101 / 2025-04-17 14:05:07

Reset Observer with Gaussian-Type Adaptation for Entropy-based Fault Detection

fault detection,reset observer

This paper introduces a novel entropy-based fault detection methodology for a class of nonlinear systems subject to potential actuator faults, nonlinear dynamics, and external disturbances. A key contribution lies in the development of an enhanced reset observer that identifies uncertain system parameters via a Gaussian-type learning coefficient-based adaptive law. This adaptive mechanism distinguishes itself through accelerated convergence properties, while the proposed entropy-driven fault detection algorithm ensures timely and false-free fault identification by extracting abnormal information from output residuals and real-time parameter estimates. The integration of entropy metrics with adaptive learning not only suppresses false alarms but also enhances detection sensitivity. A simulation example is given to demonstrate the effectiveness of theoretical findings.