New Improved SALSHADE-cnEpSin Algorithm with Adaptive Parameters

Differential Evolution algorithm is very challenging algorithm and has been found to put forth the basis of evolutionary computation. This algorithm because of its simple structure and linear nature, has been applied to a large number of optimization problems from various diversified fields. In this paper, we propose a new variant of DE by modifying the original LSHADE-cnEpSin algorithm. Two new modifications are proposed, keeping all the modifications of LSHADE-cnEpSin intact. The modifications proposed include the introduction of adaptive parameters by using Weibull distribution based scaling factor and exponentially decreasing crossover rate. Apart from that linearly decreasing population size is also used. The main reason for these adaptations is to make an adaptive algorithm so that no parameter needs to be changed from the end user perspective. The proposed algorithm has been applied to solve CEC2017 and CEC2019 benchmark problems. The numerical results prove that the newly proposed SALSHADE-cnEpSin algorithm performs better than SaDE, JADE, SHADE, LSHADE, CV1.0, CVnew, MVMO and other algorithms.