摘要
针对开关磁阻电机驱动系统具有非线性且结构参数变化范围较大的特点,提出了将混合遗传算法和神经网络相结合实现对开关磁阻电机驱动系统辨识的新方法。该方法结合混合遗传算法与神经网络各自的优点,克服了传统BP神经网络收敛速度较慢以及易于收敛到局部极小点等缺点。仿真试验表明,采用该方法能较迅速、准确地逼近实际系统,具有效性。
Based on the characteristics of the nonlinear and the large range of structures and parameters of switched reluctance motor drive system, a new method that the combination of mixed genetic algorithms and neural network is put forward to achieve the identification of switched reluctance motor drive system. The principle is expounded and corresponding algorithm and formulas are presented as well. The method combines the advantages of the optimum genetic algorithms and neural networks which overcomes the shortcomings of traditional BP networks such as the slow learning rate and liable to converge to the local minima. The simulation results demonstrate that this method is quite practicable for its fast and exact closing to its real system.
出处
《重庆科技学院学报（自然科学版）》
CAS
2009年第6期121-124,共4页
Journal of Chongqing University of Science and Technology：Natural Sciences Edition
基金
黑龙江省教育厅科学技术研究项目(11533058)
关键词
开关磁阻电机
辨识
混合遗传算法
神经网络
switched reluctance motor
identification
mixed genetic algorithms
neural network
