本研究以Chen and Hsueh (1998a,b)所建構之動態用路人均衡路 徑選擇模型為基礎，延續王中允(1999)、周鄭義(1999)及陳穎俊(1999) 之研究，應用變分不等式理論，探討含額外限制式之動態用路人均 衡路徑選擇模型相關課題及求解方式，所研究之相關課題如下： 1. 以路段流出率為決策變數之動態用路人均衡路徑選擇問題 2. 含路段流出率容量限制動態用路人均衡路徑選擇問題 3. 動態用路人均衡雙邊限制起迄/出發時間/路徑選擇問題演算法績 共計包含一個模型基礎課題及兩個含額外限制式課題。 在構建出以路段流出率為決策變數之動態用路人均衡模型後， 除了以簡單的數值例說明其正確性外，更探討其收斂特性不如原本 以路段流入率為決策變數模型之原因。鑑於此，在含路段流出率容 量限制動態用路人均衡路徑選擇問題中，仍以路段流入率為決策變 數構建此模型，並成功以變數轉換的方式求解此問題。此外，針對 動態用路人均衡雙邊限制起迄/出發時間/路徑選擇問題，構建適合路 徑演算法求解之模型，証明此模型與陳穎俊(1999)所建構之模型為對 等，並得到所發展之路徑演算法(梯度投影法，Gradient Projection Method)較陳穎俊(1999)所提出之路段演算法(Evans 演算法)有較佳之績效。 This thesis, as follow-up study of Wang (1999), Chou (1999) and Y.J. Chen (1999), attempts to further some important issues based on the dynamic user-optimal route choice model formulated using variational inequality approach by Chen and Hsueh (1998a,b). It includes the dynamic user-optimal route choice model formulated with decision variables of link exit flows, the dynamic exit-capacitated user-optimal route choice model, and comparison of the gradient projection method versus the Evans algorithm in the dynamic user-optimal doubly constrained O-D pair/departure time/route choice model. The non-convergence phenomenon in the dynamic user-optimal route choice model formulated with decision variables of link exit flows is more than with decision variables of link inflows (Chen and Hsueh, 1998a). So the dynamic exit-capacitated user-optimal route choice model is formulated with decision variables of link inflows and solved by transforming variable. Moreover, in the dynamic user-optimal doubly constrained O-D pair/departure time/route choice model, computational experience with three test networks indicates that the gradient projection method are in general superior to the Evans algorithm in terms of execution time.
