Enhanced DE with Weighted Base Vector for Unconstrained Global Optimization

Objectives: Differential Evolution (DE) algorithm has came out as a robust, effective and well-organized computational technique for solving global optimization problems. However, similar to other evolutionary algorithms of the same genre, DE has some inherent drawbacks like slow/ premature convergence, stagnation of population etc. due to its probabilistic nature. This paper aims to decrease the drawbacks and hence enhance the working of DE algorithm in term of convergence speed and accuracy of result. Method: This paper presents two improved versions of DE named Differential evolution with weighted base vector (DEwB-1) and DEwB-2 which adapts novel mutation scheme and self adaptive approach to control DE parameters. Findings: The corresponding DE versions are tested on 13 standard unconstrained problems as suggested in various literatures and a real life molecular potential energy problem. The numerical and statistical results expose that the proposed modifications assist in improving the performance of basic DE algorithm. Application: The variants can apply on more complex and constrained optimization problems.