最小方差控制中的参数递推辨识

Parameter Recursive Identification for Minimum Variance Control

从系统辨识的角度研究最小方差控制中的参数递推辨识问题。对于最小方差闭环控制中的ARMAX模型未知参数矢量,采用多新息递推最小二乘辨识和分离迭代的递推最小二乘辨识法在线辨识和估计ARMAX模型中的未知参数矢量。在白噪声干扰下,两种辨识方法都能得到未知参数矢量的无偏估计;而在有色噪声干扰下,仅分离迭代的递推最小二乘辨识法才能给出参数的无偏估计值。最后用仿真算例验证了方法的有效性和可行性。

The problem of parameter recursive identification for minimum variance control was studied from the point of system identification. For the unknown parameter vector of the ARMAX model in the minimum variance closed loop control, we proposed a multi-innovation recursive least-squares identification method and a separable iterative recursive least-squares identification method to identify and estimate the unknown parameters vector in the ARMAX model on line. When excited by the white noise, both the methods could give the unbiased estimation about the unknown parameter vector. When excited by the colored noise, only the separable iterative recursive least-squares identification method could give the unbiased estimation. Finally, the effectiveness and feasibility of the proposed strategy was verified by the simulation results.