摘要
以神经网络、小波分析和遗传算法等为代表的智能诊断技术,是故障诊断技术发展的一个重要方向。以传统故障字典法、BP神经网络、小波分析和遗传算法等基本原理为基础,将神经网络、小波分析和遗传算法与故障字典结合,用小波分解预处理故障信号提取故障特征,用遗传算法优化BP神经网络的结构和权值,对基于遗传小波神经网络的故障字典在模拟电路故障诊断中的应用进行研究,并结合实例验证其实际使用性能。
As the representative approach of intellectual diagnosis technique,neural network（NN）,wavelet analysis and genetic algorithm（GA） have been an important direction of current fault diagnosis technology development.Based on the basic theories of traditional fault dictionary,BP neural network,wavelet analysis and genetic algorithm,this dissertation has researched a new systematic method combining neural network,wavelet analysis and genetic algorithm in-depth for fault diagnosis of analog circuits,by wavelet transform extracting characteristic fault signal and genetic algorithm optimizing the connection weights and the membership functions.At last,the method is simulated with an example and the result is given,which indicates that the method fault dictionary based on genetic algorithm-wavelet neural network is effective and reliable for analog circuit fault diagnosis.
出处
《战术导弹技术》
2012年第2期107-112,共6页
Tactical Missile Technology
关键词
故障诊断
故障字典
BP神经网络
小波分析
遗传算法
fault diagnosis
fault dictionary
BP neural-network
wavelet analysis
genetic algorithm
