Data-based real-time compensation control: With application to high-precision motion systems

This article proposes a novel data-based real-time compensation control (DBRTICC) method for high-precision motion control systems. The method addresses the challenges of simultaneously achieving high-accuracy tracking, strong task flexibility, and robust disturbance rejection. DBRTICC integrates a real-time compensation control (RTICC) strategy with an offline update mechanism for the feedforward controller. Specifically, a system state prediction model is established based on system identification results. Based on the predicted states, compensation signals are generated in real time and directly injected into the plant. The RTICC strategy mitigates the impact of system uncertainties on tracking performance and further reduces the tracking error. However, the RTICC strategy cannot guarantee that the feedforward controller parameters are optimal for varying tasks. To further suppress tracking errors induced by reference variations, this article proposes a data-based feedforward parameter update (DBFPU) mechanism. The mechanism exploits an analytical expression of the residual tracking error after real-time compensation, feedback control signals, and information from adjacent varying tasks. Then, the feedforward controller parameters are updated by solving first-order optimality conditions. Notably, DBFPU applies the feedforward signal to the system, eliminating reference-induced tracking errors for RTICC. Meanwhile, RTICC provides DBFPU with more accurate updating information, ensuring the correctness of the parameter update direction. Finally, experiments validate that DBRTICC achieves high tracking accuracy while maintaining strong task flexibility and robust disturbance rejection.