Estimation of intersection origin-destination (OD) matrices is a frequently-encountered problem in transportation management operations. In particular, advanced traffic responsive arterial signalization management schemes rely on the accurate real-time estimation of vehicle queue lengths and turning ratios. This paper proposes a new method for estimating OD matrices of a signalized intersection in real time. The proposed method only requires detectors at exit legs, saving half detectors from a complete configuration assumed by previous research. Signal phase timing information is used to determine the portion of the volume measurement that corresponds to a specific phase. The estimation of turning proportions is formulated as a constrained least squares problem and solved using recursive algorithm with forgetting factor and covariance resetting. Simulation shows that the proposed method can obtain accurate results and response to parameter changes fast, while requiring significant less computational effort than a comparable solution using a constrained optimization method.
↧