Experimental Analysis of Effect of Tuning Parameters on The Performance of Diversity-Driven Multi-Parent Evolutionary Algorithm
Sumika Chauhan, Manmohan Singh, Ashwani Kumar Aggarwal
Abstract
In order to find a global optimal solution to real-world optimization issues, an algorithm that efficiently explores the search space is required. The DDMPEA is a newly developed algorithm that uses selection, crossover, and mutation to perform search operations. The search performance of this algorithm is depending upon the tuning parameters involves in this technique. There are four parameters in this algorithm which are required to tune to obtain better solutions. In this work, sixteen cases are developed which are tested on eight classical benchmark functions. The performance of proposed algorithm is tested and performance is obtained in terms of average, median, standard deviation, best and worst values. The results suggest better values of tuning parameters on which this algorithm gives global optimum solutions.