-
[show abstract]
[hide abstract]
ABSTRACT: Run-to-run control has been applied to several traditional batch processes in the chemical industry. The 24-h cycle of eating meals, measuring blood glucose concentrations, and delivering the correct insulin bolus, with the goal of achieving the optimal blood glucose profile, can be viewed in the same spirit as traditional batch processes such as emulsion polymerization. In this paper, we aim to exploit the "repetitive" nature of the insulin therapy of people with Type 1 diabetes. A run-to-run algorithm is used on a virtual diabetic patient model to control blood glucose concentrations. The insulin input is parameterized into the timing and amount of the dose while the glucose output is parameterized into the maximum and minimum glucose concentrations. Robustness of the algorithm to variations in the meal amount, meal timing, and insulin sensitivity parameter is addressed. In general, the algorithm is able to converge when the meal timing is varied within ±40 min. If the meal size is underestimated by approximately 10 grams (g), the algorithm is able to converge within a reasonable time frame for breakfast, lunch, and dinner. If the meal size is overestimated by 20-25 g, the algorithm is able to converge. When random variations in the meal timing and the meal amount are introduced, the variation on the output variables, G<sup>max</sup> and G<sup>min</sup>, scales according to the amount of variation allowed. Along with this, the insulin sensitivity of the virtual patient model is varied. The algorithm is robust for differences in insulin sensitivity less than 50% of the nominal value.
IEEE Transactions on Biomedical Engineering 07/2006; · 2.28 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: Summary form only given. Understanding regulation is a critical hurdle in unraveling complex biological systems. As gene-level architectures become known, the open challenge is to assign predictable behavior to a known structure, the so-called "genotype-to-phenotype" problem. In response to this challenge, the discipline of systems biology has emerged with an integrative perspective towards determining complex systems behavior. A property of particular interest is the robustness of the biophysical network: the ability to maintain some target level of behavior or performance in the presence of uncertainty and/or perturbations. In biological systems, these disturbances can be environmental (heat, pH, etc.) or intrinsic to the organism (changes in kinetic parameters). While preliminary results are available for simple (low-dimensional, deterministic) biological systems, general tools for analyzing these tradeoffs are the subject of active research. The gene network which underlies circadian rhythms is an ideal system for robustness studies, owing to its remarkable performance in a highly uncertain environment. Of interest for control theoretic analyses, the dominant elements of the postulated architecture for Drosophila consist of nested negative autoregulatory feedback loops controlling the expression of timeless (tim) and period (per) interlocked with a positive feedback loop established via the dClock gene. Complex formation, regulated translocation and degradation of several of these gene products, which is additionally controlled (and delayed) by protein phosphorylation, add further levels of complexity to the system. In this talk, a number of quantitative tools from systems theory are presented as enabling methodologies for unraveling robust biological regulatory systems, with an emphasis on sensitivity analysis. Our work on modeling and analysis of the Drosophila circadian rhythm gene network are detailed, and generalizations are be drawn for the mammalian analog and for more general gene regulatory networks.
Control, Communications and Signal Processing, 2004. First International Symposium on; 02/2004
-
[show abstract]
[hide abstract]
ABSTRACT: Rigorous reachability analysis is largely an open problem for nonlinear distributed systems. A programming-based reachability analysis is proposed to address nominal and perturbed systems, with the final distributed variables characterized as multi-Gaussian distributions. An application study is detailed for the feedback control of particle size distribution in emulsion polymerization, where the particulate system is described by population balance models. The approach has been applied to a semibatch styrene emulsion polymerization process. The re-mapping of the reachable region in the case of mid-course correction shows the best performance achievable through in-batch control action.
American Control Conference, 2003. Proceedings of the 2003; 07/2003
-
[show abstract]
[hide abstract]
ABSTRACT: This article discusses closed-loop blood glucose regulation
algorithms that use the intravenous route for insulin delivery to
insulin-dependent diabetic patients. Classical control methods and
advanced algorithms using implicit knowledge or explicit models
(empirical, fundamental, or “gray-box”) of the diabetic
patient are examined. Current research on characterizing patient
variability is presented, in the context of a model predictive
controller able to adjust to changes in patient glucose and insulin
sensitivity
IEEE Engineering in Medicine and Biology Magazine 02/2001; · 2.06 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: The control of particle size distribution (PSD) in semi-batch
emulsion polymerization necessitates the identification of the best
manipulative variables, with a profound influence on the PSD. The
candidate variables include the feed rate of the monomers, surfactants
and initiators. The monomer feed predominantly affects the growth rate,
but could also influence nucleation rates. The surfactant feed
influences the rates of nucleation and coagulation. The initiator feed
could influence the rates of nucleation and growth. This paper describes
the effect of these manipulative variables on the evolution of PSD, as
studied in an experimental reactor system
American Control Conference, 2001. Proceedings of the 2001; 02/2001
-
[show abstract]
[hide abstract]
ABSTRACT: This work presents a hybrid modeling approach for batch-to-batch
optimization of particle size distributions (PSD) in semi-batch emulsion
polymerization. A fundamental population balance model describing PSD
evolution is augmented by a Partial Least Squares (PLS) model. This
hybrid model is then embedded in a successive quadratic program (SQP) to
design surfactant and initiator input trajectories that drive the
process to a target PSD. PLS using a moving batch data history, in which
earlier batches are dropped from the data set as more recent batches are
added, leads to superior convergence compared with an expanding data set
American Control Conference, 2001. Proceedings of the 2001; 02/2001
-
[show abstract]
[hide abstract]
ABSTRACT: An approach to insulin injection therapy is introduced that draws from recent ideas in run-to-run control of batch chemical reactors. The method relies on measurements of blood glucose and does not require a mathematical model. Results are shown for simulation case studies involving a detailed pharmacokinetic model.
