混沌梯度耦合优化算法及仿真研究

提出一种基于自适应混沌梯度下降的单目标耦合优化算法 .它采用变步长梯度下降法得到某个局部优化值 ,通过规则来判断其为局部极小值 ,然后利用一个由小到大变化的自适应尺度混沌遍历算法来获得一个更优值来代替局部极小值以跳出局部极小状态 ,全局优化值可以通过这种反复迭代来获得 .仿真结果表明 ,该算法能充分发挥梯度法寻优的快速性和混沌法寻优的全局搜索能力 ,有效地跳出局部极小 ,并快速找到最优值

Coupled Optimization Algorithm Combined with Based on Gradient and Chaotic Searching and its Simulation

A mutative scale chaotic gradient descending optimization algorithm based on gradient descending search combined with chaotic search for single objective optimization is presented. A local minimum, which is judged by two rules, is obtained by an improved mutative step gradient descending. A more optimal minimum is obtained to replace the local minimum by a mutative scale chaotic search algorithm which scales are magnified gradually from a small scale in order to escape local minima. The global optimal value will be attained by repeatedly iterating. The simulation result shows that it will make full use of quickness of gradient search and global scope search of chaotic optimization, the algorithm can jump local minimum and attain the global optimal value.