论文部分内容阅读
针对协同微粒群算法不能保证收敛到局部或全局最优值的问题,提出一种改进协同微粒群算法(ICPSO),并证明了该算法能以概率1收敛于全局最优解.应用ICPSO建立一类非线性对象的神经网络辨识模型,并对系统的模糊神经网络自适应控制器的参数进行了离线和在线优化.仿真结果表明,ICPSO能提高系统的建模精度,增强模型的泛化能力,而且由ICPSO训练的控制器可以达到良好的控制效果.
Aiming at the problem that the PSO can not guarantee the convergence to the local or global optimal value, an improved Cooperative Particle Swarm Optimization (ICPSO) is proposed and it is proved that the algorithm can converge to the global optimal solution with the probability 1. Using ICPSO to establish a The nonlinear neural network identification model of nonlinear objects is studied and the parameters of the system are optimized offline and online.The simulation results show that ICPSO can improve the modeling precision of the system and enhance the generalization ability of the model, And ICPSO trained controller can achieve good control effect.