Research On Gear Fault Diagnosis Based On Wear Identification

The gear, which plays an essential role in power transmission, is one ofthe most commonly used drive parts in machinery and equipment. Normalworking conditions and reliability of the gear guarantee the whole drivesystem efficiency. At present, the gear fault diagnosis of the commonly usedtechniques includes traditional diagnostic techniques and intelligentdiagnostic techniques, monitoring techniques such as vibration, noisemonitoring, oil analysis, infrared thermography, nondestructive testing,acoustic emission technique and neural networks, fuzzy logic, expert systems,genetic algorithms, etc., which has important applications in gear conditionmonitoring and fault diagnosis. With the continuous development of imageprocessing technology in machinery industry has become increasinglywidespread. Accordingly, the proposed research ideas based on digitalimage processing techniques applied to fault diagnosis of gear wear forfurther studies provides new means of detection of gear wear, realizing anew identification method in gear wear fault discrimination.Our research work in this thesis aims at building an availablediscrimination system of gear wear failure for the practical application and isdivided into five major sections: At first, gear image preprocessing. Clean andtarget-characteristics-distinguished images are obtained through the distortionand noise processing to them; Then, study on the gear image edge detectionand extraction algorithm. Through the comparative analysis of the bestoperator a true reflection of the tooth profile curve on the actual tooth edgecan be extracted.And then, discrete image processing. Through thecomparative study between the standard gear tooth profile obtained from theanalytical formula and the actual tooth profile obtained from imagerecognition, the amount of gear wear can be achieved; In the meanwhile, inorder to obtain the threshold of the amount of gear wear extent, the related experiments were carried out. At last, the system were developed and a set ofvisual gear wear fault diagnosis system were established with the use ofrelevant software.This study covers a wide variety of subjects and technologies, includingdigital image processing technologies and system development technologies.Our research ideas and methods are up-to-date, while there are little relevantresearch domestically, and therefore it evokes a certain theoreticalsignificance and value of innovation. The method is also practically of greatsignificance enriching the gear fault diagnosis technology and providing anew direction for the wear fault diagnosis of gear parts.