Research On The Key Technologies Of Multi-agent Cooperative System Based Structural Health Monitoring

Structural health monitoring (SHM) technology is a research fucus in the engineering and academic domain. Much progress has been made in many aspects. However, using the technology on large practical structures, integration work must be solved in practice efficiently. Since density and different kinds of sensors have to be adopted to have a reliable monitoring of large scale engineering structures, the sensor information obtained by different sensors at different sites is distributed, diverse and heterogeneous. To coordinate and manage the distributed sensor monitoring network effectively, in this dissertation the multi-agent technology in Artificial Intelligence (AI) area is researched for SHM. Multi-agent cooperative system based SHM is developed. To evaluate the efficiency of the system, the function verification is conducted on a plate structure. In the research the mobile agent cooperation technology is also proposed to optimize the realization performance. The objective of the research is to promote the development of SHM technology in practical application.The main works and novel researched performed in this dissertation include:(1) The theory and construction principle of the multi-agent cooperative system based SHM are researched. BDI model of agent is researched. According to the deliberative agent, reactive agent and hybrid agent, different architectures are chosen to design different agents. KQML as the agent communication language is investigated. The architecture of the multi-agent cooperative system based SHM is designed adopting the composite architecture combing the advantages of network architecture and layed architecture.(2) The detailed process of constructing the multi-agent cooperative based SHM is presented. A general framework of the multi-agent cooperative system based SHM is developed. Seven kinds of health monitoring agents are designed and the cooperation, coordination and negotiation mechanism are estabished. The ontology design is presented to normalize the relationship of the concepts in the system. Facilitator design is adopted to act as the bridge among the agents. The work above realizes the automatic and effective cooperation in the system.(3) The blackboard model is proposed to solve the problem that the single damage evaluation agent can not deal with. An information fusion method on D-S evidence theory is adopted to solve the diagnosic result conflict problem. Functionally Accurate, Cooperative method is presented as the effective cooperation way among the subsystems. Sharing Information Management Agent (SIMA) is designed for every subsystem and the distributed database is estabished to provide information service for local subsystem and external information sharing. The subsystems exchange the results with each other by SIMA and cooperate together to realize the large structural health monitoring work.(4) The experimental verification is conducted on a large plate to validate the efficiencies of the system including the distributed monitoring, cooperation function and health monitoring for three typical kinds of structure states including strain distribution, bolt loosening and the impact load. Case-based reasoning (CBR) technology is presented as the effective coordination mechanism of diagnostic agents. Circular actuating/sensing in turn technology is proposed to monitor the bolt loosening for the large structure. Through the cooperation of different agents, the whole verification system can automatically choose sensing object, choose suitable signal processing method and damage evaluation method, self-organize the sensor network and discard useless sensor data to recognize the three typical structure states. Through collaboration, the static load happened in the edge area between two adjacent substructures is monitored successfully and the influence from environment parameters can be eliminated.(5) In the research the mobile agent cooperation technology is presented to optimize the realization performance. The efficiency of the technology is embodied through the verification instance.