| Since the model predictive control was proposed,it has attracted attention in the industry due to unique advantages in dealing with multiconstraint and multi-objective problems.For the control of Permanent Magnet Synchronous Motor(PMSM),especially in high-power applications,a common method to bring down the inverter switching loss is limiting the switching frequency,which causes current harmonic distortion and brings other problems such as torque ripple.We expect to take use of the advantages of MPC to optimize the current distortion while reducing switching losses.This paper combines the idea of MPC and explores the high-performance control strategy of PMSM.Based on traditional vector control,we replace the PI regulator in current loop with model predictive controller and proposes a model predictive direct current control(MPDCC),which takes the maximum current limit and the number of switching state actions into account.The simulation proves that the method can achieve better response ability and minimize current distortion rate while reducing inverter switching frequency.In the study of the model-based direct current control,considering the fact that single-step predictive control is easy to fall into a local optimum and the outcome is not ideal when dealing with multi-objective problems,we use a multi-step predictive direct current control strategy.Taking the switching loss of inverter as the optimization objective,the optimization evaluation function is established.According to the two main methods of multi-step prediction,including step-by-step calculation and hysteresis control,two multi-step predictive current control methods are designed to minimize the switching loss of the inverter.The characteristics and applicable conditions of the two methods are compared and discussed through simulation.In contrast,the step-by-step prediction method is more suitable for the control of permanent magnet motor.However,the step-by-step prediction method has a large amount of calculation.In order to reduce the complexity of calculation in the multistep online prediction process,we use a heuristic voltage sequence search method based on pre-inspired ant colony algorithm.The switching state of the inverter at continuous moments is regarded as the ant colony motion trajectory.The function leaves stronger pheromone on the better path as positive feedback to reduce the amount of calculation and speed up the convergence speed.However,the solution calculation under heuristic algorithm is still very complicated.In this regard,we deduce and simplify the evaluation function with mathematical methods and transforms multi-step current prediction into a discrete combinational optimization problem.and proposes a depthfirst branch-and-bound search method.In the search process,the global and local delimitation constraints are considered to discard the voltage sequence that cannot become the optimal solution in time.Also,the adjacent vector priority search order is used to speed up the convergence.Simulations prove that the method can improve the dynamic and static characteristics of PMSM at low switching frequency and minimize computational complexity.The significance of extending the prediction step in improving the dynamic and steady state performance and parameter robustness of the system is also discussed in the paper.Finally,2-steps MPDCC under this method is verified and compared in experiments.The results prove the effectiveness of the method and the improvement of the steady-state performance of the motor under low switching loss by extending the prediction step. |
