本研究結合進階卡門濾波與格位傳送模式,建構遞迴動態O-D矩陣推估演算法,藉由該演算法模擬在不同交通情況下車輛到達型態之交通行為,並預測各依時O-D起迄對之到達型態,以便推估動態O-D矩陣。為驗證本演算法,本研究以6個O-D起迄對之小路網為範例,每6秒為1時階,推估90分鐘的O-D矩陣,再與Greenshields巨觀模式預測車輛旅行時間及假設進入車輛會於兩時階範圍內到達迄點之演算法進行比較。結果顯示本模式推估結果之RMSE遠較Greenshields巨觀模式為低。此外,本研究另以國道1號楊梅至泰山收費站間及臺中至臺北間進行實例應用,結果顯示本模式之RMSE均在可接受範圍,說明本演算法的有效性與實用價值。
This study proposes an iterative dynamic O-D matrices estimation algorithm to effectively capture the traffic behaviors and their arrival distributions under various traffic conditions. The core logic of the proposed algorithm is to combine extended Kalman filtering with cell transmission model to simulate traffic movement behaviors so as to predict the arrival distributions of all O-D pair traffic in various time intervals, and then to estimate dynamic O-D matrices. To validate the performance of the proposed algorithm, a small-scale corridor example with six O-D pairs is tested, in which a set of 90-minute O-D matrices, varying at every six seconds is estimated. For comparison, the Greenshields macroscopic model, which predicts the travel time by assuming that entered traffic will arrive at their destinations within two time intervals, is also tested in the same corridor example. The results show that the proposed algorithm can obtain a relatively accurate estimation result with RMSE value much smaller than the Greenshields model. To further investigate the applicability of the proposed algorithm, two case corridors on Taiwan Freeway No.1: Yangmei Toll Station to Taishan Toll Station and Taichung Interchange to Taipei Interchange are conducted. The results show that the proposed algorithm can obtain satisfactory RMSE values, suggesting the effectiveness and applicability of the proposed algorithm.