New Online Parameter Identification And Self-tuning Strategy For Deadbeat Predictive Control Of PMSM

Permanent magnet synchronous motors and their drive systems are the main solutions for electric drives in new energy vehicles and many industrial fields.Solving the non-linear,time-varying and coupling problems of permanent magnet synchronous motors is the focus of the research on motor control methods.As the core link of permanent magnet synchronous motor control,the current inner loop is generally implemented by proportional-integral control,but due to the lower control bandwidth,the performance is difficult to further improve,and it is expected to be predicted by the deadbeat with excellent dynamic performance in the future Control replaces.But deadbeat predictive control based on motor model is inherently sensitive to model parameter deviations.Motor model parameter deviations and control system nonlinearities will cause transient overshoot or undershoot and steady-state errors.For solving the problem of parameter deviation,the existing researches are mainly divided into two categories.One is to add an integrator to form a closed-loop structure to eliminate the steady-state error,and the other is to directly obtain the model parameters through the parameter identification observer to modify the model.The high dynamic performance of deadbeat predictive control is far better than proportional-integral,but the combination of integrator or traditional parameter identification observer and deadbeat predictive control takes a long time to eliminate steady-state errors,so that deadbeat predictive control The original high dynamic performance advantage has been offset.This paper focuses on solving the problem of parameter dependence on deadbeat predictive control of permanent magnet synchronous motors,and proposes a new online parameter direct identification method and self-tuning strategy.The convergence speed of the new method is much higher than that of the traditional identification method 1 to 2Orders of magnitude.Compared with the existing research,the combination of the new method and deadbeat predictive control not only fully retains the advantages of high dynamic performance of deadbeat predictive control,but also solves the parameter sensitivity problem of deadbeat predictive control and eliminates the steady state.Errors improve the dynamic performance of the system.The second chapter of this article introduces the continuous-time model and discrete mathematical model of permanent magnet synchronous motors.For the first time,the unit cycle calculation delay and deadbeat predictive control command update sequence of nonlinear systems are analyzed through intuitive methods such as schematic diagrams.At the end of the second chapter,we intuitively analyze the parameter sensitivity characteristics of traditional deadbeat predictive control from the perspective of time domain and frequency domain.The third chapter of this paper innovatively proposes a new online parameter identification method.The method is mainly divided into two steps.The first step: do not introduce any additional sensors,only use three-phase sampling current,the voltage command and speed information of the deadbeat predictive control to calculate the equivalent deviation value covering all the motor parameter deviations.In the second step,based on the motor model and the equivalent deviation value calculated in the first step,the deviation size of each motor parameter can be obtained directly by one-step analytical solution.Because the proposed algorithm only uses the information of one operating point to directly calculate the parameters,the convergence speed is extremely fast;at the same time,because the proposed method does not include any integration and derivation process,the calculation process is simple,the controller computer resources are few,and it is easy to implement.High engineering application value.The simulation results show that the proposed identification method can complete the convergence within 0.002 s,and the convergence error range is 1.2%(100% larger inductance,300% larger resistance)～7.9%(25% smaller inductance,75% smaller resistance).In contrast,the fastest convergence of the traditional integrator or parameter identification observer also requires0.035s(resistance identification)and 0.02s(inductance identification),and the parameter identification speed is far lower than the proposed algorithm.Chapter 4 proposes a simple and feasible controller parameter self-tuning strategy,using the parameter identification results of Chapter 3 to re-tune the controller parameters,so that the model parameters of the deadbeat predictive controller always match the real parameters of the motor.In this way,the current steady-state error is eliminated,and the dynamic performance and stability of the system are improved.The simulation results show that the proposed parameter auto-tuning strategy can successfully auto-tune the model parameters to the true value of the motor,make the equivalent deviation value representing the motor parameter deviation tend to 0,and successfully correct the current steady-state error.In Chapter 5 of this article,a 2k W,4k RPM permanent magnet synchronous motor pairing platform system is built.The system includes the host computer control software based on Lab VIEW and the motor controller based on the DSP28335 chip.Experiments are carried out on the proposed new online parameter direct identification method and self-tuning strategy.Parameter identification transient experiments show that the proposed parameter identification method can also converge in about 0.01 s with the addition of a low-pass filter,which is an order of magnitude better than the same type of parameter identification algorithm.The parameter identification steady-state experiment shows that the error identification result of resistance and inductance error is less than5%.The overall parameter identification and self-tuning strategy experimental results show that under the initial condition of 50% parameter deviation,the controller model parameters can be updated online and automatically with more accurate parameters,the steady-state current error is quickly corrected,and the parameters sensitive of deadbeat predictive control is solved.