Model-Free Adaptive Control for Nonlinear Systems Under Dynamic Sparse Attacks and Measurement Disturbances

In this paper, the tracking control problem is studied in the model-free adaptive control (MFAC) framework for a class of discrete-time single-input single-output nonlinear systems affected by dynamic sparse attacks and measurement disturbances. The system outputs are measured by multiple sensors, but an attacker can manipulate nearly half of the sensors simultaneously in a time-varying manner. First, considering the communication burden caused by multiple sensors, a voting-based event-triggered mechanism is introduced to minimize data transmission under attacks. The triggering condition is designed according to tracking performance so that the system is updated only at the triggering instants while maintaining satisfactory control performance. Then, to minimize the effects of measurement disturbances and dynamic sparse attacks on the control performance of the MFAC algorithm, two data fusion algorithms are developed to estimate the system output from the transmitted data. Moreover, an event-triggered extended state observer is designed to mitigate the negative impact of nonlinear residual terms caused by estimation errors on the MFAC algorithm, and based on this, a controller that updates only at the triggering instants is designed. Finally, simulation examples confirm the effectiveness of the proposed MFAC algorithm.