Research And Design Of Distributed Intelligent Condition Monitoring And Fault Diagnosis System

The distributed intelligent monitoring and diagnosis system (DIMDS), which is of complex systems oriented, has become the extremely pressing requirement to the modern industrial and technological development. In this dissertation, the theory, developing method and application technique of DIMDS are studied. The main and creative research works are as follows: 1. Based on the elements and characters of DIMDS analyzed by using process algebra, an agent-based hierarchical logical architecture is presented, as well as the frameworks and functions of the primary agents. The AO system analysis, design and implement flow is generalized on the base of practical system development. 2. The multi-agents diagnosis system (MADS), which is the key part of DIMDS, is studied in detail. After analyzing its logical structure, the organizing framework, communication mechanism and collaborative problem solving mechanism are researched. Further, the two different distributed collaborative diagnosis modes are brought forward, including multi-agents-based diagnosis and mobile-agents-based diagnosis. 3. Aimed at improving the analysis and design work of DIMDS, a combined technique is presented that includes extending π -calculus to model the MADS's problem-solving behaviors, and using formalized frame method to analyze and describe the processes of distributed collaborative diagnosis, involving the MADS's problem-solving goal, structure and rule. 4. Focused on the computational space description of DIMDS information fusion, the information fusion computational theory is discussed. A theorem is proved that the information fusion problem-solving of fault diagnosis is computable, and enhances the diagnostic veracity. And then the information fusion framework of DIMDS is put forward. Furthermore, the fusion strategies to the fault diagnosis of the typical distributed communication system are presented, including the "0-1" distribution fusion algorithms and the probability distribution fusion algorithms based on D-S evidence theory. 5. Applying the research results mentioned above, a distributed communication system's network element management system (BSS-EMS) has been developed. Two parts of the system analysis and design work are expounded, in which includes the faults and alarms management function and the task process cooperation function. The overall work carried out based on the theory of process algebra, MAS and information fusion, etc, has built a design and study architecture of DIMDS, and provided a practical direction to the system development. Applied with the theory and technology presented in this dissertation, the intelligence and reliability of the distributed monitoring and diagnosis system are improved. Moreover, the system's preferable practicability and its promising situation have confirmed that the dissertation has a reasonable and farseeing objective.