Design of model reference adaptive-PID controller for automated portable duodopa pump in Parkinson's disease patients

摘要

This paper focuses on the design of an automated portable Duodopa pump (PDP) for Parkinson's disease (PD) patients, with the aim of achieving robust performance on intra-day, intra-patient and inter-patient variability under various disturbances, parameter variations and uncertainties. For achieving this objective a suitably modified Hacisalihzade's pharmacokinetics-pharmacodynamics (PK-PD) model of Levodopa has been chosen. By using Bernoulli's equation for pressure balance and Hagen-Poiseuille law for flow rate, a positive displacement piston type drug dispensing syringe connected to PEG-J (percutaneous endoscopic gastrostomy- Jejunal) tube is modeled under the assumption of Newtonian, incompressible and laminar flow properties. A PDP is proposed employing model reference adaptive-PID (MRA-PID) as the control strategy and the closed-loop responses are compared with a PID controller tuned by particle swam optimization (PSO-PID). In MRA-PID controller an impeccable tracking of reference model is accomplished by a parameter adaptation mechanism such as a gradient decent method named MIT rule. The overall performances of both the controllers are evaluated under two different perturbation protocols as well as sensor noise and the robust stability is analyzed theoretically for structured uncertainties in system gain and time constant by using a cohort of eight in- silico subjects with large inter-patient variability. Even though the responses from both the controllers are stable with zero steady state error and zero undershoots, the dynamic performances of proposed controller is superior and safe under unannounced meal disturbances, intra and inter patient variability.