Realisation of a Riccati equation-based controller using gradient-type neural networks

This paper presents a solution to an algebraic Riccati-matrix equation-base d robust control law using a set of gradient-type neural networks. The prop osed neural network solves a representative Riccati-matrix equation, which is commonly encountered in robust control problems. The class of neural net works is a variant of the continuous-time Hopfiled network. To verify the p roposed feedback control scheme in real-time applications, a high-speed dig ital signal processor has been used to emulate the network operations. This allows an on-line implementation that is adaptable to system parameter cha nges. Finally, illustrative examples show the potential of simulation under hard real-time conditions. (C) 2001 Elsevier Science Ltd. All rights reser ved.