基于在线差分进化算法的平面开关磁阻电机约束预测位置控制  

Constrained Predictive Position Control of Planar Switched Reluctance Motors UsingOn-line Differential Evolution Algorithm

作　　者：黄苏丹[1] 刘付榕 曹广忠[1] 陈龙 李玲珑 敬刚[2] 刘岩[2] HUANG Sudan;LIU Furong;CAO Guangzhong;CHEN Long;LI Linglong;JING Gang;LIU Yan(Guangdong Key Laboratory of Electromagnetic Control and Intelligent Robot(College of Mechatronics and Control Engineering,Shenzhen University),Shenzhen 518060,Guangdong Province,China;Guangdong Key Laboratory of Optomechatronics(Research Institute of Tsinghua University in Shenzhen),Shenzhen 518060,Guangdong Province,China)

机构地区：[1]广东省电磁控制与智能机器人重点实验室(深圳大学机电与控制工程学院),广东省深圳市518060 [2]广东省光机电一体化重点实验室(深圳清华大学研究院),广东省深圳市518060

出　　处：《中国电机工程学报》2023年第13期5239-5250,共12页Proceedings of the CSEE

基　　金：国家自然科学基金项目(51907128,52277061);广东省自然科学基金项目(2021A1515011704)。

摘　　要：模型预测控制方法在解决平面开关磁阻电机非线性、不确定干扰、约束等问题具有突出的优势,有望实现平面开关磁阻电机的高精度约束最优控制,目前平面开关磁阻电机约束预测控制亟需解决在线计算量大、内存大的问题。为快速、精确地求解平面开关磁阻电机约束预测控制在线优化问题,并获得小内存控制器,提出基于在线差分进化算法的约束预测位置控制方法。基于电机的多步预测位置模型,建立考虑输入约束的电机预测位置控制器,采用设计的差分进化算法在线求解最优控制量,构建具有摩擦补偿的电机约束预测位置控制系统。为提高差分进化算法的收敛速度并获得更优解,采用缩小种群范围、生成分区种群个体以及优化初始种群的操作产生初始种群,利用优化的变异因子和交叉因子对种群个体进行变异和交叉操作,将预测控制的代价函数作为适应度函数对种群个体进行选择操作。仿真和实验结果验证了所提方法的有效性。Model predictive control method shows outstanding advantages in solving problems such as nonlinearity,uncertain interference,and constraint of planar switched reluctance motors(PSRMs),and this method is promising to achieve high-precision constrained optimal control of PSRMs.Currently,the constrained predictive control of PSRMs urgently needs to solve the problem of heavy computation and large memory.In order to solve the online optimization problem of the constraint predictive control of PSRMs quickly and accurately and to obtain a small-memory controller,the constrained predictive position control of PSRMs using on-line differential evolution(DE)algorithm is proposed in this paper.A predictive position controller considering the input constraint is established based on the multi-step prediction position model of the PSRM.A differential evolution algorithm is designed to solve the online optimal control law of the controller.A constraint predictive position control system of the PSRM with friction compensation is developed.To improve the convergence speed of the DE algorithm and to obtain a more optimal solution,the operations of reducing the population range,generating sectionalized population individuals,and optimizing the initial population are applied to generate the initial population;the optimized mutation and crossover factors are employed to mutate and crossover the population individuals,respectively;the cost function of the predictive control is usedas the fitness function to select population individuals. Theeffectiveness of the proposed method is verified via thesimulation and experimental results.

关 键 词：差分进化算法 模型预测控制 平面电机 位置跟踪控制 

分 类 号：TM352[电气工程—电机]