Abstract:

There are many ways to solve a linear equation，and various solving methods will result in different convergence speeds.Two different error functions are designed for one linear equation,and then two gradient neural network models are established according to the gradient descent method.Matlab is used for computer simulation,the solutions of linear equation are given according to the different gradient neural network models,and the feasibility of these two gradient neural network models is confirmed.Finally，the convergence speeds in solving the linear equation are compared.

Key words: error function, gradient neural network, system of linear equations, Matlab simulation