发电机组组合的混合蚁群优化算法  

Hybrid ant colony optimization algorithm for generation unit commitment problem

作　　者：王剑[1] 刘天琪[1] 

机构地区：[1]四川大学电气信息学院,四川成都610065

出　　处：《电力系统保护与控制》2010年第20期85-89,95,共6页Power System Protection and Control

摘　　要：电力系统中的机组组合作为一个非确定型多项式困难问题,一直难以获得其理论最优解。针对算法的精度和速度这一矛盾,提出了一种结合启发式算法和蚁群算法的混合优化算法。用优先级排序法获得次优解,并在附近形成一个搜索邻域;采用蚁群算法在此邻域内寻优,减小了蚁群算法的空间复杂度。同时,在蚁群算法中引入了人工鱼群算法的拥挤度概念。拥挤度阈值在迭代过程中是自适应变化的,从而增强了算法的遍历寻优能力,也保持了较快的收敛速度。经济负荷分配采用简化梯度法。对一个10机系统算例仿真计算,验证了所提算法对解决机组组合问题具有很强的搜索能力和快速收剑性。As a hard nondeterministic polynomia（lNP） problem in power system,the unit commitment problem is difficult to find its optimal solution theoretically.This paper presents a hybrid algorithm based on heuristic method and ant colony optimization（ACO）for the contradiction between precision and speed of the optimization algorithm.Firstly,this paper attains the sub-optimal solution by priority list（PL）method,and forms a local area for search.Then,it completes the optimization process in the local area by ant colony algorithm,which reduces the space complexity of the ant colony optimization.Meanwhile,a swarm degree in the artificial fish school algorithm（AFSA）is used in the ant colony optimization.The threshold of a swarm degree varies self-adaptively in the process of iteration,so it enhances the searching ability of the algorithm,and also insures the algorithm to have a quick convergence rate.The simplified gradient method is used for the economical load dispatch problem.Simulation results of 10-units system prove that the algorithm proposed has good search capability and quick convergence rate to solve the unit commitment problem.

关 键 词：电力系统 机组组合 优先级排序 混合蚁群算法 人工鱼群算法 经济负荷分配 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]