Research On Online Intelligent Diagnosis System Of Reciprocating Compressor Based On Multi-agent Technique

Reciprocating compressor is one of the most widely used mechanical equipment in the petrochemical industry. As a key equipment of industrial production, its functioning has an important impact to the safety of the entire production line. Condition monitoring and fault diagnosis of reciprocating compressors has practical significance and economic value. Traditional intelligent method has certain limitations in the face of the fault diagnosis of large and complex systems, the advantages of Multi-Agent responsiveness, sociality and initiative can make up the defects of current intelligent diagnosis in engineering applications, and achieve intelligent diagnosis on large and complex system failure. The application of reciprocating compressor fault diagnosis based on Multi-Agent Technology is in-depth researched.On the basis of analyzing basic structure, working principle, studying technical information and previous experience in fault diagnosis of the reciprocating compressor, typical failure of the reciprocating compressor and its fault characteristics is summarized.The overall framework of a reciprocating compressor fault diagnostic system is proposed in this research, which is based on Multi-Agent. The structure, function and collaborative relationship of each Agent of the framework are also designed. And the framework has better scalability.The reciprocating compressor fault diagnosis method based on fuzzy information fusion is proposed. Hybrid decomposition strategy is used in Multi-Agent reciprocating compressor fault diagnosis system, which is using structure decomposition in high-rise and using fault decomposition in the bottom. Improved contract net approach was used in the diagnosis task allocation. Three intelligent diagnostic methods, expert systems, neural networks and fuzzy logic, and fuzzy information fusion method are Introduced.The validity and accuracy of the fuzzy information fusion method is verified by simulation.Reciprocating compressor fault diagnosis system based on Multi-Agent is developed. The overall program of reciprocating compressor online condition monitoring and fault diagnosis system is designed, and the diagnostic object, hardware components and the structure of the database of the diagnostic system is expounded. And then using LabVIEW and MATLAB hybrid programming, the system of reciprocating compressor fault diagnosis based on Multi-Agent Technology is developed, the diagnostic system has been used in reciprocating compressor in a petrochemical workshop. After three months of trial operation, the diagnostic system is verified to be stable and effective.