递阶遗传算法优化的模糊神经网络的故障诊断应用

Application of Fuzzy Neural Networks Optimized by Hierarchical Genetic Algorithm to the Machine Fault Diagnosis

提出一种利用递阶结构的混合编码遗传算法与进化规划相结合优化模糊神经网络学习的新算法,利用该算法同时优化模糊神经网络的结构和参数,剔除网络的冗余接点和冗余连接,提高网络的处理能力。分析和实验结果表明,所构建的机械故障诊断模糊神经网络结构简洁,而且具有良好的诊断效果。

A fuzzy neural network learning algorithm which combined a genetic algorithm based on hierarchical structure and evolutionary programming was presented. The shooting algorithm has been used to optimize the fuzzy neural network structure and train the connection weights, raise the processing ability of networks and eliminate the redundant nodes and weights. The analysis and experiment result showed that the method could get concise network structure of fuzzy neural networks and high performance of machine fault diagnosis.