Fault Diagnosis Expert System Study Based On Artificial Neural Network For The Condensation Equipment

It is significatively that the Condensation Equipment works safely and steadily in the Nuclear Power Plant or the Regular Power Plant. However, the faults of the Condensation Equipment have pilosity nature, uncertainty and disguise which make difficulties for fault diagnosis of the Condensation Equipment. The fault sign and the fault reason of the Condensation Equipment have complex non-linear mapping relations, therefore it can not use simple model to describe. This has brought the puzzle for the classic fault diagnosis theory. The intelligent diagnosis method based on Artificial Neural Network (ANN) and Expert System (ES) can solve above problems greatly and the diagnosis results are more accurate and reliable.The problem which how to realize fault diagnosis of the Condensation Equipment with intelligent diagnosis has been researched in this paper.All kings of diagnosis technology and methods have been analyzed in this paper. The model of fault diagnosis ES based on ANN which is called ANNES for short has been established. This system is made up by memory system of knowledge, learning systems, reasoning machine, explaining device and human-computer interaction interface. Detailed structure and design method of every part have been provided in this paper.ANNES has merged the advantage that formidable adaptive learning ability of ANN and explicit knowledge expression of ES, simplified the progress of learning and gaining data of ANN and building inference rule. ANN is important part of ANNES and it is the criterion that ANNES is good or bad. This paper has approved that adaptive study rate has more advantages than regular BP whether precision or training speed. So, adaptive study rate method has been applied in this paper. The flow of this ANNES is:â‘ Form every Children ANN models and samples with the experience of experts or knowledge in the books. The system has used the adaptive study rate studying algorithms to train samples. Through adaptive algorithm of network, the system revises right values' distribution constantly so as to meet goal. Distribute the heuristic knowledge and experience knowledge to network interconnection and right values.â‘¡Through front computation of the network, the output obtains the output vector.â‘¢The explanation model of ES explain to users.In the end, this paper has realized the fault diagnosis of the Condensation Equipment of the Nuclear Power Plant based on ANNES intelligent diagnosis and it has proven this fault diagnosis method feasibility. Meanwhile, it has more proven this intelligent fault diagnosis method which has merged the advantages of ANN and ES can solve the equipment fault diagnosis problem that faults have pilosity nature, uncertainty and disguise such as the Condensation Equipment. The diagnosis results are precise and credible.