摘要
通过对模拟退火算法(简称SA算法)进行深入研究发现,SA算法在寻优过程中随机产生的新点仅与当前状态有关,而与已搜索过空间的状态及其目标函数值毫无关系,这就浪费了大量有用的信息,因而SA算法最大的缺点是在搜索寻优过程中存在较大的盲目性。为此,该文将模糊推理技术用于SA寻优过程,可利用已搜索过的空间信息确定全局最优点所在的区间,从而缩小搜索范围使算法迅速收敛于全局最优点;并可自动去除不可行解,避免了电磁场逆问题计算中大量不必要的磁场计算。应用电磁场逆问题的分析方法,对开关磁阻电动机转子磁极几何形状进行了全局优化设计,在电机主要尺寸不变的条件下,显著地提高了样机的静态转矩。
This paper focuses on the inverse electromagne-tic field problem of switched reluctance motor (SRM). Firstly, Simulated annealing algorithm is thoroughly studied and it is found that there exists blindness to a great extent in the simulated annealing. Secondly, a novel global optimization method, intelligent simulated annealing algorithm, is proposed by assembling fuzzy inference into basic simulated annealing, which has the ability to reject infeasible solution prior to cost function computation, and by which the searching area is greatly reduced and the convergence speed is improved considerably. Finally, the intelligent simulated annealing algorithm is applied to the inverse electromagnetic field problem of SRM, and the static torques of a prototype motor is remarkably increased by optimizing the pole shape of the rotor .
出处
《中国电机工程学报》
EI
CSCD
北大核心
2003年第1期126-131,共6页
Proceedings of the CSEE
关键词
开关磁阻电机
磁极几何形状
优化
智能型模拟退火算法
转子
inverse electromagnetic field problem
intelli-gent simulated annealing algorithm
SRM (switched reluctance motor)