The Application Of Asynchronous Electric Fault Diagnosis Method To Predict

Since invented from 19^th century, motors have been very popular because of excellent performance, convenient control and operation. As the supply electricity equipment and power mechanism which is the most popularly used and high quantity in the word, the asynchronous motor has been almost occupied in all fields. However equipments are developing in big scale, high speed, continuum and high automation way, which must make motors more and more complicated. So the chance of fault will correspondingly become bigger and bigger. In order to improve the security, reliability and economy of production and reduce operation cost, it is necessary for us to monitor asynchronous motors' state and predict fault.This thesis analyzes the mathematical model of the actual industrial asynchronous motor and sets up its simulation model in Simulink. With the help of virtual instrument development tools such as LabVIEW, it is easy to realize the monitoring of each type condition parameter when the motor is running. The solution is supplied for the prediction of the motor's fault by taking advantage of the monitoring data processed by the simulation software and the prediction model based on intelligent algorithm or mathematics modeling.The method of Trend extrapolation is an important way that gets the prediction goal by reasoning based on the logic thinking. When fault prediction is going on asynchronous motor, the change of stator and rotor's current can be predicted under the interior rules of asynchronous motor, by analyzing the continuous changeable current of stator and rotor and making use of the prediction technology of time sequence.The prediction model adopted by the thesis is based on BP neural network. BP neural network, an already mature network, with very strong non-linearity imitation ability, can approach to the prediction arithmetic operators excellently, which may greatly improve the accurateness of prediction. Compared to the other methods, it is general, objective and scientific to exercise BP neural network to predict the asynchronous motor's fault. Aimed at the convergence problem of BP algorithm, the thesis adopts the improved training algorithm, LM algorithm. Under the help of the algorithm, the training goal can be got in very short time.Furthermore, the state-of-the-art and the trends of the condition monitoring and the condition monitoring and the fault predict of the asynchronous motor are summarized and reviewed in this dissertation, indicating the expansive development of the software and hardware simulation and the main work that need to do further in the system simulation, also giving a prospect of the application in asynchronous motors by virtual instrument technology.