The Research Of Recognition System Of Wear Particle Atlas

Ferrography analysis technology is a unique method which realizes condition monitoring and fault diagnosis of mechanic equipments by observing the wear particles to find out the wear condition of inner friction pairs without shutdown or disassembling. The recogntion of wear particles is a key point of ferrography analysis technology. With the development of computer image dispose technology and artificial intelligence, the recognition of ferrography is intelligentizing and becoming a hotspot and difficult part in ferrography analysis technology area recently.In this paper, the method of IA-SVM ( Immune Algorithm to optimize parameters of Support Vector Machine) is applied to the recognition of ferrography analysis, and the WP_IA-SVM ( Wear Particle Classifier Based on IA-SVM) is designed, both of which are an exploration of new methods under finite specimems.This paper is summerized as follows:The background and significance of the topic are discussed: the status quo of technologies such as oil analysis and ferrography analysis both home and abroad are introduced. According to the request of this topic, the main research and main content are brought forward.The mechanism and classification of wear, the classification and characters of wear particles are analyzed and discussed. The character parameters of figure, texture and color and their formulations are presented.The methods of obtaining ferrography wear particle images, pretreatment process and character pick-up are expatiated. The wear particle images are obtained by using ferrogragh and pickup camera; the pretreatment of image is realized by combining the function of image treatment provided by MATLAB; The pick-up methods corresponding different eigenvalue of wear particles are analyzed.A new method is proposed based on SVM and immune algorithm: Immune algorithm to optimize parameters of SVM. This method can get the optimized recognition mode of SVM by training SVM using immune algorithm. And then SVM can utilize the trained mode to distinguish input. The IA-SVM is applied to the foundation of WP_IA-SVM. A application program with an operating interface is programed to classify normal, sliding, cutting, fatigue erosion wear and to do systemic simulation experiment, with which get a favorable effect of classification.This project is supported by a grant from the National Natural Science Foundation of China, which the authorizing number is 50375141.