改进蚁群算法和生成树相结合的中压配电网架规划  

Medium-voltage distribution network planning based on improved ant colony optimization integrated with spanning tree

作　　者：梁英[1] 关洪浩[2] 

机构地区：[1]中国电力科学研究院,北京100192 [2]中国农业大学信息与电气工程学院,北京100083

出　　处：《农业工程学报》2013年第A01期143-148,共6页Transactions of the Chinese Society of Agricultural Engineering

摘　　要：由于涉及许多变量和约束,中压配电网规划是一个非常复杂的大规模组合优化问题。蚁群算法是一种具有正反馈特性的贪婪性、分布式的现代启发式搜索方法,适合于路径的寻优,但是易陷入局部最优。该文将蚁群算法和生成树算法结合起来,用于带有交叉点的中压配电网网架规划。针对基本蚁群算法易于陷入局部最优的问题,提出了动态设定待选路径信息素阀值和动态调整路径选择策略的方法,以便提高蚁群算法的全局搜索能力。在考虑中压配电网辐射状和连通性约束时,提出了带有交叉点的生成树方法,大大减少了在规划中不可行解的产生。算例仿真结果表明采用该方法求解中压配电网架规划方案是有效的。At present there are a lot of literatures referring to transmission network planning, but less about medium-voltage distribution network planning, especially for the network with cross-points of line corridors. Ant colony optimization （ACO） is a heuristic algorithm with positive feedback, distributed computation and greedy characteristic. It is very suitable to search optimal path in a graph but prone to fall into local optimum. This paper integrates ACO with spanning tree algorithm to solve medium-voltage distribution network planning with cross-points. The model of medium-voltage distribution network planning is established, which takes the minimum investment cost of expansion lines and minimum energy loss as objective functions, power flow balance as equality constraints, and maximum line current, upper and lower limit of node voltage as inequality constraints. Meanwhile,the network topology graph is radial and connected. The dynamic pheromone threshold of candidate paths and dynamic adjustment of path selection strategy are introduced in order to reduce the possibility of falling into local optimum in general ACO. According to the maximum iteration number, the current iteration number and the maximum pheromone value of all paths in the current iteration, the dynamic pheromone threshold of the candidate paths is given,which can ensure that the difference between the pheromones of the paths is as small as possible to increase the diversity of solutions at the early search stage, and the difference between the pheromones of the paths is as large as possible to speed up the algorithm convergence at the later search stage. The probability to be selected of one candidate path is decided by its length,investment and pheromone quantity. The shorter length, the smaller investment and the more pheromone can make the greater opportunity to be chosen. The dynamic adjustment of path selection strategy can help the search process tend to minimal objective function. Considering the radiation and connectivity constraints

关 键 词：配电网 规划 蚁群算法 生成树 交叉点 辐射型网络 

分 类 号：TM715[电气工程—电力系统及自动化]