| Summary: | 碩士 === 國立交通大學 === 運輸與物流管理學系 === 105 === With the development of urban city, the growth of population and transportation mode and the change of land use, the improvement and construction of transportation are much more important, therefore, transportation planning has become an important issue. Estimation of O-D matrices is an important step for transportation planning. It’s an essential part of network design, fare setting and operation analysis. Past methods faced the problem with accuracy and fitting of pattern. In this paper, we propose an estimation method which can fit the pattern though traffic counts of origin and destination and the cost matrix of the network. We improve Sen-Smith method, included Hyman method and Deming-Stephan-Furness procedure.
To calibrate the parameter of deterrence function, Sen-Smith method required a set of known O-D matrices. However, a set of known O-D matrices are difficult to find. To solve the problem mentioned previously in this study, we change the initial value of Hyman method, and the initial values of DSF procedure are also altered, which can reduce the number of iterations of the DSF procedure. In addition, we develop a statistical test which allows a specifit error tolerance of data based on the Kolmogorov-Smirnov (K-S) test. Numerical examples display in this study shows that the method of our research can fit the real traffic pattern.
|
|---|
