Tensor Product model based PID controller optimisation for propofol administration

Abstract

This study investigates the computer-regulated propofol administration in anesthesia during medical interventions considering output feedback and robust PID control. The paper applies the Affine Tensor Product Model Transformation to derive the appropriate polytopic quasi-LPV representation of the closed-loop dynamics. This model form enables the use of LMI-based optimisation techniques to evaluate the closed loop performance. Despite the highly non-convex nature of this output feedback problem, the PID gains can be locally tuned through simplex optimisation. The proposed method provides a systematic way of tuning PID-controlled propofol administration for individual patients with theoretically established worst-case performance measures.