Discrete-time prescribed performance control with switchable structure: Experimental validation in a system with unknown dynamics and constraints

Abstract

This paper proposes an adaptive Prescribed Performance Control (PPC) framework for discrete-time systems with unknown dynamics, subject to practical input–output constraints, strong nonlinearities, and abrupt disturbances. The proposed controller explicitly addresses two operating conditions: ensuring the tracking error remains within predefined funnel boundaries and restoring performance when violations occur. A hybrid control law is directly derived using an adaptive network, called the Multi-input Fuzzy Rules Emulated Network (MiFREN), in conjunction with a switchable tracking error and its transformation. This design incorporates switching logic and adaptive learning mechanisms and is theoretically validated by selecting parameters directly from the input–output characteristics based on experimental data, without relying on a mathematical model. The scheme avoids singularities near the funnel boundary by introducing a switching threshold and reduces constraint violations while maintaining system stability. Experimental validation on a DC motor torque control system demonstrates the controller’s effectiveness, showing superior tracking accuracy, disturbance rejection, and energy efficiency compared to existing methods.