Ramesh R. Rao

Model Predictive Control of Drug Infusion Systems

Publications | Presentations | Simulation & Expt. Software Dowloads

Critical care patients such as those in intensive care or undergoing surgery require close monitoring of hemodynamic variables. Physicians maintain patient states within acceptable operating ranges by infusing several drugs and/or intravenous fluids. For example, sodium nitroprusside (SNP) and phenylephrine (PNP) are used in regulation of mean arterial pressure (MAP) and dopamine and intravenous fluids are used for increasing cardiac output (CO). In addition, they may be required to administer anesthetics and monitor the depth of anesthesia (DOA) during surgical procedures. Current clinical practices usually involve manual regulation of drip IV lines administering the drugs. Programmable pumps are also used to either deliver the drugs at a constant rate or a variable rate to achieve a desired concentration. Such pumps are based on averaged pharmacokinetic data and are essentially open-loop (i.e) there is no feedback of the patient's states and require regular intervention to adjust the drug flow rates by the attending physician or nurse. It is desirable to have an automated system that closes the loop on primary variables, but monitors secondary variables and performs diagnostics. This allows the physician to spend more time monitoring the patient for conditions that are not easily measured, and assures that the physician is always "in the loop".

The overall control objective is to maintain two hemodynamic variables, mean arterial pressure (MAP) and cardiac output (CO), at desired setpoints by automated infusion of inotropic and vasoactive drugs. SNP is administered for arterial vasodilation. Dopamine (DPM) is used as an inotrope to enhance cardiac performance; Phenylephrine (PNP) is used for arterial vaso-constriction. The controllers was initially evaluated and tuned in closed-loop simulations using a elaborate non-linear canine circulatory model as the "patient" before moving to the experimental phase.

In this research we developed a novel approach combining the MPC and MMAC strategies for regulation of hemodynamic variables. A multiple model strategy is used to provide a prediction model for an MPC framework. This approach has the combined advantage of allowing model adaptation to handle inter- and intra-patient variability and the ability to handle explicit input and output constraint specifications often desired by the critical care physicians. The controller was demonstrated in experiments on three canines that were pharmacologically altered to exhibit hypertension and depressed cardiac states.

Advanced Control of a Glass Cooling Forehearth

Publications | Presentations

Cooling forehearths are used in glass manufacturing processes. Most forehearths cool the flowing glass from high temperatures where melting and refining take place to lower temperatures required for forming. Due to the dependence of the glass flow rate (pull) on the viscosity, which in turn depends on the temperature, some cooling forehearths actually operate at a point that would be unstable in the absence of control. This paper shows that the unstable operating point is one of three operating points, the other two being stable but undesirable. This situation is similar to that of the exothermic continuous stirred tank chemical reactor with a cooling jacket.

After demonstrating the instability of the system, three control approaches are applied to a generic three zone cooling forehearth which is represented by a nonlinear multi-stage lumped model derived from first principles mass and energy balances. The most common control strategy in current manufacturing processes is multiple PI loops, which often have adequate performance at constant operating conditions. A change in the desired glass product results in a change in temperature profile setpoints. The multiple PI loop strategy often requires a substantial amount of time before the new desired operating condition is achieved.

Better control performance for operating condition changes is achieved by model-based control. The two model-based approaches used in this paper are Multivariable Generalized Predictive Control (GPC) and Extended Kalman Filter-based Nonlinear Model Predictive Control (EKF-based NMPC). Unconstrained multivariable GPC is coupled with Recursive Least Squares (RLS) to provide better performance than PI, at the cost of a significantly more complicated algorithm and the requirement of a disturbance phase during which the process is identified. NMPC with an EKF-based state estimation strategy allows modeling and estimation of parameters and unmeasured disturbances as appended states, while explicitly handling constraints. The advantages and disadvantages of the three control strategies are compared.

Gain Scheduling

Publications | Presentations

Gain scheduling is a prevalent method used to design controllers for systems with widely varying nonlinearities. Autotuning is a convenient tool often used for tuning PID-based controllers local to a region of operation. The proposed work explores non model--based approaches to integrate autotuning methodologies in a gain scheduled framework and analyze the associated interpolation, switching and bumpless transfer strategies. The study will focus on chemical processes that require operation near regions of zero or infinite process gain and/or those exhibiting input/output multiplicities. A stability analysis based on numerical determination of regions of attraction is also proposed.