随机线性系统参数估计的偏差补偿方法

Bias-eliminated Least-squares Method for Parameter Estimation of Stochastic Linear System

将偏差补偿最小二乘法（ＢＥＬＳ）推广到一般单输入单输出系统的辨识。结果表明，这种推广的偏差补偿最小二乘法可以在有色噪声下获得系统参数的无偏估计而不需对噪声建模。仿真例子验证了理论分析的正确性。

Based on the studies of [6] and [7], this paper extends the proposed Bias-eliminated Least squares method (BELS) to parameter estimation of general stochastic linear systems in the presence of colored noise. It is shown that the extended BELS method can obtain asymptotically unbiased estimates for the system parameters even if the noise model is not exactly known. The theoretical analysis is verified by simulation examples.