An elitist whale optimization algorithm with the nonlinear parameter: Algorithm and application

Abstract

Abstract To address the problem that the whale optimization algorithm tends to fall into the local optimum and fails to maintain a balance between exploration and exploitation, an elitist whale optimization algorithm with the nonlinear parameter (EWOANP) is proposed in this paper. An elitist strategy based on the random Cauchy mutation is used in the shrinking encircling mechanism to increase the chance of escaping the local optimum. Cleverly, the strategy is to generate mutation solutions based on the random Cauchy mutation, after which the better population is selected to proceed to the next iteration. Then, a nonlinear parameter is used in the logarithmic spiral mechanism to balance exploration and exploitation. Various numerical optimization experiments are performed based on the IEEE CEC2020 benchmark suite and compared with eleven other algorithms. The results show that EWOANP outperforms most competitors in numerical optimization. Finally, the backpropagation neural network is optimized by EWOANP to build a prediction model for the sulfur content in the molten iron. The experimental results based on production data indicate that the proposed prediction model has a relatively small fluctuation in errors. Compared to the other seven competitors, the proposed model has a better prediction performance with RMSE=0.001457 and R2=0.916619.

Keywords

• backpropagation neural network
• elitist strategy
• sulfur content
• whale optimization algorithm