有噪输入输出动态系统辨识方法

Identification of Dynamic Noisy Input-Output System

研究了基于带噪观测数据的动态系统的辨识问题.针对输入输出观测数据均含有加性噪声的情况,提出了一种改进的偏差补偿最小二乘算法.该算法引入一个后向输出预测算子,通过考察最小二乘误差的自相关函数及最小二乘误差与后向输出预测误差的互相关函数的性质,得到渐进偏差的估计,并利用偏差补偿原理,得到输入输出带噪系统参数的一致估计.仿真结果验证了该方法的有效性.

Identification of linear dynamic systems from noisy input and output measurements is studied.In order to deal with this task,a modified bias compensation least square（BCLS） method is proposed.In the proposed method,the backward output predictor（BOP） is introduced.With the help of analyzing the properties of auto-function of least square（LS） error and cross-function of the BOP error and the LS error,the estimate of the asymptotic bias is obtained.Then based on the principle of bias compensation,the consistent parameter estimate of the noisy input-output system can be obtained.Simulation results are given to illustrate the effectiveness of the proposed algorithm.