基于改进PSO算法的GMA迟滞模型参数辨识

Parameter Identification of Hysteretsis Model for GMA Based on Improved PSO Algorithm

使用改进的粒子群优化算法辨识Jiles-Atherton模型参数。针对J-A模型对超磁致伸缩致动器(giant magnetostrictive actuator,GMA)迟滞特性建模中磁化参数互相嵌套难以辨识的特点,改进磁滞模型并建立了考虑超磁致伸缩材料磁机耦合特性的动态磁滞模型;为了克服普通粒子群算法实际求模型参数时计算量大,运行时间长的缺点,提出基于粒子群算法和遗传算法的改进算法--带交叉因子的粒子群优化算法,将模型仿真所求的磁化强度和实验测得的磁化强度的差值的平方作为适应度函数,并结合最小二乘法思想对J-A模型的几个参数进行辨识;最后,在Matlab 7.0上进行仿真,给出了模型辨识后的结果。在不同预压力和驱动频率下的仿真结果与GMA已有实验数据进行对比,验证得出辨识后的模型可较好地与实验数据拟合,磁致伸缩位移误差在5%以内。

Using the improved particle swarm optimization algorithm to identify parameters of Jiles-Atherton model. Aiming at the problem that parameters of the hysteresis nonlinear model are mutual nested and difficult to identify, an improved hysteresis model is proposed by established a dynamic coupling hysteresis model; General particle swarm optimization （PSO） model is computation-intensive and time-consuming when calculating model parameters, so art improved algorithm with crossover factor of particle swarm optimization is put forward, which based on the particle swarm optimization algorithm and the genetic algorithm; The square of the magnetization differ- ence between model simulation and experimental results provides the fitness function, and the parameters of J-A model are identified with this improved algorithm, combined with least square method. Finally, the simulation is carried out with Matlab 7. 0 and the result of parameter identification is presented. The computer simulation result under different pre - pressure and driving frequency are com- pared with GMA experimental data. It verifies that the model can be well fitting the experimental data, and the error in hysteresis strain curve is controlled within 5 %.