A Constraint Handling Method Based on Cyclic Random Sampling

Real-world optimization problems often come with constraints that limit the feasibility of solutions, posing challenges for finding viable solutions. While various constraint handling methods have been proposed, such as the penalty function method, superiority of feasible solutions, and stochastic ranking, analysis and experiments have revealed their limitations in tackling complex real-world optimization problems. This paper proposes a constraint handling method based on cyclic random sampling. The method involves narrowing down the scope of decision variables through cyclic sampling, altering the composition of original samples, and ultimately reducing the search for infeasible solutions, termed as the Boundary Sampling Update (BSU) method. The integration of the BSU method with the Differential Evolution (DE) optimization algorithm is applied to real-world optimization problems. The results demonstrate that the proposed BSU method effectively enhances the search performance of the algorithm.