Learning-Based Adaptive Estimation with Guaranteed Prescribed Performance for Nonlinear Sandwich System Subject to the Quantised Sensor

High-quality identification schemes are essential for controller design of nonlinear systems in the field of automatic control. In control communities, the prescribed performance technology has received wide attention as a promising methodology because of its quantitative description of the steady-state and transient performance of system control over the past several years. The improved transient performance of the parameter estimation is widely known to enable simplified controller design, reduce system regulation time, and prevent system divergence. However, in system identification communities, few studies on the transient performance of parameter identification have been published because of the difficulties in designing an error variable that reflects the transient performance of parameter identification. To address this problem, this study provides a solution by integrating the prescribed performance technique into the estimator design. In this study, we introduced an adaptively prescribed performance parameter identification for nonlinear sandwich systems subjected to quantitative observations. First, a low-pass filter and forcing variables are developed to construct an identification error expression. Subsequently, an improved prescribed performance function that characterises the error bound of parameter estimation is introduced. Second, the error transformation concept is used to obtain a new system by transforming the raw system such that constraint conditions are avoided. A novel adaptive law is proposed to guarantee original parameter identification using the prescribed performance. Finally, simulation and process examples are presented to illustrate finding results.