Optimal PD-type networked iterative learning algorithm based fault estimation for repetitive systems with delays, packet losses, sensor saturation and sensor failure

For a class of repetitive networked control systems (NCSs) with packet loss, network-induced delays, bounded disturbances, additive sensor saturation constraint and sensor failure, an optimal proportional derivative-type (PD-Type) iterative learning algorithm (ILA) based sensor fault estimation (FE) is designed with the aim to evaluate and estimate the effect of sensor fault on system between every iteration. To do so, state variables, Markov chain process of random packet losses, network-induced delays, bounded disturbances, additive sensor saturation constraint and sensor failure are introduced to establish an extended-state-space system model. Then, based on this model, the iterative learning algorithm ILA based sensor FE is designed. Using the linear matrix inequalities (LMIs) technique for linear repetitive processes, sufficient conditions are developed with the Lyapunov Krasovskii technique and H ∞ approach to calculate the iterative learning gain matrices and the observer gain matrix. In addition, MATLAB optimization based on YALMIP is applied to improve the performance of proposed scheme. Finally, the feasibility and effectiveness of the proposed design method is illustrated on a networked dynamic hydro-turbine governor system based on Matlab/Simulink and TrueTime toolbox.