Study On The Application Of Data Fusion Intelligent Diagnostic System To Nuclear Power Plants

In order to satisfy the high safety requirement of nuclear power plants (NPP), many operation support systems appear continuously, in which fault diagnosis system has increasingly draw much attention from researchers as one kind of important assistant system. The function of fault diagnosis system not only limits to the only fault diagnosis, the problems of fault forecast, fault manages and so on have also been integrated in the fault diagnosis system function. In allusion to the research of fault diagnosis system as present and the self characteristic of NPP, building the synthesis and intelligent diagnosis system are proposed, which includes the functions of condition monitoring, trend forecast, fault diagnosis, fault location and repair decision. The research of this diagnosis system has very important engineering and reality meaning to further enhance the safety and reliability as well as the whole effectiveness of NPP.This thesis is supported by the National Defense Advanced Research Key Project "XXX real time fault diagnosis technology", and takes the main coolant system of primary loop as research object. The technology and method of intelligent fusion diagnosis system realization and the basic structure are deeply researched in this thesis.The research work and the major achievements can be summarized as follows:1. The perfect fault diagnosis system of NPP should have each basic function was proposed. In order to satisfy the demand of NPP operation safety, the diagnosis system should include the functions of trend forecast, status evaluation, repair decision, fault diagnosis, fault location and fault post-processing etc.2. The diagnosis method of data fusion and dynamic ordinal fault lattice was put forward in the based of the data fusion's basic classification idea and the advanced intelligent diagnosis algorithms have been designed.3. The grey Markov GM(1,1) model is used as the automatic forecast method through further research grey system theory and the prediction analysis to dynamic data of relative grey forecast models. The forecast models can choice according to the instance in order to satisfy the prediction of no abnormal parameters which facilitate the contrast analysis.4. The Ant Colony Optimization is used to fault location, and the implementation technology is research emphasis. The NPP are graded then using the Ant Colony Optimization searches abnormal parameters to location fault step by step.5. The fuzzy variable weight evaluation and improved repair optimization method are researched, that evaluates the state of NPP and calculates the detection or maintenance cycle of equipments, which can determine whether the equipments need detection or examination and the mode of repair.6. The intelligent diagnosis system has been successfully developed using Visual Basic 6.0, and has realized each kind of anticipated design function. Friendly man-machine interface has been designed for the system. The problems of the diagnosis system's extensibility and the diagnosis knowledge unceasing renewal and so on are solved through the design of diagnosis knowledge updating subsystem. The knowledge database of diagnosis knowledge updating subsystem can be expanded and revised through man-machine interaction easily by user with deepening to NPP security characters and accumulating of fault data.7. The system successfully realized the functions of automatic forecast, diagnosis and alarm automatically.In order to validate the NPP data fusion diagnosis system and its theories, first the examples are used to validate the every function, then the Communication Protocol is used to data communication with the simulator of NPP and the simulator validates the diagnosis system generally. The test results showed the diagnosis system is suitable for the NPP real-time fault diagnosis. They also confirmed the feasibility and the validity of fault diagnosis according to the fault premonition. The accuracy of the designed fault diagnosis system has also been verified. These are the basis of the diagnosis system for the further research and application.