To improve the performance of genetic algorithms by using a novel selection operator

The Genetic Algorithm (GA) was developed as a search engine for difficult non-deterministic polynomial optimization problems. However, it suffers from internal weaknesses, such as premature convergence and low computation efficiency. One critical aspect of the GA is the selection process, which determines new paths and ultimately guides the algorithm towards a solution. This paper details a novel selection procedure that is a perfect blend of the two extremes, namely exploitation and exploration. The proposed technique eliminates the fitness scaling problem by changing the selection pressure continuously during the selection stage. Utilizing traveling salesman problem library (TSPLIB) instances, a performance comparison of the proposed method with a few traditional selection methods was conducted, and the proposed strategy yielded much better outcomes in the form of standard deviations and mean values. A two-sided t-test was also developed, and the results revealed that the proposed strategy enhanced the performance of a GA substantially.