永磁同步电机双率残差增广最小二乘参数辨识

Residual based extended least squares identification method of permanent magnet synchronous motor dual-rate sampling data system

针对永磁同步电机系统输入输出的本质多采样率特性,结合多项式变化技术,构建模型转换矩阵,推导出基于电流和电压的永磁同步电机双率采样数学模型,获得其扩展参数向量,并构建其回归模型,针对输入输出量中不可测干扰量采用残差估计,进而建立永磁同步电机双率残差增广最小二乘算法(DR-RELS),对其算法收敛性进行分析。仿真实验结果表明,DR-RELS算法对变换后的永磁同步电机双率采样数据模型参数估计一致收敛,同时噪声方差的大小影响该算法收敛效果。

Permanent magnet synchronous motors（ PMSM） control system has multi-rate sampling charac-teristic in itself on the system input and output. The model transformation matrix was proposed based on polynomial transformation technique and the dual-rate mathematical model of PMSM was induced under the current and voltage, and the extended parameters vector and regressive model of dual-rate sampled system was acquired. The residual was used to estimate the unmeasurable noise, and then the residual based extended least squares （ DR-RELS） was proposed and used to the parameters identification, and the convergence properties of the algorithm was analyzed. Simulation results show that DR-RELS is uni-formly convergence for the parameters of PMSM dual-rate sampling system, and meanwhile noise variance affects the convergence performance of this proposed algorithm.