Using Artificial Neural Networks for Solving Engineering Problems

Abstract

SummaryAn artificial neural network (ANN) is a ‘black box’ capable of learning and reproducing a relationship between its inputs and outputs. Once trained, ANNs can calculate a set of outputs very quickly and it is this speed of computing that encourages their application in practical problems, in particular, the inverse kinematic problem for a two-link robot arm. Characteristics such as accuracy and the effects of network and training set size are discussed. Network accuracy is improved by an iterative process and thus becomes a practical solution for this problem. In general, without careful design ANN methods can have large errors that can cause their application to fail.