應用條件逆矩陣方法推估依時性起迄對旅次矩陣之研究

A Conditional Inverse Approach for Inferring Time-dependent Origin-Destination Trip Matrices

本研究利用條件逆矩陣方法可求得非方陣型式矩陣之條件逆矩陣的特性，應用於求解路段流量推估依時性起迄對旅次矩陣的研究中。研究結果顯示，在測試路網中，條件逆矩陣的求解程序在不同的測試情境下，均能準確反推出依時性的起迄對旅次矩陣，且能同時在符合及不符合動態使用者最佳化的路網流量狀況推估出正確的依時性起迄對運輸需求，較傳統的路段流量反推依時性起迄對旅次矩陣的模型更具準確性及一般性。

In this study, we proposed a procedure to solve the time-dependent Origin-Destination (O-D) trip matrix estimation problem which adopted the Conditional Inverse Matrix approach. We tested this procedure in different scenarios of the test network. The results showed that the estimation, is fairly accurate. Compared to the traditional O-D demand trips estimation models, the model developed in this research can handle all of the link flow situations, whether the link flow is restricted by dynamic user optimal (DUO) conditions or not. The present study is a more general time-dependent O-D trip matrix estimation method than most traditional models.