Application Research On Genetic Algorithm For Track Arrangement Of Locomotive Depot Paddock

With the rapid development of railway, to ensure that locomotives in locomotive depot departure faster, more safely and more punctually and consequently to improve the efficiency of the railway operation are important issues we now need to address. Without a good and efficient track arrangement, the normal traffic plan will be seriously affected, and thus the efficiency of the railway operation. Therefore, to make a reasonable and efficient scheme for tracks arrangement of locomotive depot can guarantee that locomotives in locomotive depot can departure faster, more safely and more punctually thus can improve operational efficiency and safety of the railway.This paper first analyzes the basic principle of genetic algorithm, introduces some of its basic concepts and implementation techniques, and elaborates its mathematical foundation, development history and improvement direction. Then, this paper gives the mathematical model of tracks arrangement in service area of locomotive depot. It needs to meet the six requirements and two constraints for getting a more reasonable and more efficient tracks arrangement of locomotive depot. These six requirements construct six target functions, and the lower the value of target functions, the closer they meet the requirements. Finally, these six target functions are multiplied by the respective weighting value and then summing to obtain a total target function.The achievement part of the algorithm mainly tells about the algorithm implementation steps. This paper uses encoding of T×M two-dimensional matrix to code chromosome for scheme of track arrangement of locomotive depot that satisfy constraints. This coding generate initialized group P. After decoding each chromosome of P, by the total formula of the total objective function, I calculates the values of the objective function for each chromosome.Then through the formula of fitness function I convert total objective function and obtain the fitness value of each chromosome, while saving the globally optimal solution(the fitness value is the theoretical optimum value)that is found in this generation.Next, using roulette selection method、method of random point crossover of two-dimensional and method of random point mutation of two-dimensional I produce next generation for genetic manipulation. When the calculating time of the algorithm has not yet terminated. the next group P as the initialization group P of next iteration continue to perform calculating fitness value、 saving the globally optimal solution, selection crossover and mutation operations. This process is not executed cyclically before the calculating time of the algorithm terminates. After reaching the end of the calculating time, I decode the individual in this iteration to find the globally optimal solution. At the time of simulation, population size Z、 crossover probability pc、mutation probability pm、constant C are used to experience value and set to100、0.85、0.05and1.Termination calculating time is set to20seconds. Through simulation, we can see that this algorithm can solve the track arrangement of locomotive depot and is better than methods of manual transmission.Finally, in order to improve the performance of the algorithm, I optimize parameters of genetic algorithms. After parameter being optimized, we can search the number of average-optimal solution for track arrangement of locomotive in the unit time form18.3up to the28.2, and its standard deviation has been improved from2.1to0.46.Bothspeed and stability of algorithm has been greatly improved.