Self-adaptive differential evolution algorithm for the optimization design of pressure vessel
Huixian Huang, Pengfei Hu · 2016
Considering both exploitation and exploration capabilities, a self-adaptive differential evolution algorithm is proposed in this paper. All populations are divided into two parts, and two mutation strategies are selected. The one sub-population use the DE/rand/1 approach, in which scale factor F is changed according to corresponding model of exponential declining function; The second sub-population adopt DE/best/2 method in which best fitness value's information is kept in each generation and scale factor is self-adapted according to distance between current and optimal individual. To strengthen information sharing between individuals, it is crucial that two groups are integrated into a population after mutation operator, Meanwhile, the adaptive crossover probability CR is to adaptively adjust the value by decreasing quadratic function. In the following benchmark function experiments, the proposed algorithm performs obviously better than the basic DE algorithm. Finally, it is found that the improved differential algorithm(IDE) could also be well applied in optimization design of pressure vessel.