Development of a new metaheuristic algorithm for optimization problems

Optimization is an effort to achieve the best result by making choices on the inputs of an objective function under certain criteria and is a problem that is frequently encountered in almost every field of science and in real world problems. In multidimensional complex optimization problems, it is impossible to search for all possible solutions or combinations in an acceptable time frame and find the best one. Metaheuristic algorithms are an important area of research that has made significant progress in solving complex optimization problems in acceptable time. Since the applicability of these algorithms, their running times, and the accuracy of the solutions they produce vary according to different problem types, many studies have been and are being conducted in the literature to propose new metaheuristic algorithms. The development of new metaheuristic algorithms is an important process to provide alternative solutions for different optimization problems. In this thesis, a new metaheuristic algorithm called Flood Algorithm is proposed for optimization problems. This algorithm is inspired by the flow of flood water on the Earth's surface. It has a single basic parameter and can be applied to different types of optimization problems. The proposed algorithm is tested on fifteen functions with different properties (unimodal, multimodal, separable and non-separable in different dimensions) from the benchmark functions commonly used in the literature for testing metaheuristic algorithms, on three engineering design problems, and on a real-life exam seating plan preparation problem. The results obtained are compared with the most widely used metaheuristic algorithms in the literature, namely Genetic Algorithm, Simulated Annealing and Particle Swarm Optimization algorithms. The comparison results show that the proposed algorithm has a competitive performance in terms of solution accuracy and convergence performance with the other metaheuristic algorithms used in the comparison. Especially in multidimensional optimization problems, it outperformed the other algorithms. Wilcoxon Ranksum Test comparisons on the raw results also show that the Flood Algorithm produces better results for all problem types.