The Fault Diagnosis Of Locomotive Converter Based On Data Mining

The electric locomotive is a mechanical system with very complicated structure. Because of its non-stop operation state, there is high demand to operation reliability of the locomotive system. The real-time fault diagnosis of locomotive is an important means to improve transportation efficiency.Data Mining, a new generation of tools and techniques for automatic and intelligent database analysis, is an active area with the promise for a high payoff in many business and scientific applications. On the other hand, knowledge acquisition has been a "bottleneck" with the rapid development of railway information technology. To deal with this problem, Data Mining technology is studied and implemented for locomotive fault diagnose in this paper.The system of locomotive is huge and complicated. According to the fact, we choose the locomotive converter as the research object in thesis. The foundations of several riper algorithms are suggested at first, and three methods are adopted: Decision Tree, Rough Set and Association Rules to solve problem.The SS8 converter is the main study object in this paper. The converter's simulation model is founded through the analysis of the converter's work principle in MATLAB 6. Through the simulation, we got the converter's output voltage waveforms when different components are in disconnection fault. And then, wavelet transform is adopted to make the power decomposition and constructed relevant character vectors.Owing to the redundant character vectors, it might cause that some fault cannot be recognized, and, too many vectors mean too many sensors. So, this paper adopted Data Mining to find out and delete the redundant character vectors. Further, a group of typical character vectors are used and the courses of mining are analyzed in detail.Finally, three algorithms are validated by constructing a Data Mining system.