基于蚁群优化的Elman神经网络在故障诊断中的应用研究  

Study on Fault Diagnosis Based on Elman Neural Network Evolved by Ant Colony Optimization

作　　者：曹龙汉[1] 牟浩[1] 张迁[1] 余佳玲[1] 李景南[1] 

机构地区：[1]重庆通信学院 控制工程重点实验室,重庆400035

出　　处：《北京联合大学学报》2013年第4期30-35,共6页Journal of Beijing Union University

基　　金：重庆市应用开发计划项目(cstc2013yykfA40004)资助

摘　　要：指出普通Elman神经网络BP学习算法的不足,将蚁群算法和Elman神经网络相融合,采用信息素挥发因子ρ和信息素τ更新策略自动调整的自适应蚁群算法优化Elman神经网络的权阚值,并将其应用到柴油机涡轮增压系统故障诊断中。仿真实验表明,优化的Elman神经网络有较快的收敛速度和较高的故障诊断精度,可以有效地诊断柴油机涡轮增压系统的故障,能够更好地避免局部最小,实现了对柴油机涡轮增压系统故障的有效诊断。Points out the shortage of BP learning algorithm of common Elman neural network. Combining the ant colony algorithm and Elman neural network, we optimized Elman neural network values with adaptive ant colony algorithm which adjusts automatically by using pheromone volatilization p and pheromone T update strategy, and use it in fault diagnosis of diesel engine turbocharging system. Simulation results show that the optimized Elman neural network gets higher convergence speed and higher precision in fault diagnosis. It can also diagnose the faults of diesel engine turbocharging system effectively, and local minimum can be avoided, so that the faults of diesel engine turbocharging system can be diagnosed effectively.

关 键 词：ELMAN神经网络 蚁群算法 柴油机 涡轮增压系统 故障诊断 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]