Comparative Control of a Nonlinear First Order Velocity System by a

Abstract

Neural Networks, due to their approximation capabilities of Multilayer Perceptron (MLP) are promising to become a popular tool for modeling nonlinear systems and implementing general – purpose nonlinear controllers. One of them, for prediction and control, is the NARMA–L2 (or Feedback Linearization) controller. In this work its capabilities are tested, in a first order velocity control system, and compared with classic PID control. The comparison between system responses, clearly showed that NARMA-L2 controller gives the best control results for both kind of inputs (step or ramp), hereby minimising the steady state final errors of response and, at the same time, improving its velocity. Ill. 11, bibl. 10 (in English, Summaries in Lithuanian, English, Russian).