Enhanced Genetic Algorithm for Variable Minimization through Modified Crossover Operator

Genetic algorithms (GAs) are commonly employed in optimization techniques and one of the crucial components of GAs is the crossover function, which often encounters the issue of premature convergence (PC). The proposed innovative crossover operator, known as Flip Multi-Sliced Average Crossover (FMSAX), significantly enhances the performance of GA by addressing variable minimization. This approach introduces a novel method of generating offspring by dividing the chromosomes into three equal segments, producing a head-body-tail structure. This is achieved by flipping each gene and computing the average to generate a new offspring. The FMSAX operator represents a more optimized and efficient technique for variable optimization, aimed at mitigating premature convergence issues. The simulation results, two types of datasets, each containing 30 and 40 variables, were utilized. The results clearly demonstrate that the enhanced crossover operator within the GA outperformed the original GA with an average crossover operator by 25 variables (62.50%) and 10 variables (25%) eliminated, respectively. Moreover, it yielded Root Mean Squared Error (RMSE) and Mean Absolute Percentage Error (MAPE) values of 0.0066 and 0.0738, respectively.