Improving the efficiency of computational methods for solving some classes of systems of differential equations

This paper describes ways to improve the efficiency of numerical solutions of systems of differential equations that describe the corrosive wear. In order to control the accuracy of their solutions is proposed to use artificial neural networks to determine the numerical integration of rational parameters for given values of maximum permissible error of the result. Application of neural networks can significantly improve computational efficiency of the algorithm by minimizing the number of iterations for solving a system of differential equations.