Case Studies

Abstract

This chapter presents three case studies that illustrate the identification and control strategies presented in this book. First, in Section 7.1, a dynamic neural network model is trained based on measurements taken from a pressure pilot plant, and the effects of correct initialisation of the hidden states of the network both during training and during usage, are highlighted. The globally linearising control strategy discussed in Chapter 6 and based on the identified dynamic neural network model was applied on-line to the pressure plant. Secondly, Section 7.2 presents a case study involving the approximate feedback linearisation of a simulated single link manipulator. Finally, Section 7.3 presents a case study involving identification and approximate feedback linearisation and decoupling of a simulated evaporator process.