摘要
针对单一灰色预测方法下磁特性曲线建模对电机不同运行状态区分能力差、概括性不强,由此导致估计误差较大的问题,提出基于支持向量机分类细化特性曲线区,提高用以灰色GM(1,1)预测建模数据指数光滑度,改善转子信息估计精度的灰色近似支持向量机分类预测算法。将此预测方法用于永磁同步电机的矢量控制当中,数值仿真结果证明,引入先期近似支持向量机分类算法后的转子位置灰色预测法可以在较少测试数据集上达到较高的估计精度。
The magnetic characteristics curve modeling based on single gray model has shortcomings in distinguishing different running states of motors and summarizing, which causes a big prediction error. This paper proposes a proximal SVM classifiers based grey model prediction method （PSVMC-GMP） which classifies the characteristics curve areas in detail based on SVM classifiers, and improves the exponential smoothness of the GM （1,1） modeling data as well as the rotor information estimation precision. This prediction method is applied to vector control of PMSM, and the simulation results show that it has high estimation precision in less measured data setting.
出处
《电力系统保护与控制》
EI
CSCD
北大核心
2010年第23期97-102,共6页
Power System Protection and Control
关键词
永磁同步电机
转子位置自检测
灰色近似支持向量机分类预测算法
无传感器控制
permanent magnet synchronous motors （PMSM）
rotor position self-sensing~ proximal SVM classifiers based greymodel prediction（PSVMC-GMP）
sensorless control
