Master Thesis

Infrastructure Interdependency Simulation Using Agent-Based Modeling

Abstract

Infrastructure is important to our lives. Most infrastructure facilities consist of nodes and edges of high dependency. These kinds of facilities are considered as traditional complex systems. The research of using computer technologies to analyze such systems has grown recently. We can classify recent research into two categories: 1. Dependency analysis, and 2. System dynamic analysis. The former focuses on computation methods such as topology analysis to find vulnerable nodes while the latter focuses on large-scale simulation to find potential vulnerable infrastructure facilities. Common dependency analysis uses topology analysis to find vulnerable nodes and apply the Monte-Carlo method to their model. However, it merely applies an infrastructure’s parameters and rules to their model. In this thesis, our contributions are on applying such parameters and rules to our model by taking Taiwan Power Company as an example. On this basis, we use agent-based modeling to simulate the context.  We used the Push-Relabel algorithm to dispatch the power flow. And we used Repast Simphony as a tool of agent-based modeling. We established different situations to simulate different disasters such as earthquake or tsunami, and to provide a chance estimate and reduce the damages of such a disaster in advance. Besides, we also used indices of an IEEE standard to evaluate our result. In other words, we can find the vulnerable nodes or potential threat in a power grid with agent-based simulation. Besides, we provided various user interfaces for users to observe the information of the power grid more easily and efficiently. Our user interfaces can dynamically present the change of information on the power grid, and the result can be saved and loaded for future uses.

以代理人基方法之基礎設施互依性模擬

摘要

基礎設施是民生不可或缺的元素之一。重要基礎設施通常由許多節點構成，也因為彼此之間相依性極高，所以也被視為一種典型的複雜系統。近年來以資訊科技輔助分析這類重點基礎設施的研究也越來越多，從文獻上來看主要可以分為兩大類：(1)互依性分析(2)系統分析。前者著重於使用像拓樸分析這類數值分析的方式，找出弱點基礎設施；而後者則是使用大量模擬的方式來找尋潛在的弱點基礎設施，並模擬不同情境的危機情境。常見的基礎設施分析，多以拓樸分析為基礎找出弱的節點，並套用蒙地卡羅演算法到模型內，做為模擬的依據，但較少見到以基礎設施的各項屬性和規則當作模型的模擬。因此，本篇論文的貢獻是希望以台灣電力公司之電力基礎設施運轉規章、基礎設施的屬性等這些資料為例，當作建立模型的基礎，將Push-Relabel演算法的步驟分散至各代理人上，作為電力調度平衡的計算方式，並採用Repast Simphony作為我們代理人基模擬的工具。藉由操作這些節點的狀態，我們設計了七個不同情境的模擬，並以IEEE指標做為結果的評估方式，藉以說明我們可以透過代理人基模擬找出基礎設施中的弱點，並能夠建立不同的情境模擬各式的危機和災難，進而提供預防的機會。此外，我們的模擬工具亦提供不同的視覺化呈現結果，讓使用者能夠簡單的看出模擬過程的變化，並能夠將結果儲存並重現。

Demo

Note:
The source code is not provided due to the embed classified data. You are welcome to discuss Repast Simphony with me.

本論文涉及敏感資料故無法提供原始碼下載，如果有 Repast Simphony 相關問題歡迎寫信和我討論。