针对Hammerstein模型的典型系统辨识方法

Typical system identification methods for Hammerstein model

针对目前Hammerstein模型辨识中存在的成果多对比分析少的问题,文中研究了递推最小二乘、粒子群以及混合蛙跳方法对Hammerstein模型的辨识问题.在理论分析的基础上,通过实验研究了3种方法在辨识误差、收敛速率等方面的性能差异,最后给出了3种方法存在的优势以及潜在的不足,并深入分析其原因.文中结果可为3种典型方法的实际工程应用提供指导,并可为方法未来改进提供思路和指导.

For the problem that the achievements of Hammerstein model identification is plenty but the comparative analysis is lack,the identification methods of Recursive Least Squares,Particle Swarm Optimizer and Shuffled Frog Leaping Algorithm are studied. First,the Hammerstein model and the three methods are introduced and analyzed briefly. Then,based on the theory analysis,the system identification error and convergence rate are studied through experiments. Finally,the advantages and inferiority are pointed. The results of this paper can provide guidance for practical engineering of the three typical methods,and also provide ideas for the improvement of the corresponding methods.