摘要
探索单绕组磁悬浮开关磁阻电机的结构优化设计,研究了一种基于极限学习机与粒子群的设计方法。该方法在有限元仿真分析的基础上给出各结构参数对悬浮力影响的一般规律,并选择定子极弧、转子极弧作为优化对象,利用仿真数据进行极限学习机的模型训练,给出了样本空间设计,对模型的训练效果进行了定性定量的描述,并与支持向量机进行了训练效果对比,最后选用粒子群算法对电机训练模型进行寻优以提高悬浮力输出。结果表明,基于极限学习机的训练模型精度高、模型回归速度快,粒子群算法能快速准确地寻取最优解。
To realize the optimization design of the single winding bearingless switched reluctance motor, a design method is studied using extreme learning machine and particle swarm optimization algorithm. Firstly, general rules of effects of various structure parameters on radial forces are given based on finite element analysis, and stator pole arc and rotor pole arc are selected accordingly for optimization. Then nonlinear regression model is trained by extreme learning machine with the data from the finite element simulation. Besides, sample space is presented, and effects of the trained model are showed qualitatively and quantitatively, and then comparisons with support vector machine are made. Finally, the particle swarm optimization algorithm is used to search for the optimal solution. The results prove that the nonparametric model has good precision and fast speed of regression, and the particle swarm optimization algorithm is able to find the optimal solution quickly and accurately.
出处
《控制工程》
CSCD
北大核心
2015年第5期939-945,共7页
Control Engineering of China
基金
国家自然科学基金(51377074
51307077)
江苏省优势学科建设工程资助项目(苏政办发(2011)6号)
南京工程学院校级科研基金
关键词
单绕组磁悬浮开关磁阻电机
优化设计
极限学习机
粒子群
Single winding bearingless switched reluctance motor
optimization design
extreme learning machine
particle swarm optimization
