Design and experimental evaluation of a data-driven PID controller using cerebellar memory

Abstract

In industrial process control, the proportional–integral–derivative (PID) control scheme is well-recognized and widely utilized. However, due to the distinctive characteristics of real systems, their control design primarily aims at achieving optimal production performance, constrained by uncertainty and variations. This paper initially discusses a database-driven PID (DD-PID) control scheme that was previously proposed. This scheme combines the DD-PID with the cerebellar model articulation control to minimise computational and memory requirements for industrial application. Subsequently, a hydraulic system is introduced, detailing its characteristics and control necessities. Furthermore, both the DD-PID and the proposed cerebellar model articulation control memory-based DD-PID control schemes are implemented and evaluated through experimental examples on a hydraulic system. Lastly, as a practical validation of the theoretical approach, a quantitative assessment compares the two methods, discussing the practicality and efficacy of the proposed scheme in reducing computation and memory consumption.