线性系统参数最小绝对误差和辨识的有效集算法

Parameter Identification by Minimizing the Sum of Absolute Value of Residuals:An Active Set Approach

本文介绍线性系统参数最小绝对误差和辨识的有效集算法，提出了采用摄动法处理退化问题的策略，还提出了中间参数和绝对误差和函数的递推计算方法。

Parameter identification for linear systems is treated as a problem of overdetermined linear equations solved in this paper. By employing an active set approach, a procedure for minimizing the sum of absolute value of residuals is presented and applied to obtain the parameter estimates of linear systems, with degeneracy of the equations based upon perturbation method. Recursive algorithms for evaluating middle parameters and absolute residuals sum function are also suggested, which greatly reduce the computational complexity. [WT5”HZ〗