Energy Saving Train Control for Urban Railway Train with Multi-population Genetic Algorithm

Abstract

The problem of urban rail train energy saving control with specified running time is a typical multi-constrains, non-linear optimization problem. By applying minimum principle to differential motion model of trains, the energy saving control strategies are obtained. An approach for optimizing problem based on variable-length real matrix coding multi-population genetic algorithm (MPGA) is presented. The train running is simulated by a multi-particle simulator considering complicated line conditions and influence of train length. The GA chromosome consisting of a variable-length two dimensional real matrix represents the train control sequence. A variable length operator based on annealing selection is introduced to enhance global search performance. Fitness sharing keeps population's multiplicity. Multi-population parallel search improves convergence rate and evolution stability. The correctness and advancement of the optimization control method have been validated through the simulation platform of train operation.