摘要
电力市场下发电机组检修计划的安排不当是造成近几年大停电的原因之一。电力市场机组检修计划需要兼顾可靠性与经济性两个方面。采用Benders分解法可以将这个高维度、非线性、混合整数随机规划问题分解为主问题和子问题求解:主问题是一个多目标整数规划问题,而子问题则是一个非线性随机问题。针对Benders分解法求解效率不高的问题,求解主问题时利用了机组检修连续性的特点,对Balas隐枚举法中的前向搜索部分和回溯部分进行了改进;求解子问题时,利用了半不变量法的偏导数解析表达式来求对偶乘子。对IEEE-RBTS和IEEE—RTS系统中所有发电机组安排了全年的检修计划,测试表明所提出的算法快速、有效。
The unsuitable generator maintenance scheduling under the electricity market environment is one of the important factors that causes the power system blackout, while the proper one should be designed to stress both the reliability and economies. The Benders decomposition method can divide this large scale, non-linear, mixed-integer stochastic programming problem into two problems: a deterministic multi-objective integer programming master problem and a stochastic, non-linear operation sub-problem. Several methods are used to increase the efficiency of the Benders decomposition method. The continuity of generator maintenance is used to modify the forward detection part and the backup trace part of the implicit enumeration method to solve the master problem. The analytic method based on the Cumulants method is utilized to calculate (he dual multipliers of Benders cut. The proposed method is applied to make maintenance scheduling for all the generators through 52 weeks in the IEEE-RBTS and IEEE-RTS system. The test results show that this algorithm is effective and fast.
出处
《电力系统自动化》
EI
CSCD
北大核心
2004年第16期41-44,78,共5页
Automation of Electric Power Systems
基金
国家自然科学基金资助项目(50107005)。
