Application of genetic algorithms for solving the scheduling problem with moving executors

A new version of a genetic algorithm is proposed. For determination of crossover and mutation probabilities the learning algorithm is used. The algorithm is applied for solution of the tasks scheduling problem with moving executors. The learning procedure is performed with respect to different execution times in the scheduling problem. A basic scheme of genetic algorithm with generation of the initial population, selection and two reproduction algorithms is used. As the fitness function the makespan is assumed. The results of simulation experiments which evaluate the learning procedure as well as the effect of learning are presented. They show the slight improvement of the solution algorithm quality after applying the learning procedure for the crossover probability.