Dual preferred learning embedded asynchronous differential evolution with adaptive parameters for engineering applications

Balance in exploration and exploitation is the basic requirement of any optimization algorithm, lack of which can easily lead to premature convergence of algorithm. Asynchronous Differential Evolution (ADE), a variant of Differential Evolution (DE) algorithm has strong exploration and parallel optimization characteristics. It immediately updates the population with better individuals unlike DE in which the population is updated in next generation only. This feature leads to faster convergence but increases the chances of getting stuck in local optima. To improve the performance of ADE, the mutation operation of the algorithm is enhanced with dual preferred learning (DPL) mutation, and to balance exploration and exploitation, the control parameters are made adaptive in this work. The proposed algorithm is named as DADE (DPL based adaptive ADE). DPL enables learning from individuals having better fitness and diversity hence the proposed combination enhances the convergence rate and population diversity. In addition, inclusion of adaptive control parameters make algorithm more robust. The algorithm is investigated on 25 widely used bench-mark functions and compared with several state-of-the-art algorithms. Non-parametric statistical analysis of the proposed algorithm is also presented backing its performance. Further, it is also tested for engineering design problems. The simulation results show that the proposed work provides promising results and outperforms the competitive algorithms.