Estimation of Origin-Destination Matrix from Uncertain Link Counts Using Mixed Intelligent Algorithm

The link counts are usually inaccurate under the influence of measurement and sampling errors, all kinds of adventive events and succedent data processing errors. Ignoring data inaccurateness may lead to unacceptable error of O-D matrices estimation. To make the best of the fuzzy information acquired from traffic survey, this paper presents a bi-level programming model with fuzzy expected value model as upper-level and a user equilibrium assignment as lower-level by taking the link counts as fuzzy number. A mixed intelligent algorithm is also designed to solve the problem and the steps are illustrated. Using a simple road network, a simulation test is made to check the validity of this new approach.