Study On Fault Prevention And Diagnosis System Of Large-scale Rotary Machinery

Fault diagnosis system used in modern industry can solve the problem of machinery fault identification. Safeguard system can solve the problem of fatal accident prevention. DCS can solve the problem of technical parameter adjust. But there is no such flexible fault preventing and protecting system in modern industry that machinery isn't in the best condition in operation. Based on the main 3 kinds of industry-aided systems, the opinion of fault prevention and diagnosis of rotary machinery is presented.Based on sufficient study of working condition, mechanical property and product requirement of the research object, the theory design of fault prevention and diagnosis of rotary machinery is presented in this paper. The whole system includes three parts: data collection, data process and conclusion deduction, and control system. The system terminates fault inducement and makes machines run continuously as long as possible in healthy state.The relationship between matter condition and mechanical state is proved by experiment system we constructed. The experiment system uses air compressor as core, adding software we programmed. Lots of credible data are obtained by multiple tests under various conditions. The software includes vibration data acquisition with A/D card and property data acquisition with RS232 series communication, executing signal process module by means of VC linking Matlab C mathematics function database, transferring the signal process results into figures. Comparing and analyzing the figures, the qualitative relation between main-shaft mechanical state and inlets flux of the air compressor is observed.Data analysis and processing is another emphasis in the paper. Using Matlab as mathematical tool, neural network predict of main-shaft vibration, bearing temperature and surge is studied, and main-shaft vibration symptom of rotating stall is studied combining FFT with Wavelet analysis. The rotating stall process from initial phase to stable phase is wholly analyzed. What's more, the applications of Wavelet analysis in long period signal trend analyzing, single frequency picking up and signal denoise are discussed.