基于小波包频带能量检测的神经网络故障诊断技术

利用小波包频带能量检测技术对观测信号进行处理,从中获取反映故障特征的信息,以此作为输入对网络进行训练,可对故障进行比较可靠的分类。同时介绍了利用小波包频带能量检测技术提取观测信号特征向量的方法和步骤,以及基于松散小波神经网络的故障诊断方法。最后结合某型导弹舵系统故障诊断的实例,给出仿真试验,证明了该方法的有效性。

Neural Network Fault Diagnosis Based on Wavelet Packet Frequency Band Energy Detection

In neural network fault diagnosis technology,the wavelet frequency band energy detection technology can be used to process the observation signal,gain the information that reflect fault characteristics,and train the network by the information,and it is a rather effective method to classify the faults.This article mainly introduces the method and steps in using the wavelet frequency band energy detection technology to pickup the characteristic vector from observation signal,and the methods based on loose wavelet neural network fault diagnosis.Finally some examples of missile rudder system failure diagnosis are given,a simulation experiment is presented,and the effectiveness of this method is proved.