CSME系统的EIV-RLS辨识建模法及其精度分析

CSME System EIV-RLS Identification Modeling Method and Its Accuracy Analysis

非接触式同步电机励磁(CSME)系统因其补偿网络与拓扑电路的多样性,对建模方法的通用性与可移植性要求很高。针对传统建模方法严重依赖明确的电路拓扑和系统工作状态的缺点,提出初值为估计值的递推最小二乘(EIV-RLS)辨识建模法。采样CSME系统的输入、输出数据后,先进行数据变换,利用赤池信息准则(AIC)判断系统阶次,通过EIV-RLS算法得到系统参数估计值,建立系统的小信号模型。通过Matlab/Simulink软件,以串-串(S-S)型半桥-全波CSME系统为建模对象,分别对最小二乘(LS)、初值为零的递推最小二乘(ZIV-RLS)及EIV-RLS辨识法进行建模和仿真验证,对比分析了三种方法的精确度,并验证了EIV-RLS辨识建模法的通用性。最后,通过对CSME系统样机的实验测试,验证了EIV-RLS辨识建模法的有效性和精确性。

Contactless synchronous motor excitation （ CSME） system has diversiform compensation networks and topological circuits, so the modeling method requires a high degree of versatility and portability. In view of the problem that traditional modeling method relies heavily on clear topology circuit and specific working state, this paper proposes an estimate initial value recursive-least-squares （ EIV-RLS ） identification modeling method. After sampling and transforming the input and output data, judged system order by akaike information criterion（ AIC）. Then obtained system parameter estimation by EIV-RLS algorithm and established system small signal model. Through Matlab/Simulink respectively simulated least-squares （ LS） , zero initial values recursive- least-squares （ ZIV-RLS） and EIV-RLS identification modeling for series-series（ S-S） type CSME system. Compared and analyzed the accuracy of three methods and verified the generality of EIV-RLS identification modeling method. Finally, through the test of CSME system prototype, verifies the EIV-RLS identification modeling method is effectiveness and accuracy.