Abstract: A new algorithm，which is named as chaotic hybrid particle swarm optimization BP neural network（CPSO-BPNN），is proposed.CPSO integrates chaotic mechanism for its ergodicity，stochastic property，and regularity，which enhance the global exploitation of PSO.BP neural network has strong nonlinear approximation ability，but its nature of gradient descent algorithm determines that it’s easy to fall into local optimum and sensitive to the initial values.The CPSO-BPNN algorithm is in order to take the advantages of the two algorithms.It is applied to the non-linear modeling of Switched Reluctance Generator（SRG）.The efforts suggest that the IPSO-BPNN model has strong generalization ability，it can expression the flux and torque characteristics of SRG perfectly.

Key words: chaotic, Particle Swarm Optimization（PSO）, Neural Network（NN）, swarm intelligence, Switched Reluctance Generator（SRG）

摘要： 粒子群算法是解决非线性、不可微问题的一种优秀算法。利用混沌映射的随机性与遍历性，引入防早熟机制，加强了粒子群的全局搜索能力，但该算法仍然容易在进化后期出现速度变慢现象。BP神经网络具有很强的非线性处理能力和逼近能力，但BP算法是基于梯度下降的方法，存在容易陷入局部最优及初值敏感的缺点。将两种算法优势互补，构建了一种混沌粒子群优化BP神经网络（CPSO-BPNN）的算法。该算法应用到开关磁阻发电机（SRG）的非线性建模中，建模效果表明CPSO-BPNN算法的泛化能力很强，可以比较完美地表达开关磁阻发电机的磁链和转矩特性。

关键词: 混沌, 粒子群优化, 神经网络, 群智能, 开关磁阻发电机