Research On Agent And Support Vector Machine Based Remote Intelligent Diagnosis Technique

Technique of state monitoring and fault diagnosis can improve thereliability and the maintain ability in effect. As the popularization andapplication of Internet technique, the research and application of the distributedremote intelligent diagnosis system becomes the current hotspot. Thecomplexity of their structure and the variety of their environment lead to thehigher requirement for intelligence and the distribution of the fault diagnosis.The Agent technique with the character of intelligence and distribution, and thenew learning method named the support vector machine, become the other tworesearch hotspot.Under the systemic lucubrating, both the virtual instrument technique andthe wavelet technique are used to collection of the shake signal of the pumpshell in this dissertation. Support vector machine, Agent and Internet are fusedand then introduced to the remote intelligent fault diagnosis.The main works in this dissertation are showed as follows.(1)The Agent technique is particularly summarized and introduced into thefault diagnosis field. A type of Agent model based the complex framework isadvanced for diagnosis and maintenance. The Multi-Agent technique issummarized and a new hiberarchy of fault diagnosis system based Multi-Agenttechnique is advanced.(2)The support vector machine technique is particularly summarized, thechoice principle of some kernel function parameters is researched. Research ofthe pump's fault diagnosis is done with it.(3)Two types of frameworks, C/S and B/S, used in the remote faultdiagnosis system is particularly summarized and compared. The project ofremote intelligent fault diagnosis is constituted with the B/S. Tomcat server ischose to build the Web server, MySQL is chose to build the database, and theJSP for diagnosis is compiled with the fusion of Agent and support vectormachine.(4)LabVIEW is used to design the data acquisition, noise abatement andcharacter extraction system for the pump with the wavelet technique. The shakesignal of the pump shell is collected, noise-abated and character-extracted, andthen the remote diagnosis is done.