Robust data-driven model-free adaptive control to uncertainties and disturbances for NARMAX systems

Tite: Robust data-driven model-free adaptive control to uncertainties and disturbances for NARMAX systems (DOI)
​​​​​​​Authors:  Mohsen Heydari, Alireza B Novinzadeh, Morteza Tayefi.
Journal: Transactions of the Institute of Measurement and Control
​​​​​​​  ​​​​​​​
Abstract: This paper introduces an enhancement to the model-free adaptive control (MFAC) method based on a new control input cost function for full-form dynamic linearization (FFDL) for NARMAX systems. The new control cost function includes the one-step-ahead tracking error, OER, and output error integral. The main feature of the approach is that it has a high ability to overcome disturbances in the control input and measured noise in the output. Analytical and extensive simulations have demonstrated that the novel control law exhibits superior response and robustness in comparison to the prototype MFAC. Moreover, it guarantees stability and convergence of the tracking error by enforcing bounded-input bounded-output (BIBO) criteria. In addition, 20 Monte Carlo simulations were conducted to test the initial conditions and random control parameters, and the results demonstrated that the proposed control outperformed the prototype MFAC.