Research On Intelligent Detection And Diagnosis Of Three-Phase Asynchronous Motor Fault

Asynchronous motor is the most widely used power driven equipment in today's industrial production.Its operation status directly affects the normal operation of industrial production.Once a fault occurs,it will lead to the paralysis of the entire production system,and even affect people's life and property safety and national security.Therefore,it is of great significance to monitor and diagnose the early faults of asynchronous motor to ensure that the production and living system can operate safely and efficiently with high quality and energy saving.In this paper,firstly,the method of asynchronous motor fault detection based on sliding mode observer is proposed in the method of analytical model;Secondly,the method of nonlinear observer design based on analytical model and BP neural network is proposed to solve the problem of observer design difficulty in nonlinear system,so as to realize the early fault detection of asynchronous motor;Finally,the traditional fault diagnosis needs to be complex because the signal processing technology only stays in the detection process,it can't realize the accurate diagnosis of the fault.The specific research work is as follows:(1)The research status of fault diagnosis of three-phase asynchronous motor is introduced.On the basis of reading a large number of domestic and foreign literatures,this paper first analyzes the research background and significance of asynchronous motor fault diagnosis,then focuses on the introduction of analytical model and deep learning in the field of fault diagnosis,and gives the hot issues in the research of fault diagnosis,finally gives the focus of this paper.(2)The existing fault diagnosis methods of asynchronous motor is described.Firstly,the mathematical model of asynchronous motor is built in different coordinate system,and the state equation in d-q coordinate system is established.Then several typical fault types of asynchronous motor are introduced in detail.Finally,the realization principle and process of fault diagnosis method are given based on analytical model,signal processing and machine learning.(3)A fault detection method of induction motor based on sliding mode observer is designed.Firstly,the basic theory of observer is introduced,including the observer design of linear system,the observer design of linear system with unknown term and the observer design of nonlinear system.Then,the design process and parameter selection of the sliding mode observer used in this paper are given.A simple and effective method is used to select the error feedback gain matrix,and the stability of the proposed observer is analyzed.(4)An intelligent fault detection method of induction motor based on improved BP neural network observer is proposed.At present,most nonlinear observers are designed based on Lipschitz condition,and their application has some limitations.Based on this,in this paper,a nonlinear observer is proposed based on the analytical model and BP neural network.This method uses the cuckoo algorithm to optimize the BP neural network(CS-BP)to predict the nonlinear part of the asynchronous motor.The designed nonlinear observer can accurately estimate the currents and speed of the motor.Finally,the winding fault experiment of asynchronous motor is carried out,and the online fault detection of asynchronous motor is realized through the analysis of current residual.(5)An intelligent fault diagnosis method of asynchronous motor based on deep learning is proposed.The method of fault diagnosis based on analytical model needs accurate mathematical model,but it is difficult to establish accurate mathematical model for nonlinear,strong coupling and multivariable asynchronous motor.Based on the feature extraction,analysis and selection of signal processing methods,researchers need to have sufficient theoretical basis for fault diagnosis.Deep learning has powerful expression ability,which can integrate signal feature extraction and pattern recognition.Therefore,this paper proposes a PCA-SVCNN based fault diagnosis method for asynchronous motor.Firstly,the collected current signal is processed by principal component analysis(PCA),and then the reduced data is processed by hybrid support vector machine(SVM)and convolution neural network(CNN).Finally,the intelligent fault diagnosis of asynchronous motor is realized.