机构地区: 黑龙江大学电子工程学院
出 处: 《黑龙江大学自然科学学报》 2003年第2期55-58,共4页
摘 要: 提出了多变量滑动平均(MA)模型参数估计的两段最小二乘法。第一段将多变量MA模型用高阶多变量自回归(AR)模型近似代替,用多变量递推最小二乘法(MRLS)估计高阶AR模型参数。第二段用最小二乘法解不相容矩阵代数方程组得MA参数估值。同多变量递推增广最小二乘法相比,可提高精度,仿真例子说明了其有效性。 Two-stage least squares method of parameter estimation for multivariable moving average (MA) model is presented. In the first stage, the multivariable MA model is replaced with the high order Multivariable autoregressive (AR) model, whose parameters are estimated by the recursive least squares (RLS) method. The second stage, the MA parameters are obtained by solving the inconsistent matrix algebraic equations by the least squares method. Compared to the multivariable recursive extended least squares (RELS) method, its accuracy is higher. A simulation example demonstrates its effectiveness.