Graph-based vulnerability analyses on power grid and associated protection strategies against cascading failures

於歷史事件中我們知道電力網路中些許的傷害會導致大停電事件的發生，而電力網路的安全是學者們一直以來關心的議題，複雜網路發展至今，已有不少研究人員利用複雜網路的方法來分析電力網路，本論文即利用複雜網路的分析方法，分析台灣電力網路脆弱性，並提出一新穎的保護策略。 文章中引用三種分析方法分析台灣電力網路的脆弱性，第一為根據網路效率(efficiency of complex network)，第二為根據靜態負載的連鎖性失效(static load cascading failure)，第三為根據動態負載的連鎖性失效(dynamic load cascading failure)，三種脆弱性分析的方法得到結果的交集處，在文中被認為是電力網路中最具風險的變電站。根據靜態負載的連鎖性失效的結果，我們在不改變網路結構和整體網路的負載量之下，提出一新穎的安全策略，試圖降低靜態負載時連鎖性失效帶來的傷害，並分析此安全策應用至ER隨機網路(ER random network)、BA無尺度網路(BA scale-free network)和台灣電力網路，且量化其結果。 三種脆弱分析的結果皆有共通的部份，表示台灣電力網路在拓樸分析下並不堅固，其結果顯示此安全策略是有效的。本篇論文分析的網路均是無權重也無方向性的網路，如此，未來在網路脆弱性與安全策略的研究還有進一步的研究空間。From blackout events in history, we know the damage of a few substations or transmission lines can lead to a big blackout, and the robustness of power grids are always a great concern. Recently the topological analyses of power grid network have developed rapidly and its achievements have become a center of attention. This thesis aims to investigate the vulnerability of Taiwan’s power grid through topological analyses and propose a novel protection strategy. This thesis introduces three methods to investigate the vulnerability of Taiwan’s power grid. The first method is based on efficiency of complex network. The second method is based on static load cascading failure. The third method is based on dynamic load cascading failure. The common results of the three results are considered the high risks in Taiwan’s power grid. According to the static load cascading failure, we propose a protection strategy against cascading failure without changing the structure of network and the whole load of the network, and then demonstrate the effectiveness of this strategy on an ER random network, a BA scale-free network and Taiwan’s power grid numerically. The three methods having the common results, one may say Taiwan’s power grid isn’t robust under the topological analyses, and our protection strategy may be useful for decreasing the damage after cascading failure