Control parameters self-adaptation in differential evolution based on intrisic structure information
Zhigang Shang, Zhihui Li, Jing Liang, Ben Niu · 2012
In this paper, A self-adaptive strategy to determine the control parameters of Differential Evolution (DE) is proposed based on the elaborate analysis of intrinsic structure. The projection information of fitness function in differential direction is used to get the scale factor, while the difference between the local distance and global search range is applied to determine the crossover rate. This approach named as ITDE is tested by the benchmark functions including unconstrained uni-Modal, multi-Modal and bound-constrained ones, and the test results indicate the ITDE has good performance to balance the explorative and exploitative capability of algorithm than standard DE and easy to be implemented widely with many variants of DE algorithm.