AI-PID Control for Suppressing the Effects of Sudden Disturbances

A quality-related problem at manufacturing sites is the occurrence of sudden disturbances in the control system caused by variation and fluctuation in materials and machinery. Sudden disturbances cannot be sufficiently suppressed by the proportional-integral-differential (PID) control method that has come to be widely used at industrial sites. We therefore propose a method that extends PID control by adding compensation using a machine-learning model to suppress the effects of sudden disturbances. In an experiment using actual equipment, the proposed method reduced the convergence time of instability in the controlled variable due to a sudden disturbance by 90% on average relative to the convergence time for conventional PID control.