论文部分内容阅读
提出了一种基于混沌变量的多层模糊神经网络优化算法设计.离线优化部分采用混沌算法,将混沌变量引入到模糊神经网络结构和参数的优化搜索中,使整个网络处于动态混沌状态,根据性能指标在动态模糊神经网络中寻找较优的网络结构和参数.在线优化部分采用梯度下降法,把混沌搜索后得到的参数全局次优值作为梯度下降搜索的初始值,进一步调整模糊神经网络的参数,实现混沌粗搜索和梯度下降细搜索相结合的优化目的,能较快地找到全局最优解.最后对二阶延迟系统进行仿真,结果表明混沌优化方法控制精度高、超调小、响应快和鲁棒性强.
A new design method of multi-layer fuzzy neural network optimization algorithm based on chaos variables is proposed.The offline optimization part uses chaos algorithm to introduce chaos variables into the optimal search of the structure and parameters of the fuzzy neural network, makes the whole network in dynamic chaos state, Index to find out the optimal network structure and parameters in the dynamic fuzzy neural network.The gradient optimization method is used in the online optimization to take the global suboptimal value of the parameter obtained after chaos search as the initial value of the gradient descent search and to further adjust the parameters of the fuzzy neural network , So that the global optimal solution can be found quickly by the combination of chaos rough search and gradient descent search.Finally, the second-order delay system is simulated and the results show that the chaos optimization method has the advantages of high control precision, small overshoot and fast response And robust.