Exploitation Of Differential Evolution Research Articles

Two Strategies to Improve the Differential Evolution Algorithm for Optimizing Design of Truss Structures

The performance of differential evolution (DE) mostly depends on mutation operator. Inappropriate configurations of mutation strategies and control parameters can cause stagnation due to over exploration or premature convergence due to over exploitation. Balancing exploration and exploitation is crucial for an effective DE algorithm. This work presents an enhanced DE (EDE) for truss design that utilizes two new strategies, namely, integrated mutation and adaptive mutation factor strategies, to obtain a good balance between the exploration and exploitation of DE. Three mutation strategies (DE/rand/1, DE/best/2, and DE/rand-to-best/1) are combined in the integrated mutation strategy to increase the diversity of random search and avoid premature convergence to a local minimum. The adaptive mutation factor strategy systematically adapts the mutation factor from a large value to a small value to avoid premature convergence in the early searching period and to increase convergence to the global optimum solution in the later searching period. The outstanding performance of the proposed EDE is demonstrated through optimization of five truss structures.

Differential evolution with k-nearest-neighbour-based mutation operator

Differential evolution (DE) is one of the most powerful global numerical optimisation algorithms in the evolutionary algorithm family and it is popular for its simplicity and effectiveness in solving numerous real-world optimisation problems in real-valued spaces. The performance of DE depends on its mutation strategy. However, the traditional mutation operators are difficult to balance the exploration and exploitation. To address these issues, in this paper, a k-nearest-neighbour-based mutation operator is proposed for improving the search ability of DE. The k-nearest-neighbour-based mutation operator is used to search in the areas which the vector density distribution is sparse. This method enhances the exploitation of DE and accelerates the convergence of the algorithm. In order to evaluate the effectiveness of our proposed mutation operator on DE, this paper compares other state-of-the-art evolutionary algorithms with the proposed algorithm. Experimental verifications are conducted on the CEC '05 competition and two real-world problems. Experimental results indicate that our proposed mutation operator is able to enhance the performance of DE and can perform significantly better than, or at least comparable to, several state-of-the-art DE variants.

Differential evolution algorithm with multiple mutation strategies based on roulette wheel selection

In this paper, we propose a differential evolution (DE) algorithm variant with a combination of multiple mutation strategies based on roulette wheel selection, which we call MMRDE. We first propose a new, reflection-based mutation operation inspired by the reflection operations in the Nelder–Mead method. We design an experiment to compare its performance with seven mutation strategies, and we prove its effectiveness at balancing exploration and exploitation of DE. Although our reflection-based mutation strategy can balance exploration and exploitation of DE, it is still prone to premature convergence or evolutionary stagnation when solving complex multimodal optimization problems. Therefore, we add two basic strategies to help maintain population diversity and increase the robustness. We use roulette wheel selection to arrange mutation strategies based on their success rates for each individual. MMRDE is tested with some improved DE variants based on 28 benchmark functions for real-parameter optimization that have been recommended by the Institute of Electrical and Electronics Engineers CEC2013 special session. Experimental results indicate that the proposed algorithm shows its effectiveness at cooperative work with multiple strategies. It can obtain a good balance between exploration and exploitation. The proposed algorithm can guide the search for a global optimal solution with quick convergence compared with other improved DE variants.