【作者】 王淑芬；

【导师】 万仲平；

【作者基本信息】 武汉大学 ， 计算数学， 2005， 硕士

【摘要】 电力系统中,各级电网的电压水平互不相同,他们为了自身的利益在自己的控制范围内寻求最优网损和电压稳定水平。然而,这种只顾局部利益而忽略整体利益的做法,往往会导致整个电网的电压水平不稳定,网损也不是最优,甚至造成电网瓦解事故。 二层规划是一种具有二层递阶结构的决策优化问题。上层和下层各有目标函数和约束条件,上层问题的目标函数和约束条件,不仅与上层决策变量有关,而且还依赖于下层问题的最优解或最优值。下层问题的最优解又受上层决策变量的影响,其解(或最优值)反馈到上层而影响上层规划问题的最优解。调度中心和各级电网的关系就成了一个典型的二层决策问题。 在综合考虑电力系统的实际问题和二层规划理论的基础上,提出了基于二层规划理论的无功优化模型。对模型的求解考虑无功优化多约束、非线性、连续变量和离散变量混合、组合优化的特点,采用不要求目标函数的可微,变量的连续,可以方便的处理离散变量的遗传算法和全局收敛性好的模拟退火算法对模型进行了求解。对IEEE-6节点系统和IEEE-30节点系统的数值实验结果表明了此模型和算法的可行性。更多还原

【Abstract】 In the electric power system, each network has different voltage level, some local lines seek to the optimal plosses and stability voltage level for themselves profits in their controlled fields, which blindly adjusted without considering global profit and results not only the instability voltage level of the whole system but also not the best plosses, even the whole networks ellapse accidents.The bilevel programming is an optimization problem which has bilevel hierarchy decision .The first level and the lower level both have their objective functions and constraint conditions, the objective function and constraint conditions of the first level not only have relations with the upper decision variable, but also depends on the optimal value(or the optimal solution) of the lower level. The lower optimal solution is also influenced by the upper decision variable, which is feed back to the upper level, furthermore influence the upper programming problem’s optimal solution. The relations between the dispatch center and each networks is a canonical two-level decision problemConsidering the electric power system’s factual problem and the theory of the two-level programming, we present the reactive optimization model on the basis of the two-bilevel optimization theory. Because the reactive optimization is a problem that has character multiple constraint、 non-liner, continuous variable and discrete variable、 combinatorial optimization, the genetic algorithm which has no messages of the objective functions’ differential and continuous variable, can simply solve the continuous variable and discrete variable ,and the good global convergence simulated annealing algorithm are applied to solve the model. Testing results of applying the algorithm to an IEEE-6 bus system and an IEEE-30 bus system shows feasibility and efficiency of algorithm.更多还原