Research On Development And Application Of Maintenance Decision System Based On Bayesian Network

With the development and application of high-tech weapons, their ability of fast fault diagnosis and maintenance support is becoming more and more important in modern war to enhance the behavior of weapon repair in battlefield, improve the rate of combat readiness and rebirth of battle effectiveness, and guarantee task success.This dissertation has been supported by two projects. One is the Research on Weapon's Fast Fault Diagnosis Technique Based on Data Fusion, and the other is the portable intelligent test system for secondary power supply. On the purpose of solving some problems in the weapon maintenance, a Bayesian network based fault diagnosis and maintenance decision software system has been developed, and applied in repairing of helicopter secondary power supply instrument.The main research work in this paper can be summarized as follows:(1) The key problems and popular description of fault diagnosis and maintenance decision behavior were analyzed first. Then, the expression method and mathematical description of the Bayesian network were described as well. After those, the function of Bayesian network based decision system was analyzed and its implementing scheme was given.(2) To solve the technical problems of the Bayesian network's usage in the fault diagnosis domain, including model construction, model inference and the model learning, respectively, the implementing scheme was realized and maintenance decision software was accomplished as a result.(3) With the integration with the portable intelligent test system for secondary power supply, the maintenance decision software has been developed as an online fault diagnosis platform. During the maintenance practice of helicopter, the maintenance decision system was tested and has been successfully improved its reliability.In a summary, this dissertation has developed a Bayesian network based maintenance decision system. With the integration of a test instrument, it realized a successful application case of transferring Bayesian network from theoretical research into maintenance practice.