Identification of a Nonlinear Pneumatic Servo System Using Modular Neural Networks

Abstract

Sometimes, in the case of highly nonlinear systems the traditional approaches of identification and control could be difficult to implement. In this case, a good alternative are the neural networks. In this paper a modular neural network for the identification of a pneumatic servo system is proposed. This approach is based on the partitioning of static characteristic of the pneumatic system. The neural modules are implemented with multilayer neural networks.