Research On Motor Fault Detection Based On Neutral Network And Clark Transformation

With the rapid development of modern industry, motor has been necessary and widelyused in our daily life. Security, reliability, efficience have been some primary factors in motorapplication, However, the drives of load grow more and more complex, which will causemotor fault inevitably. Motor fault not only can damage motor itself, but also can causesudden stop, collapse the production line, and make the economic have a heavy loss when itbecomes serious, even bring a serious threat to life safety. Therefore, it is important toaccurately, timely find and diagnose the motor fault as early as possible in the operation. Themain work is study the fault diagnosis methods based on the common problems of motor faultin this paper.This paper firstly introduces the common fault of asynchronous motor and motor failuremechanism and characteristics, as well as the induction motor mathematical model. Then,based on this principle, the motor stator fault and rotor fault model are set byMATLAB/SIMULINK. Based above, motor fault diagnosis is realized through the MCSA(Motor Current Signature Analysis) method by using stator current.In this paper, on the one hand, extracting the fault characteristic information by usingwavelet analysis, and taking it as neural network input date, building the asynchronous motorfault feature library through training, the motor fault diagnosis method of using artificialintelligence is realized, on the other hand, realizing the fault detection by stator current trackusing Clark transformation. Lastly, researching the advantages and disadvantages of the twoways, and putting forward a dual mode combinative fault detection way, and the feasibilityreliability and superiority of the method was verified by using simulation analysis.