Improved GA of Train Operation Simulation Model

In order to optimize and solute train operation simulation model, the paper develops the Changeable Chromosome Length Multi-objective Improved GA. Gene encoding is based on control shift position and then get encoding strategy. The fitness function includes three train control indexes: stop deviation, time deviation and energy consumption and is unitary through weighing each control index. The fitness of individual is computed by train operation simulator. In order to increase adaptability of the genetic algorithm to complex line, individual valid examine method, elitist strategy, punish function and chromosome length variety algorithm are designed. At last, through many tests indicate that the genetic algorithm can apply to complex line, and the optimized solution can satisfy multi-objective demand. The optimized solution can save energy 5%–25% than traditional multi-particle train operation simulation system. So the algorithm is effect.