大规模电力系统机组组合问题的近似动态规划模型与算法

发布时间：2018-07-01 12:27

本文选题：安全约束 + 截断技术　；参考：《广西大学》2014年博士论文

【摘要】：随着智能电力调度系统的建设与发展,电网运行部门对机组组合计算精度和计算速度的要求越来越高。同时,由于电网互联规模不断扩大,使机组组合问题的求解愈加复杂和困难。因此,研究探索适合于求解大规模电力系统机组组合问题的理论和方法,对提高发电能源利用率、节能降耗具有重要的现实意义和经济价值。本文涉及大规模电力系统机组组合问题的近似动态规划模型与算法。针对动态规划求解大规模电力系统机组组合的“维数灾”问题,首次将数学规划中的最新成果——近似动态规划理论应用于电力系统的机组组合问题。分别对近似动态规划在日前常规机组组合问题和安全约束机组组合问题中的应用,开展了较为深入细致的研究工作,尝试提出一种可快速求解大规模机组组合问题,且易于考虑各种电力系统安全运行条件的近似动态规划方法。首先探讨近似动态规划解决“维数灾”问题的数学原理,分析了四种近似值函数模型及现有的函数近似方法,较详细介绍了近似值迭代与近似策略迭代两种近似迭代算法的原理。在结合机组组合问题对近似动态规划7个基本概念进行定义的基础上,通过引入观测成本函数,给出了决策函数、立即成本和近似值函数的具体公式,构建了采用决策后状态变量的机组组合近似值函数。采用所提机组组合近似值函数,在IBM兼容PC机上用Matlab-2013a编程,对10～100机24时段6个系统的实例进行了计算,首次比较了近似值迭代和近似策略迭代两种近似迭代算法求解机组组合问题的效果,指出了近似策略迭代算法对机组组合问题的适用性。针对直接套用近似策略迭代算法求解机组组合问题运算时间偏大的问题,根据电力系统实际运行特征,提出了可快速求解大规模系统机组组合问题的策略迭代近似动态规划法。首先结合机组组合问题对近似策略迭代算法进行改进,提出了适合于机组组合问题的近似策略迭代算法,更新所提机组组合近似值函数。本文给出了该算法的具体流程和实现细节,对相邻时段近似值函数的更新、机组出力爬升约束的处理和压缩状态空间的传统序列截断技术未考虑机组最小启停时间特性等问题进行了重点研究,提出了每次用一个预决策状态更新决策后状态近似值函数、采用动态的机组出力两界约束出力爬升约束以及扩展的序列截断技术等具体解决方案。通过10～1000机96时段大规模系统的计算,验证了所提方法的正确性和实用性。结合上述对近似动态规划理论的研究结果及其在日前常规机组组合问题上的具体应用,将所提策略迭代近似动态规划法扩展到了安全约束机组组合问题。针对潮流方程、线路潮流约束和节点电压限制等约束条件使问题的求解难度和计算量显著增加的问题,在不改变近似策略迭代算法的前提下,提出了通过调整立即成本的计算处理潮流约束的方法。通过IEEE-30～300节点系统和波兰2737节点系统的实例,验证了所提方法求解安全约束机组组合问题的正确性和实用性。本文研究成果是国家973项目(2013CB228205)第5课题“特性各异电源及负荷的能量互补协同优化调控”成果之一,在该项目与国家自然科学基金项目(51167001)共同资助下完成。研究成果为制定大规模电力系统节能发电调度计划,提供了一种可靠方案与技术支撑；不仅在电力系统的智能优化调度方向具有重要应用价值,还为近似动态规划理论的发展开拓了新领域。
[Abstract]:With the construction and development of the intelligent power dispatching system, the requirements for the calculation precision and speed of the unit combination are getting higher and higher. At the same time, the solving of the unit combination problem is more complex and difficult because of the continuous expansion of the interconnection scale of the power grid. Therefore, the research and exploration is suitable for solving the unit assembly of large scale power system. The theory and method of the problem have important practical significance and economic value for improving the utilization of power generation energy and saving energy and reducing consumption. This paper deals with the approximate dynamic programming model and algorithm of the unit combination problem of large-scale power system. The latest achievement of the approximate dynamic programming theory is applied to the unit combination problem of power system. The application of approximate dynamic programming to the problem of daily conventional unit combination and the combination of safety constraint unit is carried out in a more thorough and meticulous study, and an attempt is made to quickly solve the problem of large scale unit combination. The approximate dynamic programming method for the safe operating conditions of various power systems is easily considered. First, the mathematical principle of the approximate dynamic programming to solve the "dimension disaster" problem is discussed. Four approximate value function models and the existing approximation methods are analyzed. Two approximate iterative calculations are introduced in detail. On the basis of defining the 7 basic concepts of approximate dynamic programming in combination with the unit combination problem, by introducing the observation cost function, the specific formula of the decision function, the immediate cost and the approximate value function are given, and the approximate value function of the unit combination with the state variable after the decision is constructed. The value function is programmed by Matlab-2013a on the IBM compatible PC machine, and the examples of 6 systems in the 10~100 machine and 24 period are calculated. The results of the approximate iterative and approximate iterative iterative algorithm for solving the unit combination problem are compared for the first time, and the applicability of the approximate strategy iteration algorithm to the unit combination problem is pointed out. The approximate strategy iterative algorithm is applied to solve the problem of large operation time of the unit combination problem. According to the actual operation characteristics of the power system, a strategy iterative approximate dynamic programming method is proposed to quickly solve the unit assembly problem of large scale system. The approximate strategy iteration algorithm of the unit combination problem is used to update the approximate value function of the proposed unit combination. This paper gives the detailed process and implementation details of the algorithm, updates the approximate value function of the adjacent period, the processing of the unit output climbing constraint and the traditional sequence truncation of the compressed state space without considering the minimum start and stop time of the unit. The characteristics and other problems are emphatically studied, and some specific solutions are proposed, such as updating the state approximation function after decision making with a pre decision state, using the dynamic force two bound constraints and the extended sequence truncation technology. The proposed method is verified through the calculation of the large-scale system in the 10~1000 machine and 96 period. In combination with the results of the above approximate dynamic programming theory and its application on the problem of the daily conventional unit combination, the proposed iterative approximate dynamic programming method is extended to the problem of the safety constraint unit combination. The constraints of the power flow equation, the line flow constraint and the node voltage limit are made. The problem is difficult to solve and the amount of calculation is increased significantly. On the premise of not changing the approximate strategy iterative algorithm, a method to deal with the current constraints is proposed by adjusting the calculation of the immediate cost. Through the example of the IEEE-30 300 node system and the Poland 2737 node system, the proposed method is proved to solve the problem of the safety constraint unit combination problem. The results of this study are one of the results of the national 973 Project (2013CB228205) fifth project, "energy complementation and coordinated optimization control of different power and load of different characteristics", completed jointly by the project and the National Natural Science Foundation (51167001). The research results are for the formulation of large-scale power system energy saving power generation. The degree plan provides a reliable scheme and technical support, which not only has an important application value in the direction of intelligent optimal dispatching of power systems, but also opens up a new field for the development of approximate dynamic programming theory.
【学位授予单位】：广西大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TM73

【参考文献】