Fault Diagnosis System Of Machinery Vibration Based On The Neural Network

With the development of machine equipment complexity and the improvement of automatic level, the subject of machine equipments fault diagnosis is more and more important. So choosing fit fault diagnosis method seem to be more important to the result whether the precision.Along with the computer technology popularization, the breakdown intelligent diagnosis demonstrates the enormous superiority. The artificial nerve network imitates the physics structure of person's brain, with its strong ability of parallel computation and association, it is very suitable to diagnose the machine equipments fault.First, this article system introduction mechanical breakdown diagnosis technology importance and the domestic and foreign development present situation, in the tendency foundation, had pointed out the artificial neural networks theory applies in the mechanical breakdown diagnosis technology has the enormous application value and the development potential. Simultaneously also pointed out that, our country the artificial neural networks theory is at present imperfect in the mechanical breakdown diagnosis domain application, still was at the development phase.Then, the development of fault diagnosis technology,the fault diagnosis technology of rotating machinery have been discussed, then introduces the characters of rotating machinery faults and the mechanism of rotating machinery. At first the structure of the back propagation neural networks is presented. Aiming at the problems that BP algorithm has slow convergence rate and is likely to fall into local minimum point, Levenberg- Marquardt (LMBP) algorithm improved on numerical optimization has been introduced. Then it has built a suitable BP model for rotating machinery faults diagnosis and has used in the factory.This article puts forward to fault diagnosis system of machinery vibration based on the neural network. By a concept of object-orientated, the system is programmed in the operating platform of Window and Delphi, the developing environment. This system mainly includes the document management module, the parameter establishment module, the signal analysis module the neural network training and the diagnosis module as well as the system help module.In the last, according to the examples and analyses, this articles confirm that the improved BPNN can recognize kinds of familiar faults of rotating machinery truly and quickly, it is adopted that the neural networks diagnose the fault. The result indicates the effectiveness of this method having the very good application prospect since that it can improve the efficiency of diagnosis and the service efficiency, also can reduce the equipment service cost.