火电厂机组负荷优化分配与决策研究

发布时间：2018-02-20 17:09

本文关键词： 火力发电负荷分配多目标优化动态规划自适应网格多属性决策　出处：《东北大学》2014年硕士论文　论文类型：学位论文

【摘要】：随着我国电力体制改革的不断深化,“厂网分开,竞价上网”机制的推行,使发电企业从生产型企业转变为经营型企业,电厂作为独立的经济实体参与市场竞争,电网不再对单元机组下发负荷指令,而是对电厂中所有机组下发总的负荷指令。因此,科学合理地分配每台机组的负荷,将有利于优化机组的运行水平,有利于降低全厂的供电煤耗,为发电厂在电力市场中竞价上网提供科学依据。同时,国家的对企业的节能减排和可持续发展政策,也给发电企业带来了新的要求。火电厂在发电过程中会产生大量有害气体,对环境造成严重的污染,污染物排放已成为火电厂机组负荷分配中不可忽略的因素。因此,电厂机组负荷优化分配问题的研究具有非常重大的现实意义。本文介绍了机组负荷优化分配的基本概念,根据国内外机组负荷分配的研究现状,并结合火电厂机组的实际运行情况,选定供电煤耗作为负荷分配的经济性指标,确定单元机组的供电煤耗特性曲线是火电厂机组负荷优化分配的基础,采用最小二乘法对单元机组的煤耗特性曲线进行拟合。在对单目标负荷优化分配算法的研究中,利用机组的煤耗特性曲线建立了基于经济性的单目标负荷优化分配数学模型,然后详细介绍了等微增率法和动态规划法应用于负荷优化分配的方法,并对比了它们的特点,选用动态规划法作为单目标负荷优化分配算法,并对算法进行详细的设计和说明。算例结果表明,动态规划法可以在一定程度上优化全厂机组的运行,从而降低全厂煤耗,保证了机组负荷分配的经济性。本文在单目标优化模型基础上,进一步考虑减排目标,建立了基于环境和经济的多目标负荷优化分配模型,设计了基于自适应网格的多目标粒子群算法,并对该算法得到的Pareto解集进行多属性决策,得到最满意的负荷分配方案。对基于自适应网格的多目标粒子群算法的设计主要包括：对等式约束和不等式约束的处理；自适应网格法对Pareto外部档案的维护；粒子个体最优位置和全局最优位置的选取。文中通过实例计算,将实验结果与基于遗传算法的多目标优化算法比较,验证了该方法的有效性。文中对多目标算法产生的Pareto解集的多属性决策,首先采用客观赋权的信息熵法对经济和环境两个属性进行权值计算,然后用逼近理想解的排序方法(TOPSIS)对Pareto解集给出排序,得到最满意的解。本文还将上述机组负荷优化分配方法应用到某火电厂生产运营管理系统中,搭建了基于Flex+Spring+Hibernate的系统框架,设计了简洁美观的界面,实现了机组负荷基础数据的管理、机组煤耗特性曲线的拟合、机组单目标负荷优化分配和多目标优化分配,提高了企业的管理和生产效率,为企业带来了经济效益。
[Abstract]:With the deepening of China's electric power system reform, the implementation of the mechanism of "separation of power plants from power plants and power grids, bidding for the Internet" has transformed power generation enterprises from production-oriented enterprises to profit-oriented enterprises, and power plants, as independent economic entities, have participated in market competition. The power network no longer issues the load instruction to the unit unit, but issues the total load instruction to all the units in the power plant. Therefore, the scientific and reasonable distribution of the load of each unit will help to optimize the operation level of the unit. It will help to reduce the coal consumption of the whole plant, and provide the scientific basis for the power plants to compete for electricity in the electricity market. At the same time, the national policies on energy saving and emission reduction and sustainable development of enterprises, Thermal power plants will produce a large number of harmful gases in the process of generating electricity, which will cause serious pollution to the environment, and pollutant emission has become a factor that can not be ignored in the load distribution of thermal power plants. It is of great practical significance to study the optimal load distribution of power plant units. This paper introduces the basic concept of unit load optimal distribution, according to the research status of unit load distribution at home and abroad. Combined with the actual operation of thermal power plant, the coal consumption of power supply is selected as the economic index of load distribution, and the characteristic curve of power supply coal consumption of unit is the basis of optimal load distribution of power plant. The coal consumption characteristic curve of the unit is fitted by the least square method. In the study of the single objective load distribution algorithm, a single objective load optimal distribution mathematical model based on economy is established by using the coal consumption characteristic curve of the unit. Then, this paper introduces in detail the methods of load optimal distribution using the equal differential rate method and dynamic programming method, and compares their characteristics, and chooses dynamic programming method as the single objective load allocation algorithm. The algorithm is designed and explained in detail. The results show that the dynamic programming method can optimize the operation of the whole plant to a certain extent, thus reducing the coal consumption of the whole plant. On the basis of single objective optimization model and considering emission reduction target, a multi-objective load distribution model based on environment and economy is established in this paper. A multi-objective particle swarm optimization algorithm based on adaptive mesh is designed, and the Pareto solution set obtained by the algorithm is used for multi-attribute decision making. The design of multi-objective particle swarm optimization algorithm based on adaptive mesh mainly includes the processing of equal constraints and inequality constraints, the maintenance of Pareto external files by adaptive mesh method. In this paper, the experimental results are compared with the multi-objective optimization algorithm based on genetic algorithm. The effectiveness of this method is verified. In this paper, for the multi-attribute decision of the Pareto solution set generated by the multi-objective algorithm, the weights of the economic and environmental attributes are calculated by the objective weighted information entropy method. Then the Pareto solution set is sorted by using the ranking method of approximating the ideal solution, and the most satisfactory solution is obtained. In this paper, the optimal load distribution method of the unit is also applied to the production and operation management system of a thermal power plant. The system frame based on Flex Spring Hibernate is built, and the simple and beautiful interface is designed. The management of basic data of unit load, the fitting of unit coal consumption characteristic curve, the optimization of unit single objective load distribution and multi-objective optimization distribution are realized. It improves the management and production efficiency of the enterprise and brings economic benefits to the enterprise.
【学位授予单位】：东北大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM621

【参考文献】