核电站报警触发式智能诊断专家系统的研究
发布时间：2018-11-20 17:47
【摘要】：核电站的安全运行始终是人们关注的焦点。为了使核电站对严重事故或威胁核电站安全运行的潜在事故发生的可能性降至最低,研究和开发核电站故障诊断系统,有利于提高核电站的可靠性,且对核电站的安全和经济运行具有重要作用和意义。本文提出了一种基于报警信息的核电站智能诊断专家系统。首先根据运行机理分析、核电仿真机和领域专家经验获取故障类型的征兆特征,同时,根据核电系统的特点,考虑了多样性原则;然后按照不精确知识表示方式构建以报警为单元的置信规则库;其次采用信息融合技术进行多重故障诊断,以提高诊断的准确性;最后采用改进层次分析法建立合理的推理控制策略,以提高诊断的快速性。其次,首次提出了智能诊断专家系统的自学习改进方法。在证据获取过程中,利用BP神经网络建立工况或故障程度与置信度函数的阈值和限值之间的数学模型,使系统可在全工况下进行故障诊断,提高了系统的工况自适应能力。最后,利用网络技术获取现场实时数据,集成数据库技术实现诊断过程中知识的连接与存储,采用模块化设计并利用VC++进行开发,根据人因工程学设计原理,在NetControl工业组态软件中实现系统的人机交互功能与诊断画面显示,并为操作人员提供故障原因分析和操作指导。仿真诊断结果表明,报警触发式故障诊断方法可行,对核电的故障诊断具有一定覆盖面,并且能够快速、准确地识别当前报警下的故障,进而提高核电站运行的可靠性。
[Abstract]:The safe operation of nuclear power station has always been the focus of attention. In order to minimize the possibility of serious accidents or potential accidents threatening the safe operation of nuclear power plants, the research and development of fault diagnosis systems for nuclear power plants is conducive to improving the reliability of nuclear power plants. And the safety and economic operation of nuclear power plants have an important role and significance. This paper presents an intelligent diagnosis expert system for nuclear power plants based on alarm information. First, according to the operation mechanism analysis, the nuclear power simulator and field experts experience to obtain the symptom characteristics of the fault type, at the same time, according to the characteristics of the nuclear power system, the diversity principle is considered. Then the confidence rule base with alarm as the unit is constructed according to imprecise knowledge representation, and the information fusion technology is used to carry out multiple fault diagnosis to improve the accuracy of diagnosis. Finally, a reasonable reasoning control strategy is established by using improved analytic hierarchy process (AHP) to improve the speed of diagnosis. Secondly, the self-learning improvement method of intelligent diagnosis expert system is proposed for the first time. In the process of obtaining evidence, BP neural network is used to establish the mathematical model between the threshold and limit value of the confidence function and the degree of failure, so that the system can diagnose the fault in the whole working condition and improve the adaptive ability of the system. Finally, using network technology to obtain real-time data, integrating database technology to realize the connection and storage of knowledge in the diagnosis process, using modular design and using VC to develop, according to the principle of human engineering design, The man-machine interaction function and diagnosis picture display of the system are realized in the NetControl industrial configuration software, and the fault cause analysis and operation instruction are provided for the operators. The simulation results show that the alarm trigger fault diagnosis method is feasible and has a certain coverage for the fault diagnosis of nuclear power, and it can quickly and accurately identify the faults under the current alarm, thus improving the reliability of the operation of the nuclear power plant.
【学位授予单位】：上海电力学院
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TM623
[Abstract]:The safe operation of nuclear power station has always been the focus of attention. In order to minimize the possibility of serious accidents or potential accidents threatening the safe operation of nuclear power plants, the research and development of fault diagnosis systems for nuclear power plants is conducive to improving the reliability of nuclear power plants. And the safety and economic operation of nuclear power plants have an important role and significance. This paper presents an intelligent diagnosis expert system for nuclear power plants based on alarm information. First, according to the operation mechanism analysis, the nuclear power simulator and field experts experience to obtain the symptom characteristics of the fault type, at the same time, according to the characteristics of the nuclear power system, the diversity principle is considered. Then the confidence rule base with alarm as the unit is constructed according to imprecise knowledge representation, and the information fusion technology is used to carry out multiple fault diagnosis to improve the accuracy of diagnosis. Finally, a reasonable reasoning control strategy is established by using improved analytic hierarchy process (AHP) to improve the speed of diagnosis. Secondly, the self-learning improvement method of intelligent diagnosis expert system is proposed for the first time. In the process of obtaining evidence, BP neural network is used to establish the mathematical model between the threshold and limit value of the confidence function and the degree of failure, so that the system can diagnose the fault in the whole working condition and improve the adaptive ability of the system. Finally, using network technology to obtain real-time data, integrating database technology to realize the connection and storage of knowledge in the diagnosis process, using modular design and using VC to develop, according to the principle of human engineering design, The man-machine interaction function and diagnosis picture display of the system are realized in the NetControl industrial configuration software, and the fault cause analysis and operation instruction are provided for the operators. The simulation results show that the alarm trigger fault diagnosis method is feasible and has a certain coverage for the fault diagnosis of nuclear power, and it can quickly and accurately identify the faults under the current alarm, thus improving the reliability of the operation of the nuclear power plant.
【学位授予单位】：上海电力学院
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TM623
【参考文献】
相关期刊论文 前7条
1 陈国云;范杜平;;核能发电的特点及前景预测[J];电力科技与环保;2011年05期
2 闫修平;彭敏俊;成守宇;;分布式故障诊断策略及其在核电站中的应用[J];核动力工程;2009年02期
3 慕昱;夏虹;刘永阔;;基于时序数据挖掘的核电厂故障诊断技术研究[J];核动力工程;2011年05期
4 张素兰;胡骏;;基于仿真与数据挖掘的故障诊断方法研究[J];计算机仿真;2007年01期
5 段隽U，
本文编号：2345565
本文链接：https://www.wllwen.com/kejilunwen/dianlilw/2345565.html
教材专著
