最佳權重法應用於交通資料融合

Abstract

先進旅行者資訊系統(ATIS)為智慧型運輸系統(ITS)發展的重點之一，其路況資訊可由許多來源獲得，如偵測器、探針車等，每個來源能提供的資料內容不同，亦有各自的使用範圍及限制，因此不同來源的路況資料必須透過資料融合(Data Fusion)的方法來處理，以得到一個較可靠、較精確的交通資訊。 本研究以最佳權重法來融合多種異質的交通資料，最佳權重法是將系統總不確定性最小化，使得融合後的資料有最低的不確定性，模式中以Shannon熵表示不確定性，並提出距離權重法來改善明確分類方式所造成的偏誤。本研究以電腦模擬的方式，來評估模式的適用性，測式結果顯示以距離權重法來分類資料，可降低分類邊界的影響，也可改善資料過度集中或分散所造成的偏誤。

Advanced Traveler Information System (ATIS) is one of the key elements of Intelligent Transportation System (ITS). Traveler information could be obtained via various sources such as VD and probe vehicles. Therefore, various information must be merged into a distinct and reliable information. The purpose of this research is developing a date fusion methodology for merging various traffic data source. The optimal weighting approach is proposed for fusing various traffic data. It minimizes total system uncertainty while fusing data. We use Shannon entropy to represent uncertainty. Besides, a weighted distance approach is used to reduce the impact of data classification. We use simulation data to evaluate the performance of our model. The results show that our proposed approach could reduce influence of data classification and reduce the bias of centralized and separated data.