摘要
针对传统网络收敛速度慢、隐层节点数选取盲目的问题,提出了一种基于递阶结构的自适应遗传算法。该遗传算法采取基于递阶结构的编码方式和自适应调整遗传算子,以网络的复杂性和准确性为目标函数,同时优化小波网络的结构和网络参数,并将优化网络用于飞控系统舵机的故障诊断,通过与传统的BP算法比较,结果表明基于递阶结构的自适应遗传算法的网络结构优化能力很强,且网络的收敛性能和诊断能力都有了很大的改进。
An improved adaptive genetic algorithm （AHGA） based on hierarchical structure for wavelet neural network is presented, which adopts hierarchical chromosomes and adaptive genetic operators. This optimal, algorithm with objective function of network＇ s complexity and accuracy can optimize the structure and parameters of wavelet neural network simultaneity, and is used to fault diagnosis for flight control system, The simulation results show that the improved algorithm has strong structure optimization ability and high convergence speed,
出处
《测控技术》
CSCD
2008年第1期91-93,共3页
Measurement & Control Technology
关键词
小波网络
递阶结构
自适应遗传算法
故障诊断
wavelet neural network
hierarchical structure
adaptive genetic algorithm
fault diagnosis
