配电网络重构中的智能优化算法

Intelligence Optimization Algorithm in Distribution Network Reconfiguration

详细介绍了配网重构问题中的各种智能优化算法,分析了各种算法的优缺点:SA算法一般可以得到与初始解基本无关的全局最优解或次最优解,但收敛速度较慢,其收敛性依赖于退火方案的选取;GA不受目标函数连续性、可导性等约束条件的限制,在解的空间多点并行搜索,有较高的搜索效率,但收敛性能较差,容易出现早熟收敛;TS算法灵活的记忆功能,避免了搜索时陷入局部最优,但对初始值有较强的依赖性,搜索过程是单对单的串行操作,而非并行操作;ANN优化算法不需要复杂的潮流计算,大大节省了配网重构的时间,但优化结果依赖于提供的训练样本。

Modem intelligence optimization methods for distribution network recorfiguration have been introduced in detail. Their strong and weak points have been discussed. Simulated annealing algorithm can work out optimal or near optimal solution independent of initial solution, however, they are time - consuming and sensitive to parameters of algorithms. Genetic algorithm cannot be constrained by the features d the object function and the performances of search are efficient owing to parallel operation, but they sometimes converge to local optimal solution. Tabu search algorithm can avoid local optimal solution for their flexible memory function, however, they are sensitive to initial solution and the search operations are point- to- point. The algorithm based on artificial neural network can find out optimal solution quickly owing to no calculation of flow, but the results depend on the training sets.