A Random Parameter Value Strategy for the Biased Random-Key Genetic Algorithms

The Biased Random-Key Genetic Algorithm (BRKGA) is a population-based metaheuristic designed to find optimal or near-optimal solutions for combinatorial problems. The performance of such algorithms heavily depends on parameter settings, which influence solution quality and search efficiency. One approach to this challenge is parameter control (or online tuning), where the algorithm adapts parameter values during the search process, utilizing accumulated information about the fitness landscape. This method also alleviates the user from manual tuning. This work proposes a novel approach by using random parameter values for each generation in BRKGA. Evaluations on three classical optimization problems—the Flowshop Permutation Problem, Set Covering Problem, and Traveling Salesman Problem—showed that this strategy produced competitive results compared to the tuned algorithm.