基于BP小波网络的故障模式识别

提出了一类新的BP小波网络，该网络采用BP学习算法，可实现信号的小波变换、特征提取和模式分类，具有结构清晰、算法简便的特点。将该网络应用于柴油机的活塞-缸套故障模式识别，并与用一般BP网络识别的结果比较，表明该网络对于非平稳时变振动信号具有很好的模式识别能力。

Pattern Recognition for Faults Based on BP Wavelet Network

In this paper, a kind of new BP wavelet network is presented. In the network, BP learning algorithm is applied, and wavelet transformation, feature extracting and pattern classification are implemented. Therefore, clear-cut structure and simple algorithm are characteristic of the networks. The networks is applied to the fault pattern recognition of the piston-cylinder of the diesel engine, and compared with recognized result by means of the general BP networks. The compared result shows that BP wavelet network possesses very good ability of pattern recognition for non-stable and time-variable vibration signals.