摘要
主要研究小波包变换和神经网络相结合的故障诊断技术。首先利用小波包的多分辨率分析的特点,对故障信号进行多尺度的分解,正交和归一化处理后,根据主成份分析原理提取故障特征向量作为神经网络的训练样本,设计故障类型识别器。仿真结果证实了该方法的有效性和可行性。
Research on the fault diagnose based on the combination of wavelet packet transformation and neural network mainly. By using the character of wavelet packet multiresolution, fault signal is decomposed at multi-scale, orthogonalation and normalization, extract feature vector according to principal component analysis theory, which as training input of neural network, design classifier of fault pattern. The validity and feasibility of the fault diagnosis method is demonstrated by a simulation.
出处
《微计算机信息》
北大核心
2006年第10S期232-234,共3页
Control & Automation
基金
广东省自然科学基金资助项目(0049496)
关键词
故障诊断
小波包
主成分分析
BP网络
Fault diagnosis,Wavelet packet,Principal component analysis,Back propagation neural network
