Özet:

Real parameter optimization problems are often used to solve metaphoric algorithms. These algorithms are tested by their designers until they obtain sufficient performance before they are applied to problem-solving. The designers suggest the algorithms they test in the n literature contains a large number of synthetic function sets presented. One of them is the set involved in the CEC 2019 competition and called the 100 scores problem. This problem contains 10 functions that are difficult to solve. The goal of this pro blame is to correctly calculate the global optimum value of all functions up to 10 degrees. Each function is given 10 points to the correct determination and the result of the determination of the whole is reached 100 points. One of the important algorithms used to solve the real parameter optimum is the Differential Development (DE) algorithm. Its simple structure, easy to realize and the successful results it obtained have led to the widespread use of DE and improved performance and the emergence of new variants. The most famous of the DE variants in literature is the JADE algorithm in the beginning. JADE has a new mutation equation for the original DE, an adaptable method for determining parameters and finally an archive strategy where the information of individuals is kept to increase the diversity of populations. In this study, the solution to the 100 points problem belonging to the CEC 2019 competition was carried out using the JADE algorithm. The results were compared with two metaphor. These are the Differential Development and Artificial Bee Colony (ABC) algorithms. The parameters of the algorithms are configured with the automatic parame tre tool to perform the trial experiments in which the three algorithms participate in a fair way. Furthermore, all the algorithms involved in the experiments were studied with the working behaviors of the algorithms running with different function call numbers. The results showed that JADE had better results than the algorithms compared to each FES value in which it worked. As the FES value increased, the success of the algorithm improved.

Anahtar Kelimeler:

Özet:

Metaheuristic algorithms are frequently used to solve real parameter optimization problems. Before these algorithms are applied to the solution of problems, they are tested by their designers until they have enough performance. There are many benchmark sets presented in the literature to test the algorithms proposed by the designers. One of them is the one included in the CEC 2019 competition and named as the "100-digit problem". One of the most important algorithms used for the solution of real parameter optimization is the Differential Evolution (DE) algorithm. Its simple structure and easy implementation have led to the widespread use of DE and the emergence of new variants by improving its performance. In this study, the solution of the 100-digit problem belonging to the CEC 2019 competition was carried out using the JADE algorithm. The results obtained are compared with two metaheuristics. These algorithms are Differential Evolution and Artificial Bee Colony (ABC). For the experiments involving three algorithms to be fair, the parameters were configured with the automatic parameter tool. Besides, all algorithms participating in the experiments were run with different function evaluation numbers (FES) and the working behavior of the algorithms was examined. The results showed that; JADE achieved better results than other algorithms at all FES values. Also, as the FES value increased, the performance of the JADE algorithm improved.