An Improved Iterative Algorithm of Neural Network for Nonlinear Equation Groups

Neural network can precisely approach the inverse function of function of nonlinear equation groups, and solution for the groups can be computed using iterative algorithms. In the experiment, the emergence of endless iterations is found in the iteration process because of premature convergence of error, and it is also found that the value of convergence not always is the minimum. Iterative algorithm is improved for the two problems, and the experimental analysis is performed by using examples. The approximate solution of the equation can be obtained even when input sample interval deviates from the right solution very far. New optimization algorithm of initialized weights and thresholds is used to reduce the error.