponding subspace.
  9. 9. The model-free predictive repetitive control method for the double three-phase permanent magnet synchronous motor based on the double subspace virtual vectors is characterized in that when control signals are synthesized, optimal vectors selected by a fundamental subspace and a harmonic subspace are overlapped on corresponding duty ratios on each phase of bridge arm to obtain a total duty ratio, and PWM waves are generated according to the total duty ratio.
  10. 10. The model-free predictive repetitive control system for the double three-phase permanent magnet synchronous motor based on the double subspace virtual vector is characterized by comprising the following components: The vector transformation module is used for decomposing the vector of the motor into mutually orthogonal fundamental wave subspaces and harmonic wave subspaces through space vector decoupling transformation based on the acquired operation parameters of the double three-phase permanent magnet synchronous motor; the super local model construction module is used for constructing independent super local models in the fundamental wave subspace and the harmonic subspace respectively; The disturbance observation module comprises a linear expansion state observer constructed in a fundamental wave subspace and a repeated expansion state observer constructed in a harmonic wave subspace and based on repeated control, and is used for observing direct current disturbance and periodic disturbance in a corresponding subspace respectively; The virtual voltage vector construction module is used for respectively synthesizing an independent virtual voltage vector set and a decoupling virtual voltage vector set in the fundamental wave subspace and the harmonic subspace according to the output voltage vector corresponding to the switching state of the inverter; the reference voltage calculation and vector selection module is used for respectively calculating reference voltage vectors in the fundamental wave subspace and the harmonic wave subspace, respectively determining the sector where the optimal virtual voltage vector and the optimal decoupling virtual voltage vector are located according to the phase angle of the reference voltage vector, and accordingly selecting the optimal virtual voltage vector and the optimal decoupling virtual voltage vector; the duty ratio calculation module is used for calculating the optimal duty ratio for the optimal virtual voltage vector and the optimal decoupling virtual voltage vector selected by the fundamental wave subspace and the harmonic wave subspace respectively based on a preset cost function; And the control signal synthesis module is used for synthesizing control signals finally applied to all bridge arms of the inverter according to the optimal vector and the corresponding optimal duty ratio so as to drive the double three-phase permanent magnet synchronous motor.

Description

Double-subspace virtual vector-based model-free predictive repetitive control method and system for double-three-phase permanent magnet synchronous motor Technical Field The invention relates to the technical field of control or regulation of motors, generators or electromechanical converters, and control of transformers, reactors or chokes, in particular to a model-free predictive repetitive control method and system for a double three-phase permanent magnet synchronous motor based on double subspace virtual vectors. Background In the fields of high power and high reliability of ship propulsion, aerospace, electric automobiles and the like, the double three-phase permanent magnet synchronous motor is particularly favored, and has the advantages of reducing voltage and current stress born by power electronic devices, reducing torque pulsation, improving control flexibility and having inherent fault tolerance. For a double three-phase permanent magnet synchronous motor driving system, the prediction control of a limited control set model is becoming a competitive alternative to the traditional magnetic field directional control and direct torque control, which mainly benefits from the visual structure, rapid dynamic response and strong capability in the aspect of processing multi-objective optimization problems. However, the motor-driven predictive model relies heavily on parameters such as resistance, inductance and flux linkage, which are affected by temperature, saturation of the magnetic circuit, motor asymmetry, etc., so that the control performance is greatly reduced by the parameter mismatch and unmodeled dynamics such as inverter nonlinearity. One solution is to estimate the lumped disturbance and compensate when calculating the reference voltage using disturbance observers, however such observers usually assume that the lumped disturbance is a slow time varying disturbance but the tracking ability is limited in the x-y subspace due to the existence of rich periodic disturbances, and another solution is to update the model parameters by online parameter identification, including recursive least squares, etc., but the parameter identification method is susceptible to noise and disturbance, resulting in a reduced identification accuracy. In order to solve the above-mentioned problems, model-free predictive current control has received a great deal of attention in recent years as an advanced control strategy that does not depend entirely on accurate motor parameters. The existing model-free prediction methods are roughly divided into two types, namely a method based on a super local model and a strategy for replacing a physical motor model by adopting a data driving model. The data driving method relies on input and output data to fit model parameters, but the method needs to perform prediction evaluation on a plurality of virtual vectors and parameter iteration in each control period besides real-time calculation on a large amount of data, so that the calculation load is obviously increased, and meanwhile, the method is also susceptible to noise to cause the reduction of prediction accuracy. The method based on the super local model generally adopts an extended state observer in an alpha-beta subspace, estimates total disturbance and current through the observer to calculate reference voltage, and adopts virtual voltage vectors to form a candidate set for modulation in the alpha-beta subspace, however, the method only indirectly controls through the virtual vectors in an x-y subspace, and lacks independent current reference values and error feedback to realize closed loop control, and at the moment, low-order x-y current harmonic waves caused by motor asymmetry, inverter dead zone effects and counter electromotive force harmonic waves cannot be effectively compensated, so that the copper loss of a rotor is increased. Disclosure of Invention The invention solves the problems existing in the prior art and provides a model-free predictive repetitive control method and system for a double three-phase permanent magnet synchronous motor based on double subspace virtual vectors. The invention adopts the technical scheme that the model-free predictive repetitive control method of the double three-phase permanent magnet synchronous motor based on double subspace virtual vectors is characterized in that based on the acquired operation parameters of the double three-phase permanent magnet synchronous motor, vectors of the motor are decomposed into mutually orthogonal fundamental wave subspaces and harmonic subspaces through space vector decoupling transformation, independent super local models are respectively built in the fundamental wave subspaces and the harmonic subspaces, linear expansion state observers are built in the fundamental wave subspaces and used for observing direct current disturbance in the fundamental wave subspaces, repetitive expansion state observers based on repetitive control are built in