基于复数域的自适应递推子空间辨识算法

Adaptive Recursive Subspace Identification Algorithm in Complex Domain

针对工业系统广泛存在的特征值分布于复数域的特性,提出一种基于复数域的自适应递推子空间辨识算法。该算法首先设计了变遗忘因子机制下Hankel矩阵的无偏更新形式;其次利用正交化子空间跟踪算法实现广义能观测矩阵的递推更新;最后,针对工业系统参数矩阵特征值普遍分布于复数域的特点,利用特征值空间的欧氏距离信息实现变遗忘因子的自适应更新。数值仿真结果表明,该自适应算法对时变参数的跟踪速度快、跟踪精度高。

An adaptive recursive subspace identification algorithm in complex domain was proposed for recursive estimation of state space model of linear time-varying systems.The unbiased Hankel matrices were updating with variable forgetting factor.Then recursive estimation algorithm of extended observation matrix was realized with the help of orthogonal PAST algorithms.Finally,the adaptive updating form of variable forgetting factor was designed with eigenvalues＇ Euclidean-distance information of industrial system parameter matrices.Two simulation results show tracking performance of new algorithm is faster and better than traditional one.