Decision Fusion In Fault Diagnosis And Maintenance Process Under ICMMS

Rapid development and increased competition of modern industry have pushed the companies to keep on improving the reliability of equipments and reducing the equipments management cost to realize steady production and customer satisfaction, which calls for advanced techniques of fault diagnosis and maintenance decision making. The fact, however, is that with the increase in scale and complexity of equipments their behavior has become more dynamic and more uncertain,which has made poor performance in accuracy and real-time properties of the traditional diagnosis system dealing with only one single information source. Moreover, most companies take the scheduled maintenance strategy which often means high cost. ICMMS (Intelligent Control-Maintenance-technical Management System) can be used to improve the performance of fault diagnosis and maintenance decision making by connecting maintenance with control and technical Management. Trying to solve the problems above, this paper has done some research as followed:1. Firstly, we introduce the background and significance of fault diagnosis using information fusion technique and maintenance decision making using MDP(Markov Decision Process) technique under the ICMMS framework;2. Give a model of the fault diagnosis system under ICMMS and the designing principle of it, and then analyze the unique properties of decision fusion;3. In the synchronous decision fusion domain, in order to overcome the evidence conflict problem, a definition of consistency and a revised D-S rule of combination are presented. We try to get rid of the distortion the subjective side of experts'decisions brings to the diagnosis result. The experiment result shows the improvement brought by the proposed method;4. In the non synchronous decision fusion domain, a new method called CWA operator is proposed which takes into consideration both the size and the position of the sub decisions in the form of intervals; and5. At last, this paper presents a degradation model of equipments with Markov process. Based on this model, MDP method is applied to optimize the maintenance plan which is proved effective by the experiment.