Research on fault diagnosis for TBM Based on wavelet packet transforms and BP neural network

Analyzed the particularity of the TBM work environment and the superiority of virtual instrument for condition monitoring, and built a virtual instrument-based TBM condition monitoring systems. Researched the collected method of certainty feature vectors based on wavelet packet transform, and verified the applicability of this approach. A combination diagnostic methods of wavelet packet transform and BP neural network for fault diagnosis was proposed. In the process of applying this method, presented the method to adjust the weights of neural netwoek by the second learning to influent oefficient weighting method. Built a TBM condition monitoring and fault diagnosis system using LabVIEW and Matlab software. And shared the system online by using the web publishing tool. The technical feasibility were validated by the results of the operation of the system.