Research On Motor Electrical Fault Location Method Based On Infrared Image Feature Analysis

Motor is an indispensable equipment in modern manufacturing and life.If it fails,it will not only affect the operation of the process industry,but also cause serious economic losses and even threaten personal safety.The electrical fault of the motor is a major type of fault during the operation of the motor.Compared with the mechanical fault,the external characteristics of the fault are hidden and develop rapidly.If the fault part is not found in time and effectively maintained,it will not only cause the motor itself fault,but also cause major accidents such as shutdown and shutdown.Therefore,it is very important to effectively identify and accurately locate the electrical fault of the motor.The traditional current signal can reflect the running state of the motor and detect the type of fault,but it cannot locate the fault location,which has certain limitations.Infrared detection is a commonly used supplementary technology for non-embedded fault diagnosis.It can effectively monitor the variation characteristics of temperature field,visualize,and locate hot spots or fault areas,and more intuitively represent the severity of faults.In this paper,by collecting and analyzing the current and infrared image characteristics of the motor during operation,a motor electrical fault identification and location method combining current signal and infrared image is proposed.Aiming at the problem that the traditional wavelet denoising algorithm cannot filter out the mixed noise and retain the details of the infrared image of the motor in the research of motor electrical fault diagnosis,an adaptive denoising method for infrared image of electrical equipment with effective edge information is proposed.On this basis,aiming at the problems of low accuracy,slow speed,and weak robustness of fault region segmentation in motor infrared image,an En FCM clustering segmentation method for infrared image of electrical equipment based on fusion reconstruction is proposed.The main research contents are as follows:The basic theory and imaging characteristics of infrared radiation are analyzed.Based on this,the infrared images of main electrical faults of AC motors are analyzed from the perspectives of histogram features,edge features and saliency features,which lays a foundation for the location and diagnosis of subsequent electrical faults of motors.In order to filter the mixed noise in the infrared image of the motor,an infrared image denoising algorithm combining improved wavelet transform and median filtering is proposed.Firstly,the adjustment parameters are introduced to improve the flexibility of the new threshold function and estimate the threshold of each layer more accurately.Secondly,the decomposition layer and the threshold are combined to quantify the high frequency coefficient,to avoid the loss of the original edge information of the target image due to excessive denoising.Finally,the residual salt and pepper noise in the reconstructed image is removed by median filtering.Aiming at the problem of segmentation accuracy and real-time performance of existing image segmentation methods on motor infrared images with strong spatial correlation and low contrast of fault area,an En FCM clustering segmentation method for infrared images of electrical equipment based on fusion reconstruction is proposed.Firstly,the adaptive morphological gradient reconstruction algorithm is used to reconstruct the image gradient,which solves the difficulty that the existing improved FCM algorithm must select different filters for different types of noise,and fuses the features extracted after saliency detection with the reconstructed gradient image to further highlight the contour of the fault area and lay the foundation for the accurate segmentation of the subsequent fault area.Secondly,the super-pixel image is obtained by watershed algorithm,which avoids the repeated operation of pixels,overcomes the problem of insufficient spatial local information of FCM algorithm and weak boundary attachment ability of segmentation results,and effectively improves the accuracy of fault region segmentation.Finally,the enhanced fuzzy C-means clustering algorithm is used to cluster the gray histogram of the image.At the same time,a faster membership filter is introduced to replace the slower distance calculation between the local spatial adjacent pixels and their clustering centers,thereby reducing the computational complexity and effectively improving the speed of the segmentation method.A motor electrical fault identification and location method based on current signal and infrared image is proposed.Mainly based on the motor stator inter-turn short circuit,the effectiveness of this method is verified by experiments.Based on the spectrum characteristics of the current signal,the fault type is identified,and the infrared image of the corresponding area of the motor is obtained under the guidance of the identification results of the current signal.The infrared image positioning method proposed in this paper is used to further identify the fault location.The experimental results show that the proposed method can effectively filter out the mixed noise,protect the image edge information,and quickly and accurately segment the position and contour of the fault,which lays a foundation for the further operation,maintenance,and safety guarantee of the motor.