Search

Predictive adaptive algorithms are powerful tools for controlling time varying systems. In particular, the Adaptive Generalized Predictive Control is widely used in different plants or processes. Nevertheless, a lack of robustness is often observed especially in case of large uncertainties in the model and in the presence of strong nonlinearities. Better solutions have been proposed using fuzzy modelling. In this paper, a direct adaptive version of the generalized predictive controller is developed. This allows an automatic online and real-time adjustment of local controller settings in order to maintain a proper level of performance. The contributions presented in this paper lie in the merging of local controllers using a fuzzy technique in order to produce a global controller using on-line parameters identification by a recursive lest squares type strategy. The stability of the closed-loop control system is demonstrated through the Lyapunov Stability Theory. Highly improved results are confirmed by simulation on an example of the commonly used Continuous Stirred Tank Reactor system. [ABSTRACT FROM AUTHOR]

In this paper we use the dual Youla parameterisation approach for the robustification of predictive linear control systems using LMI-based constraints. This method based on an approach of convex optimization, is used to robustify the generalized predictive controller GPC algorithm applied to an AutoRegressive Integrated Moving Average (CARIMA) model. The design method consists of two steps. In the first step the synthesis and the robustification of the initial polynomial controller RST-GPC is carried out. This is achieved under constraints on the system output, the control signal and its increment represented as LMI using Youla parameterisation. This controller is then robustified considering frequency and time-domain constraints. In the second step, the new controller is robustified by a dual Youla parameterisation considering frequential and temporal constraints. The effectiveness of this methodology are demonstrated and applied for position control of an induction motor drive. [ABSTRACT FROM AUTHOR]