Genetic algorithm solution for the stochastic equilibrium transportation networks under congestion

A bi-level and mutually consistent (MC) programming techniques have previously been proposed, in which an area traffic control problem (ATC) is dealt with as upper-level problem whilst the users' equilibrium traffic assignment is dealt with as lower-level problem. In this study, genetic algorithm (GA) approach has been proposed to solve upper-level problem for a signalized road network under congestion. Stochastic user equilibrium (SUE) traffic assignment is applied at the lower-level. At the upper-level, GA provides a feasible set of signal timings within specified lower and upper bounds on signal timing variables and feeds into lower-level problem. The SUE assignment is solved by way of Path Flow Estimator (PFE) and TRANSYT traffic model is applied at upper-level to obtain network performance index (PI) and hence fitness index. Network performance index is defined as the sum of a weighted linear combination of delay and number of stops per unit time under various levels of traffic loads. For this purpose, the genetic optimizer, referred to as GATRANSPFE, combines the TRANSYT model, used to estimate performance, with the PFE logit assignment tool, used to predict traffic reassignment, is developed. The GATRANSPFE that can solve the ATC and SUE traffic assignment problem has been applied to the signalized road networks under congestion. The effectiveness of the GATRANSPFE over the MC method has been investigated in terms of good values of network performance index and convergence. Comparisons of the performance index resulting from the GATRANSPFE and that of mutually consistent TRANSYT-optimal signal settings and SUE traffic flows are made.