Motor Fault Diagnosis Technique Based On Fuzzy Neural Network

Asynchronous motor as the main power source of the modern production and life, hasbeen widely used in the national economy and people’s daily life. If the motor failure, seriouseven to cause the shutdown, will cause a great loss to the production and people’s life, so thefault detection and diagnosis of motor has the important practical significance. This articlesummarize the development and current status of the domestic and foreign asynchronousmotor fault diagnosis, we research the motor vibration signal and the stator current signalfeature extraction method, and put forward a kind of asynchronous motor fault diagnosismethod:diagnosis method into ANFIS fuzzy neural network, to solve the problem of faultdiagnosis of asynchronous motor, and its effectiveness is verified by experiments.According to the asynchronous motor fault diagnosis as the research object, we firstlyaiming at the working principle of asynchronous motor, analysis the fault mechanism ofmotor in the stator circuit, rotor broken bar, rotor eccentricity three faults, and detailedelaborate three kinds of fault vibration signal and the stator current signal characteristicfrequency energy change, as the standard of judged the motor fault.Secondly, according to the fault mechanism of asynchronous motor, we design theY132M-4type asynchronous motor fault diagnosis experiment, and measure the vibration andstator current signal through the experiment, and related process the data,extract the feature ofdata.Thirdly, due to the wide application of BP neural network in fault diagnosis, we use thetraditional BP neural network to diagnose the fault of motor. Based on the data of theexperiment, the test results are not ideal, reflecting the BP neural network in fault diagnosis ofthe motor is slow convergence speed, accuracy and local minima.Finally, aiming at the disadvantages of the fault diagnosis of BP neural network, we set up a ANFIS fuzzy neural network fault diagnosis system for motor fault diagnosis on thebasis of theory of fuzzy neural network. Then we use the extraction of characteristicparameters tio train and test the fuzzy neural network inference system,and compare with thetraditional BP training results, obtain the more ideal diagnostic results. It can effectivelyaccelerate the training time, reduce the error and avoid the local minimum value. The faulttype of the ANFIS system established in this paper can accurately and efficiently diagnosemotor, and is a kind of effective method for fault diagnosis.