机组组合问题的两阶段优化算法  

Two-phase optimization approach to unit commitment problems

作　　者：葛晓琳[1] 张粒子[1] 王楠[1] 

机构地区：[1]华北电力大学电气与电子工程学院,北京102206

出　　处：《中国电力》2010年第4期14-18,共5页Electric Power

摘　　要：针对电力系统机组组合问题(UC)高维、非凸、非线性的特点,提出了一种两阶段优化方法(LR-DE)。利用拉格朗日松弛算法(LR)对UC问题进行解耦,将多机优化问题转化为单机优化问题的重复计算,使模型简化,利用动态规划法和次梯度法求出对偶解对应的拉格朗日乘子;根据对偶解信息设定拉格朗日乘子更新空间,并利用微分进化算法(DE)进行搜索,全面考虑所有约束条件,不断缩小对偶间隙,求出最优的机组组合状态。算例分析表明,该算法优化效果好,搜索能力强,能较好解决大规模机组组合优化问题。A two-phase optimization method （LR-DE） was presented for power system unit commitment （UC）, a high dimensional, non-convex, nonlinear problem. First the problem was decoupled by Lagrangian Relaxation algorithm, the optimization of multi-machine was changed into double-counting of single optimization to simply the model, using dynamic programming method and sub-gradient method to derive the dual solution Lagrange multiplier; Secondly, the space of updating Lagrange multipliers was determined by optimal dual solution, searched by Differential Evolution Algorithm （DE） with all constraints considered, the duality gap will be narrowed continually and the optimal unit commitment will be obtained. Analysis of examples shows that the algorithm can get better solutions, has the comprehensive ability to search, which is very prospective for large-scale unit commitment problem.

关 键 词：电力系统 机组组合 拉格朗日 微分进化 对偶间隙 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]