An improved particle swarm algorithm with immune mechanism for traffic matrix estimation

Owing to the shortcoming of local convergence of the particle swarm optimization algorithm, presenting relative distances between particles to enhance probability selection formula, an improved particle swarm optimization with immune theory is introduced. A particle updates its velocity and position not only by individual and global optima,but also by individual optima of a specific particle chosen by roulette method according to certain probability, to maintain population diversity and prevent precocity and stagnation. This method is used to solve the maximum entropy model, estimating OD matrix from traffic flows. By an experiment on a junction in Chongqing City, the results demonstrate that the particle swarm algorithm overcomes the defect of Newton method that strictly relies on initial values, and the improved particle swarm algorithm has much higher optimization capability than the basic particle swarm algorithm and the basic genetic algorithm.