Feedback Linearisation Using Dynamic Neural Networks

The practical application of feedback linearising techniques has been limited by the requirement for a dynamic model of the plant. This requirement has constrained considerably the number of successful applications. The use of neural networks for feedback linearisation is a relatively recently introduced practice and has the potential to increase the number of applications, as neural network models can be calculated from measured data from the plant, so avoiding the need for deriving a physical model of a possibly complex process.