| Asynchronous motor is the main force in daily life and modern industrial production.The operation of asynchronous motor will directly affect the daily life and production.Once the asynchronous motor is abnormal or fails,it may cause large and small faults in the whole system.At the same time,it may also directly affect people’s lives or cause major property losses.Therefore,it is very meaningful to diagnose and identify the fault of asynchronous motor to ensure that people’s life is safer,production efficiency is higher and quality is better.Firstly,on the method of intelligent algorithm,this paper proposes an asynchronous motor fault detection method based on improved support vector data description(SVDD)optimized by ant lion optimization(ALO);Then,aiming at the fault of asynchronous motor,a three kernel support vector data description(TKSVDD)based on health index and ALO optimization is proposed to analyze the degradation degree of asynchronous motor;Finally,aiming at the problems of insufficient speed and inaccurate identification of fault location in traditional methods,an asynchronous motor fault location identification model based on Ada Belief optimized convolutional neural network(CNN)is proposed to realize the accurate location of asynchronous motor fault location.The main research contents of this paper are as follows:(1)This paper expounds the research status and development trend of asynchronous motor,and introduces the common methods of fault diagnosis.By reading a lot of domestic and foreign documents.Firstly,the background and significance of the research content are analyzed;Then it focuses on the research status and development trend of rend in the research of intelligent algorithm and its application in asynchronous motor fault diagnosis;Based on the traditional research of asynchronous motor,common fault diagnosis methods are given,and analyze the shortcomings,which leads to the main work of this study.(2)This paper summarizes the basic methods used in the research of asynchronous motor fault diagnosis and identification.Firstly,the basic structure of common asynchronous motor and common fault types of asynchronous motor are introduced;Then,it expounds some basic concepts involved in this paper;Finally,the basic theoretical knowledge of the method used in this paper is introduced from three aspects: fault detection,degradation degree analysis and position recognition of asynchronous motor.(3)A fault detection method of asynchronous motor based on ALO optimized TKSVDD is designed.Aiming at the problems of low detection accuracy and inaccurate detection results in asynchronous motor fault detection,support vector data description is improved according to the basic theory of SVDD,and an asynchronous motor fault detection method based on ALO optimized TKSVDD is proposed.Based on the influence of noise in the current signal of asynchronous motor,in order to make the analysis results more accurate.Firstly,stochastic resonance is used to weaken the noise information in the signal and enhance the proportion occupied by the denoised signal.The reliability of the proposed method is verified by using the normal data and full fault data of asynchronous motor;Then,the model is trained according to the normal plus fault data,the asynchronous motor fault data detection model is constructed by ALO optimized TKSVDD,and the data are randomly selected to complete the asynchronous motor fault detection;Finally,the comparative simulation results show that the asynchronous motor fault detection model based on ALO optimized TKSVDD proposed in this paper can realize the fault detection of asynchronous motor quickly and accurately.(4)Based on health index and ALO optimized TKSVDD is proposed to analyze the degradation degree of asynchronous motor.Using the anti-noise performance and anti-signal instability performance of variational mode decomposition(VMD),VMD decomposition is realized for the faulty asynchronous motor data,and the signals with low noise interference and relatively stable are obtained;Then,an analysis model of asynchronous motor degradation based on health index and ALO optimized TKSVDD is proposed;Finally,according to the proposed model,the degradation degree of asynchronous motor is analyzed,and a comparative analysis method is proposed to verify the effectiveness of the method proposed in this paper.(5)A fault location identification model of asynchronous motor based on Ada Belief optimized CNN is proposed.Due to the model-based diagnosis method,it is very difficult to find the most suitable real model and the model that can simulate the operation of the machine under normal and fault conditions.The results of the deep learning method are clear and easy to understand,but general deep learning methods are prone to feature loss,and the model structure of CNN can effectively extract features.Because the performance of CNN often depends on the network parameters connected between neurons.It can be said that the network parameters directly determine the performance of the model.Therefore,this paper proposes an asynchronous motor fault location identification model based on Ada Belief optimized CNN.Firstly,the main characteristics of different fault parts are found out,the principal component of the current of the asynchronous motor is collected and analyzed by using the principal component analysis(PCA);Then the Ada Belief optimized CNN model is established;Finally,the training model of different parts of asynchronous motor faults and their main characteristics is used to test the samples to verify that the proposed method can quickly and accurately identify different parts of asynchronous motor faults,and can realize the rapid repair of asynchronous motor. |
