Modular neural networks for friction modeling and compensation

The modular neural network has been shown to be an effective alternative to multilayer feedforward networks, especially for implementing functions with sharp changes. The objective of this work is to use modular neural networks to model precision pointing systems whose performance is limited by nonlinear friction forces. The modular neural network models are used to develop friction compensation controllers. This paper also describes a new method for training modular networks, based on the Levenberg-Marquardt algorithm for nonlinear least squares. The algorithm is tested on several function approximation problems, and the performance is compared with standard steepest ascent and the Rprop algorithm