Font Size: a A A

Research On Distributed Fault Diagnosis Technology For Nuclear Power Plant

Posted on:2017-03-08Degree:MasterType:Thesis
Country:ChinaCandidate:Z Y DuanFull Text:PDF
GTID:2348330518970762Subject:Nuclear Science and Technology
Abstract/Summary:PDF Full Text Request
Nuclear power plants demand high safety and reliability because of the existence of potential radioactive hazard when they work. And after Japan's Fukushima nuclear power plant accident happened in 2011, the nations of the world pay more and more attention to the nuclear safety. Fault diagnosis technology can find out the cause of the problem quickly and accurately when exceptions occurred in nuclear power plants, so the operator have enough time to take the corresponding safety measures to ensure the safety of nuclear power plants. This paper aims to study the distributed fault diagnosis methods and key techniques by combining the characteristics of the nuclear power plants, on this basis, an effective nuclear power plant fault diagnosis system was designed and built. The main research contents are as follows:(1) By taking the PWR nuclear power plant as research object, this paper divides the subsystems of the fault diagnosis system on the basis of distributed principle and the analysis of the overall diagnostic tasks. Some typical faults and their symptoms were studied in this paper, then the monitoring parameters of each subsystem were determined and the corresponding diagnosis knowledge base was established.(2) The fault diagnosis methods of BP neural network and decision tree inference were studied, and then a diagnosis strategy, which using BP neural network and decision tree inference for local diagnosis and evidence theory for global fusion, was researched in this paper.Furthermore, some key problems on the structure of the BP neural network samples and fusion in time and space were studied in detail.(3) The fault degree evaluation methods based on the BP neural network were studied, and the degree evaluation for some typical faults was realized approximately in this paper.(4) By using C# programming language, the software function modules of distributed fault diagnosis system were designed and developed from the perspective of the overall functional requirements.(5) Some testing and validation research of distributed fault diagnosis system, which has been built, were given by using the PCTran simulator of nuclear power plant.The result shows that the nuclear power plant distributed fault diagnosis system has high reliability and expected diagnostic ability, which can not only diagnose no matter single fault and simultaneous faults quickly and accurately, but also realize the evaluation of fault degree approximately. Overall,the distributed fault diagnosis system has a good performance,and which could be reference for future study.
Keywords/Search Tags:Nuclear Power Plants, Distributed Fault Diagnosis, BP Neural Network, Decision Tree, Information Fusion
PDF Full Text Request
Related items