Evolutionary Search from the Interior of Feasible Space
Noha M. Hamza, Ruhul Amin Sarker, Daryl Essam · 2020
Constraint handling is a vital component in handling constrained problems using evolutionary algorithms, which is usually performed using an additional step instead of considering it as an integral part of the evolutionary search process. By utilizing the constraint consensus concept, this paper proposes searching from inside the feasible region. This is done by proposing a new differential evolution mutation operator that guides the selected feasible individuals to move towards the current best individual. The new approach algorithm is validated by solving a well-known set of constrained problems, with the results demonstrating the benefits the new concept adds to the algorithm in terms of the quality of solutions and computational time. The algorithm also shows superior performance over many state-of-the-art algorithms. Furthermore, the algorithm is tested on five mechanical engineering problems where the results demonstrate its effectiveness in attaining the best results.