Research On On-line Detection System Of Rolling Contact Fatigue Based On Machine Vision

Rolling contact fatigue is the main form of failure of parts that transmit motion in rolling contact.This type of fatigue is sudden and can cause abnormal operation or even damage to the machine.Accurate detection of the rolling contact fatigue of the base material helps to prevent and reduce accidents caused by actual production,but the current testing method can not quantitatively detect rolling contact fatigue,and can not accurately obtain the rolling contact fatigue performance of the material.Therefore,the development of a detection system capable of quantitative and real-time detection of rolling contact fatigue performance of materials has great significance for the research of rolling fatigue contact failure mechanism and practical production application.Aiming at the shortcomings of current rolling contact fatigue detection methods,in this paper,a set of on-line rolling contact fatigue detection system based on machine vision is designed to realize quantitative detection of rolling contact fatigue,record rolling contact fatigue generation and extend to the whole process of failure evolution.The main research work is as follows:(1)The machine vision detection system for rolling contact fatigue testing machines was built.In order to solve the interference caused by the actual working environment such as the load vibration and lubricating oil splash of the rolling contact fatigue testing machine to the machine vision inspection system.The vibration-damping bracket for installing the camera and the light source and the oil-dissipating device for image acquisition are designed.Analyzed the influence of the angle of the light source and the intensity of the light on the gray value of the image,determined that the illumination angle of the light source is 60° and the illumination of 2600 Lux is the best parameter for the auxiliary illumination during image acquisition.(2)The related algorithms of rolling contact fatigue machine vision detection are studied.It mainly includes: DoG algorithm is used to realize image enhancement in the frequency domain;according to the prior knowledge of the morphological characteristics of fatigue points,the redundant part after threshold segmentation is screened out;and the linear CMOS camera is obtained by rectangular and circular metal calibration plates.Area pixel equivalent in image acquisition environment;quantitative detection of rolling contact fatigue using a method based on area threshold and distance-based point cloud density.(3)The detection error compensation model for rolling contact fatigue machine vision inspection under oil lubrication was established.The fatigue points are divided into 10 levels according to the size of the area,and the average detection error of each level under different flow rates is obtained,and the optimal error function is derived.Then the error compensation model is established based on the lubricant flow rate,test results and error function.The experimental results show that the fatigue point detection error is compensated by this method.The average value of the detection error ranges from 1.9% to 3.04%,which is 5.8% lower than before compensation.(4)The accuracy and stability of the machine vision inspection system designed in this paper are verified.The experimental results of the whole machine joint test show that the machine vision inspection system designed in this paper judges the accuracy rate of detection fatigue points can reach 96.5%,and the average detection of the maximum fatigue point area in the failure sample.The error is 3.34%.