Real-time estimation of turning movement proportions based on genetic algorithm

Real-time turning movement proportions at intersections are important input data for adaptive traffic signal control system. To estimate them, a revised parameter optimization model with the objective function to minimize the sum of absolute deviations between measured and estimated traffic counts is proposed. A genetic algorithm is put forward to solve the problem according to its characteristics. The detailed encoding and decoding methods satisfying the inherent constraints of split parameters automatically are presented, and five other key issues are illustrated. Computational results are reported on a set of test problems using simulated as well as practical traffic data, and the capability to track dynamic turning movement proportions is compared with both least-square and Kaiman filtering methods. The results indicate that the proposed method is quite accurate, efficient and robust.