On sampled-data implementations of feedback linearizing controllers

The authors present a case study for designing a feedback linearizing controller of a nonlinear continuous-time system in the presence of sampling. Among the methods considered are exact linearization ignoring the effects of sampling, exact linearization with extended feedback to compensate for sampling, and a multirate sampling scheme implemented via a neural network. The performance of the three controllers are compared, through simulation, for a simple example from the literature. The main innovation presented is the neural controller designed using multirate sampling which gives superior performance at low sampling rates