Optimal Traffic Counting Locations For Origin-Destination Matrix Estimation

There has been substantial interest in development and application of methodology for estimating origin-destination (O-D) from traffic counts. Generally, the precision of an estimated O-D matrix depends much on the quantity and quality of input data. Many factors can influence the quality of traffic counting data, such as stochastic error of traffic counting data, the number and locations of traffic counting points. The former has been investigated extensively, while the latter has received very limited attention and the research is not perfect.The thesis discusses basic issue and theory of traffic counting location (TCL), boils existent methods of TCL down to two kinds, experiential method and analytical method. In view of the shortage of the existent theory and methods, aiming at reduction of O-D trip matrix estimating error, it finds new TCL method from two ways which reduce the maximum probable relative error up bound and restraint equation systems feasible region. It develops nonlinear programming and much practical heuristic algorithms. Then it develops TCL application by Visual Basic and Office component and illustrates with numerical examples. Finally, it analyzes the relationships between two models and their applicability.