A preferred learning based adaptive differential evolution algorithm for large scale optimization

摘要

With the help of the cooperative co-evolution, differential evolution (DE) has been applied successfully from low-dimensional problems to large scale optimization. In this paper, we propose a preferred learning cooperative coevolution DE algorithm (LDECC-DG) which focuses on the basic optimizer for large scale optimization using cooperative coevolution. The proposed LDECC-DG builds on the differential evolution with cooperative coevolution and differential grouping (DECC-DG) algorithm which possesses an accurate grouping method and an effective basic optimizer method for large scale optimization. A novel DE algorithm called preferred learning based adaptive DE (LDE) is designed as a basic optimization algorithm for large scale problems and the control parameters in LDE are selected according to the self-adaptive strategy which corresponds to the preferred learning strategy. We show that how the LDE can improve the performance of Cooperative Co-evolution framework on account of its effectiveness. In order to evaluate the performance of LDECC-DG for large-scale global optimization, we carried out numerous computational studies on the CEC 2010 benchmark functions. The results show advantages of the LDECC-DG in both solution quality and convergence rate compared to other algorithms.