Using neural networks as an aid for solving engineering problems

This paper describes applications of neural networks in solving nonlinear systems encountered in engineering problems. We will describe using neural networks to select starting points for iterative algorithms for nonlinear equations. Since input/output training data are often easily generated from a description of the problem or the system equations, a neural network can be trained to provide a rough model of the underlying (optimization) problem. After the neural network is trained, it is used to select starting points. We illustrate the method with two real engineering applications