Solving Traffic Equilibrium Assignment Problem with Genetic Algorithm

This paper proposes a parallel processing method of genetic algorithm for the traffic equilibrium assignment problem with fixed traffic demand in order to forecast the traffic volume accurately. In the algorithm design, some improved steps such as the multi-dimensional parallel crossover, variational rates of crossover and mutation, the elitism and the punishment of objective function are taken, so that the computing speed is greatly improved; the computation time and the assignment complexity are substantially reduced. It founds a new way for traffic assignment problems, and at the same time shows a potential practical prospect for the genetic algorithm to be used in the traffic programming.