مشخصات پژوهش

Navigating complex engineering systems with their dense networks of interconnected components can be challenging. Modeling these systems often requires a lot of computing power. Traditional optimization methods that rely on gradients tend to fall short. They often get stuck in local solutions and consume significant computing resources [1]. On the other hand, meta-heuristic algorithms have shown themselves to be strong and flexible options. They can efficiently search large, complicated spaces. These methods draw inspiration from various concepts, including natural phenomena, swarm behavior, and evolutionary biology. Examples include well-known techniques like Ant Colony Optimization (ACO) [2], Genetic Algorithm (GA) [3], and Particle Swarm Optimization (PSO) [4]. A key idea in this field, known as the "No Free Lunch" theorem, highlights that no algorithm is the best in every situation. A method that works well in one area might struggle in another. This principle drives the need for constant innovation. We must keep developing new solvers to meet the changing challenges of complex optimization problems. To overcome the limits of individual solvers and make the most of their strengths, researchers often use hybrid metaheuristics [5]. This approach combines parts from different algorithms. For example, it incorporates the local search accuracy of one method into the global search framework of another. This creates a stronger and more effective problem-solving technique. By blending components like the broad, population-based search of a Genetic Algorithm with the focused, path-based refinement of Simulated Annealing, hybrid models can better handle complex search environments. The main goal is to build a system that reduces the weaknesses of its individual algorithms while boosting their combined performance. This way, it achieves a balance between wide-ranging exploration and deep exploitation that is better than any single method.