摘要
为了加快模拟电路故障优化算法收敛性和优化效率,采用了一种基于自适应粒子群算法的模拟电路故障诊断方法。利用小波分解技术提取待诊断电路的测试信息作为故障特征,借助于遗传变异的思想在粒子群算法中引入"变异"理念,然后使用自适应的粒子群算法优化RBF神经网络的结构参数,实现模拟电路的故障诊断。仿真结果表明,利用自适应粒子群算法与传统的粒子群算法相比,其训练步数明显加快,扩大算法的搜索范围,从而有效提高了网络的训练速度和优化精度。
An adaptive particle swarm optimization algorithm and optimizing efficiency of fault diagnosis for analog circuit. circuit by wavelet technology. APSO is formed based on PSO rameters are optimized by APSO to realize the circuit fault （APSO） is proposed to accelerate the convergence Firstly, the fault feature is obtained from the analog algorithm and genetic mutations. Then the RBF pa- diagnosis. The simulation results show that the proposed method is faster than traditional PSO, the step is much less than PSO and the searching range is enlarged, thus the speed of network training and the optimization accuracy are improved.
出处
《电子测量与仪器学报》
CSCD
北大核心
2015年第6期901-906,共6页
Journal of Electronic Measurement and Instrumentation
基金
芜湖职业技术学院自然科学项目基金(Wzyzr201323)
关键词
模拟电路
神经网络
故障诊断
粒子群算法
analog circuit
neural network
fault diagnosis
particle swarm optimization
作者简介
张松兰，1973年出生，毕业于南京理工大学，现任芜湖职业技术学院副教授。目前主要研究方向为人工智能、系统仿真及应用测试等。E-mail：zhajamin@sina．cn
