Research On The Fault Diagnosis System Of Excavator Based On Support Vector Machine

With the rapid development of the computer, that the traditional industries enhance their competitiveness through information technology upgrade has become a trend. In the past several years, excavators developed greatly rapidly in China, but the technical backwardness, lack of technicians and other issues are gradually exposed. Therefore, it is imperative to find some ways to monitor operation-state of excavators and diagnose the excavator's faults quickly and accurately.Hydraulic excavators researched in this thesis are important construction engineering equipments, which are composed of mechanism, electric and hydraulic pressure system. Due to the uncertainties of fieldwork location, high-strength use of mechanical, high-suddenness of fault, urgent construction time, and so on, the diagnosis system is asked for diagnosing fault quickly and accurately. Access to a large number of relative books, it is found that many researches is relative to fault diagnosis in the mechanical engineering, but the intelligent fault diagnosis of excavators is very rare, almost blank. Therefore, this paper proposes an excavator intelligent fault diagnosis technology based on SVM to achieve rapid and accurate diagnosis.At the same time, this paper designed an intelligent diagnosis system based on safety clustering algorithm and support vector machine.This thesis is composed as follow. Firstly, introducing the research background and major work; secondly,analysing the principle and general faults of hydraulic excavators; thirdly particularly introducing the theory of support vector machine and some existing methods of SVM multi2class classification, then researching some common clustering algorithms, and on this basis, it proposes a new clustering algorithm based on safety and similarity. Then it introduces the excavator's fault diagnosis using a safety binary tree on detail. The process is that analysis the fault types using the security clustering algorithms at first, then raising a new safety binary tree and getting the classification function in every nodes, at last using this function to find the faults. This method has been used in the fault diagnosis system of excavator. The result means that this classification not only can find the faults, but also find the more important faults at first; finally, introducing the design and implementation of the excavator intelligent fault diagnosis system.Experimental analysis has proved that the fault diagnosis system of excavator studied in this thesis can achieve real-time fault diagnosis, improve work efficiency, and reach the expected requirement.