Study On Detection For Bearing Surface Defects Of Vehicles Based On Fuzzy Recognition

Bogie bearing is a key part of train configuration.During the running process of the train,the surface material of bearing inner and outer rings is subjected to the repeated action of high contact stress.The existence of surface defects will eventually lead to bearing failure,resulting in vehicle combustion shaft and other accidents.Paper aims at the common defects in bogie factory repair and section repair of rolling bearing inner and outer rings,through image preprocessing,feature extraction and hierarchical identification of defects,automatic identification and damage determination of defect types are realized,which has practical guiding significance for vehicle maintenance.The specific work is as follows:(1)This paper studies the common surface defects of vehicle bearings,including their morphological characteristics,locations,causes of formation and criteria for determining damage levels,analyzes the difficulties in automatic classification of defects identification,and establishes the overall framework of the identification system.(2)Preprocessing of defective images.For image filtering,the mean,median,Gaussian,and Wiener filtering are compared to study the effect of denoising on defective images.The peak signal-to-noise ratio is used as the evaluation criterion to determine the use of Gaussian filtering as the filtering method.Secondly,the histogram equalization is used to enhance the contrast of the image.Finally,the effects of denoising the defect image by Sobel,Roberts,Prewitt and Canny operators are compared and studied.Canny operator is selected as the edge detection tool in this paper.(3)To realize fast and accurate defect category identification,it is the key to choose the proper designed defect image features.For the defect image after preprocessing,the width,height,area,perimeter,elongation,rectangularity,compression degree,linear geometric feature and gray value are extracted according to the differentiated image characteristics of various types of defects.And statistics the distribution of the characteristics of various defects.(4)For the identification of defect types,in order to improve the identification efficiency,a coarse-to-fine identification strategy is adopted.For some defect image features that are significantly different from other types of defects,the corresponding features are used to gradually classify and identify the defect category by setting thresholds,and finally the damage is determined based on the image characteristics combined with the bearing segment repair procedures.For the defect classification with high similarity of defect image characteristics,the focus is on the research of the recognition method using fuzzy pattern recognition.Aiming at the defect image feature parameter values,severalfeature vectors selected for each defect are fuzzy modes.The membership function of the sample to be identified is designed,the membership degree of each feature attributable to the fuzzy set is calculated,and the classification category of the sample is finally identified based on the principle of maximum membership degree.(5)Established a vehicle bearing surface defect recognition classification system.Based on MATLAB software and GUI,a classification and classification loss system integrating image preprocessing,feature extraction,developing with the threshold classification and fuzzy classification,which can perform batch identification and determination of loss.Finally,aiming at the selected six types of defects,the threshold classification and fuzzy pattern recognition proposed adopting to give the defect types and damage grades accurately,which proves that the system can correctly realize the purpose of vehicle bearing surface defect detection and has certain application value.