Treffer: AN ADAPTIVE MULTI-STRATEGY SELECTION METHOD FOR IMPROVING GENETIC ALGORITHM PERFORMANCE

The traveling salesperson problem (TSP) is a well-known NP-hard optimization problem frequently used as a benchmark for evaluating genetic algorithms (GAs). Conventional GAs often suffer from premature convergence and a lack of population diversity, which negatively impacts the crucial balance between exploration and exploitation. This study proposes a novel selection operator that dynamically adapts its selection strategy – incorporating methods such as roulette wheel, and stochastic universal samplings – based on real-time performance metrics. This adaptive mechanism aims to maintain population diversity while simultaneously enhancing convergence speed. The efficacy of the proposed operator is evaluated using established TSPLIB benchmark instances and a real-world dataset comprising city coordinates within Pakistan. The results demonstrate that the proposed operator achieves superior performance compared to existing selection strategies, exhibiting improvements in both solution quality and convergence speed, along with increased stability as problem size increases. This research contributes to the field of evolutionary computation and offers a potentially valuable approach for addressing large-scale optimization problems, including the TSP and related challenges.