摘要
针对BP算法易陷入局部最优,提出将一种新的混沌遗传算法(CGA)用于全局优化给水管网状态神经网络模型的初始权阈值.该算法将混沌搜索与自适应遗传算法相结合,根据混沌运动的初值敏感性、内在随机性以及遍历性的特点,通过混沌映射搜索自适应遗传算法的较优初始种群,并利用自适应遗传算法进一步寻优,对混沌映射和遗传进化进行循环计算直至达到最大进化代数,最终获得BP模型的较优权阈值.实例分析结果表明,与自适应遗传算法(AGA)相比,该算法搜索稳健,全局搜索能力强,并且新算法优化模型具有更高的预测性能.
As back-propagation (BP) neural network suffers from the existence of many local minima, a novel chaos genetic algorithm (CGA) combining chaotic search with self-adaptive genetic algorithm (AGA) was proposed for globally optimizing the initial weight and threshold of neural network-based state model for pipe network. With the characteristics of sensitive dependence on initial conditions, intrinsic stochastic property and ergodicity of chaotic motion, chaotic map was used to search the optimal initial population of AGA, then AGA was employed to optimize the weight and threshold. The operations of chaotic map and AGA evolution were performed until CGA reached the maximum generation, and the optimal weight and threshold of BP neural network was achieved. The case analysis shows that CGA has better global convergence and stronger running robustness than AGA, and that the model optimized by CGA has higher forecasting performance.
出处
《浙江大学学报（工学版）》
EI
CAS
CSCD
北大核心
2005年第6期874-877,共4页
Journal of Zhejiang University：Engineering Science
基金
国家自然科学基金资助项目(50078048).
关键词
自适应遗传算法
混沌
神经网络
给水管网
状态模拟
Backpropagation
Chaos theory
Convergence of numerical methods
Neural networks
Water distribution systems
