基于混沌免疫遗传算法的神经网络及应用  

Application of Neural Network based on Chaos Immune Genetic Algorithm

作　　者：杨蕾 林红[2] 

机构地区：[1]中国石油大学(华东)理学院,山东青岛266555 [2]中国石油大学(华东)储运与建筑工程学院,山东青岛266555

出　　处：《智能计算机与应用》2011年第2X期10-13,共4页Intelligent Computer and Applications

基　　金：中央高校基本科研业务费专项资金资助（09CX04064A）.

摘　　要：借助混沌随机序列构造初始种群，将免疫机制引入传统遗传进化过程，有效克服传统遗传算法种群“退化”和“早熟”的不足，保持种群多样性，构造得到混沌免疫遗传优化算法。进而将混沌免疫遗传优化算法与BP神经网络相结合，分别用混沌免疫遗传优化算法和自适应BP算法对网络权值进行全局优化和局部二次优化，建立基于混沌免疫遗传算法的神经网络模型。利用所建立的混合神经网络模型对渤海某海域年极值冰厚进行训练预测，并将模型预测结果与实际数据以及动态拓扑预测的结果进行对比，表日周基于混沌免疫遗传算法的神经网络模型具有很高的预测精度和工程适用性。Through initializing population with chaos random sequence and introducing immune mechanism into traditional genetic evolution process, chaos immune genetic optimization algorithm is constructed to overcome population degradation and premature of traditional genetic algorithm and keep the population diversity. Combining the chaos immune genetic optimization algorithm with the BP neural network, neural network models based on chaos immune genetic algorithm is established to optimize the network weights for global optimization and locally quadratic optimization respectively. The ice thickness in Bohai sea is predicted by the model as an example, and a good fitness is revealed between the prediction and practical values. The results show that the proposed approach is feasible and effective.

关 键 词：混沌免疫遗传算法 神经网络 自适应BP算法 全局优化 海冰厚度 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]