Nowadays, there is a growing need to analyse systems using complex networks and graphs, especially in critical infrastructures. That includes transmission and distribution systems, where a single fault may cause power interruption for several consumers. A special approach to this problem uses Optimal Transmission Switching (OTS), where edges are comutated to change the network topology, and improves fault response. Because of its computational complexity, heuristics are proposed to the problem. This paper aims to introduce Variable Neighborhood Descent (VND) to the OTS problem, because of its local search feature, as well as the ability to deal with local minimuns. For that, the neighborhood structures and objetive function were adapted to address the peculiarities of the electrical grids, and a power redistribution algorithm was implemented. Failures and attacks were simulated, and the overload reduction was compared between the original topology and the one found by the VND (by line-switching). For power overload failures, results were better in intermediate overload levels, for both topologies. For node removal, best results were found in scale-free graphs, especially in intentional attacks, which shows that the local search phase, presented in VND, works well in a subset of edges limited to the proximity of the failure, especially with networks that have hubs. The computacional time shows the potential of the heuristic to be used in real time analysis.

中文翻译：

---

电网的可变邻域下降法作为减少过载的方法


如今，越来越需要使用复杂的网络和图形来分析系统，特别是在关键基础设施中。这包括输电和配电系统，其中一个故障可能会导致多个消费者的电力中断。解决此问题的一种特殊方法是使用最佳传输切换 (OTS)，通过交换边缘来改变网络拓扑，并提高故障响应。由于其计算复杂性，针对该问题提出了启发式方法。本文旨在将可变邻域下降（VND）引入 OTS 问题，因为它具有局部搜索特性以及处理局部极小值的能力。为此，对邻域结构和目标函数进行了调整，以解决电网的特殊性，并实施了功率重新分配算法。模拟了故障和攻击，并对原始拓扑和 VND（通过线路交换）发现的拓扑之间的过载减少进行了比较。对于电源过载故障，两种拓扑在中间过载级别的结果都更好。对于节点删除，在无标度图中找到了最佳结果，尤其是在故意攻击中，这表明 VND 中提供的本地搜索阶段在仅限于故障附近的边子集中效果良好，尤其是对于以下网络：有枢纽。计算时间显示了启发式方法在实时分析中的潜力。