近年來受到全球氣候變遷的影響，溫室效應促使氣候急遽變化，使得災害發生頻率和強度逐年上升。天然災害有著不可預測性且造成的破壞程度難以估計，然而災害發生後會因時間性及地域性來衍生出其他災難浩劫。隨著政府著手開放災害資料，於過往的災害歷史資料裡尋找經常發生災害的關聯序列，將可以幫助民眾對於二次災害的發生提前做準備。傳統上對於避難路徑的設計，採用災害潛勢地圖和使用問卷調查當地人的用路習慣來規劃事前最短逃生路徑，但是用事前規劃出的最短逃生路徑並未考量到災難發生後引發的道路損毀和二次災害的發生可能造成的損傷，而且傳統上對於單一災難目標規劃路徑已經不敷使用。 為了於災害發生時提升疏散的速度，且迅速地提供民眾前往安全的逃生避難路徑，和有效降低二次災害所造成的傷亡。本論文針對複合性災害進行分析，由於災害存在時序性的特徵，運用PrefixSpan演算法，以更具有系統性的方法尋找頻繁災害序列。本論文以A Star演算法做為本研究路徑演算法之求解模式，並與路網資訊和災害頻繁序列做結合，透過本研究所設計的時序複合性災害資料之避難路徑規劃系統可以預先避開疑似危險路段，規劃出安全的逃生避難路徑，協助使用者爭取更多避難和救災的黃金時間。 ;Climate change is already affecting the planet in recent years. The greenhouse effect triggers such drastic climate changes which causes the frequency and intensity of disasters increasingly year by year. Although natural disasters are unpredictable and could result in massive damages, it will derive other multiple disasters depend on its time relation and regional characteristic. As the government gradually promote open disaster data, discovering frequent disaster-related sequences from disasters historical data can assist victims to get ready for secondary disasters beforehand. The traditional evacuation path uses disaster potential maps and survey the evacuation habits of the locals to plan shortest evacuation path before disasters happened. However, the pre-planned shortest path considers neither the consequences after the disasters which could possibly destroy the road, nor the damages caused by secondary disasters. Besides, single disaster-based processing is not enough for composited disaster events. In order to accelerate the evacuation speed, provide a safer evacuation path to victims, and against the damages caused by secondary disasters when disasters occur. This research proposes to adopt a PrefixSpan method with composited disaster data and systematically search frequent disaster-related sequences. Moreover, the path finding is based on A-Star method which can effectively find a shortest path. This research also combines path finding with GIS information and frequent disaster-related sequences. In the end, this research can previously avoid the potential dangerous area and provide a safer route to user for suggestion, so that they can get more rescue and evacuation time.
