The number and size characteristics of wear particles in lubricating oil are the main basis for judging the wear degree.Because the wear particles in motion are easy to adhere,it is difficult to realize online monitoring and accurate feature extraction of wear particles.In order to improve the detection accuracy of sticky wear objects,this paper proposed a detection Algorithm based on Gaussian Mixture Modelling(GMM)and Watershed Algorithm.Firstly,the background subtraction algorithm based on GMM is used to initially detect the moving wear objects,and then the watershed algorithm is used to segment the extracted objects to complete the accurate detection of the adhered wear objects.The experimental results show that compared with the GMM algorithm,the GMM-watershed algorithm can effectively improve the detection accuracy of adhesive wear particles,and the accuracy rate is 98.86%,which provides an effective method for improving the accuracy of online feature extraction of oil wear particles.
维普
万方数据