Abstract: In order to overcome the limitations of differential evolution algorithm with typical mutation operators， an improved adaptive differential evolution algorithm was proposed with a new mutation operator with global acceleration. The global acceleration operator can balance the global search and local search， such that the algorithm has higher optimization efficiency. According to the difference vector and the population distribution， the value of mutation rate was selected to slow down the trend of search scope narrow to maintain high population diversity. The crossover rate was adaptively chosen from two intervals through learning and comparing to meet the needs of the evolutionary search and improve the versatility of the algorithm. The improved algorithm is applied to the largescale reliability problems and the experimental results show that the improved algorithm achieves better optimization performance in solving largescale reliability problems.

Key words: global acceleration, differential evolution algorithm, large scale reliability problems, twointerval selection strategy, parameter adaptation