AN INVESTIGATIVE PARAMETER ANALYSIS OF PASTORALIST OPTIMIZATION ALGORITHM (POA): A NOVEL METAHEURISTIC OPTIMIZATION ALGORITHM
Ibrahim Mohammed Abdullahi, Muhammed Bashir Mu’azu, Olayemi Mikail Olaniyi, James Agajo · ATBU Journal of Science, Technology & Education · 2019
In this paper, one of the most important parameters that affects the performance of a novel population based nature-inspired metaheuristic optimization algorithm called the Pastoralist Optimization Algorithm (POA) inspired by the pastoralists herding strategies were investigated. This is to determine the suitable range of values that should be used when applying the algorithm to solve optimization problems. The parameter that was investigated is the number of pastoralist (nP), that is the number of search agents. Eight different pastoralist size (10, 20, 30, 40, 50, 60, 70 and 80) were investigated by testing the parameter on three standard test functions; unimodal Sphere function, multimodal Dejong and Shubert functions. Each test is simulated ten times, and the average optimal value and the average convergence time(s) obtained for each function and each parameter value was recorded. The experimental results obtained show that a pastoralist population size of 20 is desirable for convergence accuracy and speed.