Abstract:

Aiming at the shortcomings of genetic algorithm which is easy to fall into local optimum, an adaptive genetic algorithm based on individual ordering was proposed in this study. In the traditional adaptive genetic algorithm, the adaptive update of the crossover probability and the mutation probability was based on the individual fitness value. But in the later stage of the algorithm, as the population falled into the local extreme value, the difference of the value became smaller, and it was difficult to reflect individual differences when updating. Referring to the idea of order optimization, in the proposed improved algorithm, the individual fitness values were sorted, and the ordering number was used instead of the fitness value. This method of replacing the value difference by using the order difference could increase the value of the crossover probability and the mutation rate in the middle and the late stage of the population, which was beneficial to avoid the premature convergence of the algorithm. Several standard functions were tested, which showed that the improved algorithm was superior to the other two adaptive improved algorithms in terms of convergence speed and convergence precision.

Key words: genetic algorithm, sorting number, self-adaptation, fitness value, test function, convergence