A Comprehensive Comparison of the Performance of Metaheuristic Algorithms in Neural Network Training for Nonlinear System Identification

Many problems in daily life exhibit nonlinear behavior. Therefore, it is important to solve problems. These are complex and difficult due their nature. It seen the literature that different artificial intelligence techniques used these One of most neural networks. Obtaining successful results with an network depends on its training process. In other words, should be trained a good algorithm. Especially, metaheuristic algorithms frequently advantages. this study, for first time, performance sixteen identification systems analyzed. aimed determine effective algorithms. The examined terms solution quality convergence speed. applications, six used. mean-squared error (MSE) utilized as metric. best mean values obtained were 3.5×10?4, 4.7×10?4, 5.6×10?5, 4.8×10?4, 5.2×10?4, 2.4×10?3, respectively. addition, test found all successful. When examined, was observed biogeography-based optimization, moth–flame bee colony algorithm, teaching–learning-based multi-verse optimizer generally more than systems.