递推子空间模型辨识算法的研究与仿真

为了很好的解决在线辨识系统模型问题，在对子空间模型辨识研究的基础上，结合递推最小二乘算法和子空问状态辨识方法。推导了子空间状态辨识的递推算法。该算法不仅解决了在线辨识问题，而且算法简单，计算方便，很好地克服了在线辨识时子空间矩阵维数的变化问题。经仿真研究表明，该递推算法克服了一次完成算法在大批量数据运算时，耗时大，专用内存多的缺点，而且对于测量和过程均有噪声干扰的多输入多输出系统，有很好的辨识效果，有较为广阔的应用前景。

Study on Recursive Subspace Model Identification

To realize the identification of system model on-line,based on the research of the subspace model identification method,and combining the recursive least square method with subspace model identification method,a recursive algorithm of subspace system identification(RSIM) is proposed.This algorithm not only resolves the identification on-line,but also calculates expediently and easily.And it overcomes the problem of dimension change when identified on-line.The result shows that the recursive subspace model identification algorithm overcomes the shortage of the subspace model identification,and gets the exact identification for the MIMO system which has progress and measurement noises.