B.W. Bequette

Model predictive control of blood glucose in type I diabetics using subcutaneous glucose measurements

A constrained model predictive controller is implemented on a simulated type I diabetic patient. A Kalman filter is used to estimate the blood glucose concentration based on a subcutaneous glucose measurement. The model predictive controller returns blood glucose to normoglycemic ranges when subjected to a meal disturbance. The settling time is similar to that of a non-diabetic.

A Dual-Rate Kalman Filter for Continuous Glucose Monitoring

A dual-rate Kalman filter is developed for realtime continuous glucose monitoring. Frequent (5 minute) sampling of a noisy, continuous glucose sensor is used for estimation of glucose and its rate-of-change. Infrequent (8 hour intervals) reference glucose meter samples enable the sensor gain and its rate-of-change to be updated. The dual-rate Kalman filter formulation accounts for uncertainty in both the continuous glucose sensor and the reference glucose meter. The method is tested on simulated and experimental data, confirming its superiority to simple one-point calibration

Estimation-based model predictive control of blood glucose in type I diabetics: a simulation study

A constrained state space model predictive controller, designed based on a 3-state model of the human glucose-insulin interaction, was implemented on a 19-state simulation model of a Type I Diabetic patient. The controller maintained the blood sugar level of the patient plant between the values of 60 and 130 mg/dl, when subjected to a meal disturbance of 50 g, significantly lowering the risk of hypoglycemia/hyperglycemia.

Automated regulation of hemodynamic variables

Experimental studies of two control methodologies for regulating multiple variables in critical care patients are described. The control strategies for the regulation of mean arterial pressure and cardiac output use vasoactive and inotropic drugs. Corresponding experimental results from the evaluation of the controllers with canines are presented.

Multiple model predictive control of hemodynamic variables: an experimental study

A multiple model predictive controller is designed to regulate mean arterial pressure and cardiac output in critical care subjects using inotropic and vasoactive drugs. The algorithm uses a multiple model adaptive approach in a model predictive control framework to account for inter- and intra-patient variability and explicitly handle drug rate constraints. The controller is experimentally evaluated on canines that are pharmacologically altered to exhibit symptoms of hypertension and depressed cardiac output.

A variably tuned multiple model predictive controller based on minimal process knowledge

A multiple model predictive controller is designed using minimal plant knowledge based on the ranges on gains, dominant time constants and time delays. The algorithm uses a weighted multiple model bank of first order plus deadtime models as the prediction model for a constrained model predictive controller. A variable tuning strategy is implemented to improve controller performance. The simulated process IS the isothermal Van de Vusse reaction in a continuously stirred tank reactor (CSTR); this system exhibits input multiplicities, making it a challenging control problem.

Multiple Model Predictive Control: A State Estimation based Approach

An augmented state formulation for multiple model predictive control (MMPC) is developed to improve the regulation of nonlinear and uncertain process systems. By augmenting disturbances as states that are estimated using a Kalman filter, improved disturbance rejection is achieved compared to an additive output disturbance assumption. The approach is applied to a quadratic tank example, which has challenging dynamic behavior, switching from minimum phase to nonminimum phase behavior as the operating conditions are changed.

Behavior of a CSTR with a recirculating jacket heat transfer system

Continuously stirred tank reactors (CSTRs) with a recirculating jacket heat transfer system may have more interesting dynamic behavior than the classical representation of a once through jacket. A particular CSTR with a single steady-state as a function of jacket temperature may have multiple steady-state behavior if the jacket inlet temperature is considered to be the manipulated input. A cascade control strategy with jacket inlet temperature as the secondary variable may not stabilize the reactor, although one with the jacket outlet temperature does. The jacket recirculation flowrate is a bifurcation parameter that determines whether or not there is multiple steady-state behavior.

Optimal estimation of blood glucose

Continuous glucose monitors based on subcutaneous sensors have the potential to improve regulation of blood glucose in individuals, while reducing the risk of hypoglycemia. In this paper we develop an optimal estimation-based approach to trade-off the probability that a sensed subcutaneous glucose change is due to measurement noise versus an actual blood glucose change. This estimator leads to significant improvement in the estimate of blood glucose, as well as its rate-of-change, compared to a finite-differences approach. This in turn allows a better future prediction of blood glucose, allowing the prediction/warning of impending hypoglycemia. This estimation-based approach can also be used to improve the performance of an automated feedback control system (artificial pancreas) to regulate blood glucose based on subcutaneous measurements.

Model predictive control of open-loop unstable cascade systems

In this paper we develop a novel state estimation-based model predictive control approach that has the same general philosophy of cascade control (taking advantage of secondary measurements to aid disturbance rejection). The model predictive control formulation can be applied to open-loop unstable cascade systems and has the additional advantage of being able to explicitly handle constraints. Traditional cascade control strategies are designed for open-loop stable systems, and there is no set procedure to tune cascade controllers for these systems. The example application is a jacketed exothermic chemical reactor, where jacket temperature is used as a secondary measurement in order to infer disturbances in jacket feed temperature and/or reactor feed flow rate. The MPC-based cascade strategy yields significantly better performance than classical cascade control when operating close to constraints on the jacket flow rate.