Research On Fusion And Decision-making Method Based On Multi-Agent System In Integrated Diagnosis

With the progress in science and technology, the reliability and security of equipments attracts more and more attentions. For a better capability of fault diagnosis and maintenance support, all the resources of testing and diagnosis should be shared and fused, i.e. the novel idea of Integrated Diagnostics. It aims at the optimized capacity of the all and the one, in allusion to the system of the equiments.In the integrated diagnostic system, fault diagnosis capability designed for the equimnets comprises several scattered units of testing and diagnosis. For instance, on-line test unit, in situ testing equipment, automatic on-site test system, remote diagnostic systems, etc. The elements of that should be integrated into the diagnostic system to improve the diagnostic efficiency. The resources used include not only local data and knowledge, but also the long-range knowledge base and experts'technical support, which are distributed in space. Therefore, one of the key problems in the integrated diagnosis is to develop the fusion and decision-making algorithm in a distributed integrated diagnostic framework.This dissertation takes the demands of distributed fusion and decision-making in Integrated Diagnosis of the diesel engine as the initiative. It is quite significant in achieving the integrated diagnostic target and improving the ability of fault diagnosis and maintenance support for new equipments.The detailed contents and innovative work can be summarized as follows.1. According to the diagnostic characteristics of the diesel engine, an integrated diagnostic model was built based on the Multi-Agent System and fusion algorithm.2. Concerning the distributed structure of the integrated diagnostic system, a practical scheme of the Multi-Agent System was designed, and the communication mechanism among the Agents was expatiated to achieve the interaction in physical layer and protocol layer.3. Based on the above model, the fusion program was studied for the system. In this part, the design requirements of the knowledge base and the relation between the knowledge base and a BP neural network were expatiated, as well as the methods of building a network and the fusion in decision-making afterwards.4. Based on the theoretical researches, the integrated decision-making system was designed, after that is an experimental verification.