基于Hopfield神经网络的多阶段配电变电站的规划优化

A New Distribution Substation Planning Algorithm Based on Hopfield Neural Network

提出了一种城市电网规划中多阶段变电站规划优化的新算法。该算法先用贪心法快速求解目标年新建变电站的座数和各变电站(包括已有变电站)的目标年容量,再利用Hopfield神经网络算法计算新建变电站的位置和各变电站(包括已有变电站)在各规划阶段的供电范围,最后确定各变电站在各阶段的真实容量及投建计划。在求解过程中,该方法考虑了已有变电站的改造问题。从全局最优的原则出发,可求得具有实际价值的最优解或近似最优解。该方法在求解变电站供电范围时无需对数据进行归一化处理,且易于编程。该方法可为变电站规划提供一种新的思路。

This paper proposes an algorithm for solving a problem of multi-stage distribution substations expansion. In the algorithm, Hopfield neural network is first applied to make the solution accurate. The algorithm uses greedy algorithm to work out the number of new substations and the size of each substation (including the existing substation) in aim year. After the size of each substation is obtained, it uses Hopfield neutral network to determine the service region of each substation and the locations of the new substations. Because the solution is obtained by resolving a group of differential equations, the computation time is much shorter than many existing algorithms. The advantage of the new algorithm is that it can involve the existing substation into optimization and need not to deal with original data. Another advantage is that the algorithm can be programmed easily. The validity of the proposed algorithm is demonstrated by a numerical example.