An improved adaptive genetic algorithm in flexible job shop scheduling

In the view of the two aspects problems of the existing genetic algorithm, one is the low quality of the initial solution, the other is that in the later stage of evolution, convergence rate is slow, so we proposed an improved genetic algorithm, which was applied to solve the flexible job shop scheduling problem. This algorithm took the method FJSP as research subject, to minimize the maximum completion time, firstly we improved the initial population, when the initial population selected machines, instead of randomly generating, we used roulette wheel selection strategy to improve the quality of initial population; Secondly, in the process of crossover and mutation, instead of the fixed probability value, the crossover probability and mutation probability could be used to change the value adaptively according to the evolution, we presented a new adaptive probability of crossover and mutation, in the cross-process, POX cross pattern has been used, the convergence rate was greatly improved. Finally, the simulation results verified the advantages of the improved algorithm at the optimum value and convergence rate.