机组组合的改进自学习粒子群算法  

Unit Commitment Model Based on Self-learning Particle Swarm Algorithm

作　　者：谢胤喆[1] 于汀[2] 陈海良[1] 赵舫[1] 郭瑞鹏[1] 蒋雪冬[1] 

机构地区：[1]浙江大学电气工程学院,杭州310027 [2]中国电力科学研究院,北京100192

出　　处：《电力系统及其自动化学报》2014年第2期14-20,共7页Proceedings of the CSU-EPSA

基　　金：国家科技支撑计划项目(211BAA07B03);国家高技术研究发展计划(863)计划资助项目(2011AA05A118)

摘　　要：安全约束机组组合是混合整数规划问题,找到高效稳定求解此问题的算法很重要。文中提出了一种新型的离散粒子群求解机组组合问题,通过松弛模型辨识出机组中必开必停的情况,减少离散变量数目,并结合机组组合问题的特性提出了对应的改进自学习策略,能较好地解决含安全约束的机组组合问题。此外,给出了一种初始粒子群生成策略,提高粒子质量。以IEEE30和IEEE118两个标准节点系统为测试算例,通过与传统算法和商业软件包CPLEX的数据对比发现此算法能较快找到最优解或次优解,效率高计算结果稳定,证明该方法可行高效。The unit commitment has commonly been formulated as a mixed-integer,nonlinear optimization problem.To find an efficient and stable method to solve this problem is important.A novel discrete particle swarm optimization to solve unit commitment was proposed in this paper.A novel identification method for the integer variables was proposed to reduce the dimensions.Besides,according to the characteristics of unit commitment and the security constraints,an improved self-learning strategy based on novel particle swarm optimization was proposed.This method can solve security constrained unit commitment well.In addition,a method to produce initial particles in the feasible region was proposed in order to improve the quality of the solution.The feasibility and effectiveness of the proposed method are demonstrated by two test systems of IEEE30 and IEEE118,and the computational results are compared with the custom benders decomposition and the commercial software CPLEX.The result shows that this method can find the optimum or suboptimum solution quickly,which proves the feasibility and validity of this method.

关 键 词：机组状态 网络安全约束 整数变量辨识 粒子群算法 自学习 

分 类 号：TM713[电气工程—电力系统及自动化]