A corporate shuffled complex evolution for parameter identification

This paper proposes a new version of the shuffled complex evolution (SCE) algorithm for solving parameter identification problems. The SCE divides a population into several parallel subsets called complex and then improves each sub-complex through an evolutionary process using a Nelder–Mead (NM) simplex search method. This algorithm applies its evolutionary process only on the worst member of each sub-complex whereas the role of other members is not operative. Therefore, the number and variety of search moves are limited in the evolutionary process of SCE. The current study focuses to overcome this drawback by proposing a corporate SCE (CSCE). This algorithm provides an evolutionary possibility for all members of a sub-complex. In the CSCE, each member is influenced by a simplex made from all other members of the current sub-complex. The CSCE barrows three actions of NM, i.e. reflection, contraction and expansion, and applied them on each member to find a better candidate than the current one. The efficacy of the proposed algorithm is first tested on six benchmark problems. After achieving satisfactory performance on the test problems, it is applied to parameter identification problems and the obtained results are compared with some other algorithms reported in the literature. Numerical results and non-parametric analysis show that the proposed algorithm is very effective and robust since it produces similar and promising results over repeated runs.