Applications of Gudermannian Neural Network for Solving the Sitr Fractal System

Abstract

This study is related to explore the Gudermannian neural network (GNN) for solving a nonlinear SITR COVID-19 fractal system by using the optimization efficiencies of a genetic algorithm (GA), a global search technique and sequential quadratic programming (SQP) and a quick local search scheme, i.e. GNN-GA-SQP. The nonlinear SITR COVID-19 fractal system is dependent on four collections: "susceptible", "infected", "treatment" and "recovered". For the optimization procedures through the GNN-GA-SQP, a merit function is constructed using the nonlinear SITR COVID-19 fractal system and its corresponding initial conditions. The description of each collection of the nonlinear SITR COVID-19 fractal system is provided along with comprehensive detail. The comparison of the achieved numerical result performances of each collection of the nonlinear SITR COVID-19 fractal system is performed with the Adams results to verify the exactness of the designed computational GNN-GA-SQP. The statistical processes based on different operators are presented for 30 independent trials using 5 neurons to authenticate the consistency of the designed computational GNN-GA-SQP. Moreover, the graphs of absolute error (AE), performance indices, and convergence measures along with the boxplots and histograms are also plotted to check the stability, exactness and reliability of the designed computational GNN-GA-SQP.