Parameter Identification Of Permanent Magnet Synchronous Motor Based On Improved Particle Swarm Optimization Algorithm

Permanent magnet synchronous motor(PMSM)has the advantages of simple structure and high power factor,and is widely used in industrial fields.The design of the PMSM high-performance controller is related to the motor parameters,but the motor parameters change with the temperature and working conditions,resulting in degradation of the control performance.To address these problems,three online parameter identification strategies for PMSM with improved particle swarm optimization(PSO)were proposed to solve the problem of PSO poor identification accuracy and nonlinear compensation of voltage source inverter(VSI),and the simulation and experimental verification were carried out.The main research contents of this thesis were as follows:(1)A fast backfire double annealing particle swarm optimization(FBDAPSO)strategy for PMSM parameter identification was proposed to address the problem of low accuracy in conventional PSO identification of PMSM parameters.The advantage of better robustness of the simulated annealing(SA)was introduced,and a double annealing process was used:one was a conventional rapid annealing to accelerate the convergence of algorithm,and the other was a tempering annealing after receiving the differential solution.The simulation results showed that the proposed method converged faster than PSO and simulated annealing particle swarm optimization(SAPSO),and maintained the better identification accuracy and robustness of SAPSO.(2)An improved self-optimizing SAPSO algorithm(SOSAPSO)for PMSM parameter identification method was proposed to solve the problem that the identification equation is under-rank and fixed value of the FBDAPSO temper.The full-rank identification model was established by time-sharing injection of _di=0 and negative-order weak magnetic current,and a dynamic opposition-based learning(DOBL)strategy was introduced for the inertia weight after simplifying the PSO velocity term.To avoid the prematurity of the algorithm,a hybrid adaptive mutation strategy of density and similarity was adopted for the PSO extremes,and a greedy optimization memory tempering annealing algorithm was introduced to solve the problem of difficult selection of tempering quantities.The simulation results showed the proposed method avoided the risk of falling into local optimization,the ability to track the changed parameters and the convergence speed were better,and the robustness of the algorithm was improved.(3)A PMSM parameter identification method based on memetic quantum annealing particle swarm optimization(MQAPSO)considering distortion voltage was proposed to overcom the influence of VSI nonlinear factors on identification and the blindness of SOSAPSO optimization.The distorted voltage as a motor parameter for simultaneous identification,the influence of the distortion voltage on the identification model can be compensated in real time;the chaos theory was introduced to enhance the ability of quantum particle swarm optimization(QPSO)to explore potential better regions,the open self-learning(OSL)and fusion mutation strategy were proposed to avoid problem of the blindness of the global DOBL exploration blindness and QPSO premature maturity.A local fine search optimization operation was carried out to solve the problem of weak search in the later stage of QPSO.The simulation results shown that the proposed method had better dynamic parameter tracking ability and stronger robustness under different working conditions,and after the compensation of the distortion voltage identified in real time,the accuracy of the identification model was higher.(4)The RT-LAB experimental platform was established to verify the correctness of three proposed methods.The experimental results showed MQAPSO can compensate the distortion voltage in real time,solved the problems of poor accuracy of QPSO and high complexity of global DOBL,and the identification speed and accuracy were better than those of FBDAPSO and SOSAPSO.