A Novel Metaheuristic Method: Gravitational-based Global Optimization

Abstract

Abstract Metaheuristic optimization methods provide a satisfactory solution for complex engineering problems. In this study, a novel metaheuristic searching approach was proposed to tackle engineering problems. The relative slope-based gravitational searching algorithm, namely XAR, provides a novel searching strategy, which is built on the law of gravity and interaction of the inertia mass. Making to move the searching agents using the slope-based gravity is the original contribution to this work. Searching agents of the method, i.e., balls, evolutionarily move regarding the relative slope of the consecutive balls. The algorithm determines a set of resolutions by the numbers of the balls aggregated. In other words, the method converges to the area in search space, where the greatest number of agents is located. The algorithm has been tested on a wide range of benchmark function sets and a complex real-world engineering problem. The implementation results confirm a notable achievement for a diverse set of cost functions. The results demonstrate that the proposed algorithm achieves a notable achievement of global optimum compared to the state-of-the-art methods. Furthermore, a satisfactory rate of convergence for all functions was found. In other word, the method converges to the optimal position (minimum cost) in search space, where the greatest number of agents is located.