Optimization with Improved Differential Evolution Algorithm having Variable Tolerance

In this paper, improved differential evolution (IDE) algorithm with variable 'del' parameter is presented. Recently developed IDE algorithm is used for global optimization of non-linear complex problems. The 'del' parameter is a predetermined level of tolerance for predefined allowable number of iterations which triggers the different mutation schemes in restart mechanism phase of IDE algorithm. The value of 'del' parameter is varied in the range [1e-1, 1e-10] in this contribution. The algorithm is tested on five commonly used benchmark functions, namely, Ackley, Griewank, Rastrigin, Schefel2.22 and Sphere to prove its findings with variable 'del' parameter. The outcomes show that IDE with lower value of 'del' parameter produced almost same or better results.