【作者】 龚迪琛；

【导师】 方方；

【作者基本信息】 成都理工大学 ， 核资源与核勘查工程， 2013， 博士

【摘要】 我国核电产业正处在快速发展阶段，按照国家的发展规划，到2020年中国的核电装机容量将由原来的4000万千瓦提高到7000万千瓦。安全高效的利用核能是保障核电高速发展的前提，核电厂控制管理的关键技术是建立在控制科学和技术的发展基础上的，控制科学的每一次发展都是对实际应用中出现的各种问题的深刻地思考和对困难的克服，智能控制是人工智能与自动控制研究的重要领域，智能控制在核电厂控制中的应用是近年来研究的热点，其中群智能的研究和应用受到了广泛的关注，群智能作为人工智能的一个分支，由于其具有应用方法可行性强、容易实现等优点，已在多个领域的研究中取得了良好的效果；在核电厂控制实体方面：随着直接数字控制技术DDC（Direct Digital Control）的发展，目前国内外普遍使用的核电厂计算机控制系统大部分采用了建立在DDC基础之上的DCS（Distributed Control System）分散式控制系统，其良好的性能保证了核电厂的正常运行；在反应堆堆型方面，国内外商用核电机组中压水反应堆占据了主要份额。本文以压水堆核电厂为对象，对反应堆功率控制、蒸汽发生器压差和给水流量控制进行了深入研究，在建立控制仿真模型基础上,从实时控制的角度将传统控制理论和智能控制理论相结合进行控制优化，从事物发展变化不确定性的角度出发，利用随机网络和随机过程的理论对核电厂的启堆运行过程、突发事故应急过程进行定量分析和算法优化，并利用现有的计算机软硬件条件，对核电厂控制运行信息管理系统平台提出了详细设计方案和技术细节分析，主要的研究工作和成果包括:（1）对核电厂反应堆功率控制的冷却剂平均温度控制方案进行了控制仿真模型的构建，对传统控制棒的棒速程序进行了模糊控制的设计和推导，在模糊棒速控制仿真结果的基础上，进行反应堆功率控制的主控制回路的瞬态过程分析和控制棒动作仿真研究，得到了模糊化棒速控制程序的效果和结论。（2）以蒸汽发生器压差和给水流量控制回路实例为研究对象，采用数字PID算法和粒子群优化相结合的思路，在粒子群优化结果作为初选目标的基础上，建立控制参数的优化采样范围。利用控制理论的稳态和瞬态相结合的分析方法，分析控制参数变化对结果的影响和变化趋势，从而最终获得控制参数优化结果，降低了寻优的盲目性。（3）从随机过程分析的角度将核电厂运行中的多个关联环节作为带有随机性质的过程进行分析。以反应堆启堆过程中冷却泵启动过程作为分析实例，定量计算冷却泵的启动成功概率、时间消耗以及与生产过程相关的条件概率。并从计算过程中提取适合于核电厂类似随机过程的分析方法。在对日本福岛核事故处理过程分析的基础上，针对核电厂的突发事件处理和核事故应急过程，以降低核事故和突发核事件的影响为目标，提出了当应急资源一定的前提下的优化计算方法。（4）我国核电厂目前普遍采用的是国外商用DCS控制系统，针对不同的应用环境对系统的需求是不同的，进口系统的功能扩展是受制于国外知识产权保护的，自主开发和设计是我国核电产业各系统的发展趋势，本文针对核电厂控制和运行信息管理，提出将实时控制信息、统计数据信息以及优化控制参数信息，通过网络化的平台进行数据整合，提出了详细的设计方案和技术细节，为核电厂相关子系统的自主软件开发提供了具体的思路。更多还原

【Abstract】 China’s nuclear power industry is in a stage of rapid development, in accordancewith the national development plan to2020, China’s installed nuclear power capacitywill be increased from40million kilowatts to70million kilowatts. Safe and efficientuse of nuclear energy provide protection for the rapid development of nuclear power.Nuclear power plant control and management of key technologies is built on the basisand development of control science and technology.Every development of controlscience is the solution of various problems in the practical application and deepthinking. Intelligent control is an important area of research in artificial intelligenceand automatic control, intelligent control applications in the nuclear power plantcontrol is a hot research in recent years. In addition the swarm intelligence researchand applications have also been widespread concern, swarm intelligence as artificialintelligence branch because of its advantages of application method is feasible andeasy to implement in many fields of research have achieved good results. In terms ofcontrolled entities: With the development of direct digital control technology(DDC:Direct Digital Control), nowadays at home and abroad widespread use ofnuclear power plant computer control system is mostly built on the basis of DCS(Distributed Control System), its good performance ensure the normal operation ofnuclear power plants. In the reactor type, the pressurized water reactor nuclear powerplants accounted for a major share of the domestic and international commercialnuclear power generating units. Pressurized water reactor nuclear power plant, andin-depth study on the reactor power control, differential pressure and feedwater flowcontrol of the steam generator is reflected in this paper. On the basis of a simulationmodel, the traditional control theory and the theory of intelligent control are usedfrom the perspective of real-time control of the control optimization. From the uncertainty of the development and changes of things, random networks and randomprocess theory are used in the start-up phase of reactor and in the accidentcontingency process analysis and quantitative algorithm optimization. Based on theexisting computer hardware and software technology，the control and operationinformation management system platform of nuclear power plant were designed andestablished in detail. The main research work and achievements include:(1) Construction of simulation model of the average temperature of the coolantcontrol in nuclear power plant reactor power control system and the comparison ofresult between traditional control rod speed program design and the fuzzy control rodspeed program.Transient reactor power control of the main control loop and thecontrol rod movement analysis and simulation were analysised in this paper. Rodspeed control of the fuzzy effect and conclusions are achieved.(2) Digital PID algorithm and particle swarm optimization combining ideas isused to get the preliminary goal and to establish the sampling range of the controlparameters in the instance of control loop of steam generator differential pressure andwater flow. The combination of steady state control and transient control analysismethod is used to analyze the affect of control parameters change and trend results,control parameter optimization is obtained and the blindness of looking foroptimization is reduced.(3) Uncertainty in the process of the control operation of nuclear power plantswere analyzed from the perspective of the random process. Cooling pump startupprocess in the start-up phase of reactor is used as a reactor startup process analysisexample, Cooling pump startup success probability, time consuming, and theconditional probability associated with the production process were calculated. Similarstochastic process analysis method suitable for nuclear power plant extract wereextracted from the calculation process. Optimization method was put forward in caseof certain preconditions emergency resources treated as unexpected events and nuclearemergency process for nuclear power plants to reduce the impact of nuclear accidentsand sudden nuclear events based on the analysis of emergency treatment process ofFukushima nuclear accident in Japan.(4) Nuclear power plants in china are currently widely used foreign commercialDCS control system. System requirements for different application environments aredifferent, the function of the system expansion is subject to the protection of foreignintellectual property. Self-development and design is the development trend ofChinese nuclear power industry. In this paper in connection with the nuclear power plant control and operation of information management, real-time control information,statistical data as well as optimal control parameter information were integrated in thenetwork platform for data integration. Detailed design and technical details were putforward. Specific ideas was provided to the independent software development for thenuclear power plant subsystems.更多还原