Engineering in Medicine and Biology Society, 2001. Proceedings of the 23rd Annual International Conference of the IEEE; 02/2001
-
[show abstract]
[hide abstract]
ABSTRACT: Modeling of particulate system is complicated by the wide range of
mechanisms that affect particles of various sizes. Even when detailed
fundamental models are available, they are often intractable for online
control. We use an approach intermediate between first principles models
and empirical modeling. By treating each of the processes of growth,
nucleation, and aggregation as a separate functionality identified
within the structure of the population balance, we create a model suited
for online control in accuracy, tractability and availability. Express
acknowledgment of a distributed framework allows the model to capture a
broad range of behavior accurately. Relative simplicity facilitates
rapid simulation. We demonstrate the application of an inverse problem
technique for identification of growth and nucleation dependences to two
examples. The crystallization example shows extraction of size dependent
growth and nucleation rates from simulated experimental data. Both rate
laws are determined from a single experiment. The emulsion
polymerization example shows the use of the technique for model
reduction and demonstrates how a growth mechanism change in time can be
handled
American Control Conference, 2000. Proceedings of the 2000; 02/2000
-
[show abstract]
[hide abstract]
ABSTRACT: An inline evaporation system for depositing Cu(InGa)Se<sub>2</sub> films from elemental sources was characterized. The system demonstrated reproducibly good deposition uniformity and device performance for translation speeds ranging from 1 to 2.5 inches/min. A source effusion model predicting film thickness, composition, and compositional gradients was developed and tested. The effusion model was reasonably successful in predicting measured film characteristics. The model can be further improved by a more accurate description of the flux profiles. In addition, effusion in the transitional flow regime and the possibility of Ga-ln interdiffusion need to be explored. Such an improved model would be a valuable tool for the design and development of commercial scale systems, as well as for improved efficiency and flexibility in manufacturing
Photovoltaic Specialists Conference, 2000. Conference Record of the Twenty-Eighth IEEE; 02/2000
-
[show abstract]
[hide abstract]
ABSTRACT: This work presents numerical studies of solutions to population
balance models for latex particle size distributions, using styrene
emulsion polymerization as a prototype. Emphasis is placed on control
relevant issues such as computational speed and particle size
distribution controllability. Simulation results using fixed and moving
grid discretizations of the population balance equations are presented.
Simulation experiments reveal controllability behavior that parallels
recent theoretical results for controllability of particle size
distributions in emulsion polymerization
American Control Conference, 1999. Proceedings of the 1999; 07/1999
-
[show abstract]
[hide abstract]
ABSTRACT: A controller which employs dynamically scheduled control parameters has been introduced in earlier work by the authors (1998). This dynamic gain scheduling (DGS) algorithm is extended for multivariable processes and systems with arbitrary relative degree. The multi-input multi-output DGS controller is demonstrated on an industrial benchmark problem-a solution copolymerization process. Simulations show that the DGS controller is significantly better than a four PI controller system, and compares favorably to a 3×3 model predictive controller
American Control Conference, 1998. Proceedings of the 1998; 07/1998
-
[show abstract]
[hide abstract]
ABSTRACT: Two synthesis methods for constrained nonlinear systems are
applied to systems with unmeasured disturbances. The first approach, an
extension to nonlinear internal model control (NLIMC), solves an
instantaneous minimization to ensure nominal performance. The second
approach utilizes Lyapunov theory to obtain a controller which
guarantees bounds on the system input and output for bounded unmeasured
disturbances. The utility of both approaches is demonstrated for a CSTR
case study
American Control Conference, 1997. Proceedings of the 1997; 07/1997
-
[show abstract]
[hide abstract]
ABSTRACT: This paper presents an application of a customized linear
programming (LP) based model predictive control strategy to the paper
machine cross direction (CD) control problem. The objective of CD
control is to maintain flat profiles of variables of interest by
minimizing worst case deviations from setpoints (defects). These control
problems can have as many as 200 actuators (inputs) and 400 sensor
measurements (outputs). This large size coupled with the stringent
real-time requirement of computing a control move in a few seconds poses
a very challenging control problem. Computational results that
demonstrate the effectiveness of this strategy are presented. For
typical disturbances this algorithm can compute provably optimal control
moves for a 400 input×400 output control problem in approximately
5 seconds versus approximately 90 seconds for a generic LP algorithm on
a HP 9000/770 workstation
American Control Conference, 1997. Proceedings of the 1997; 07/1997
-
[show abstract]
[hide abstract]
ABSTRACT: Traditionally, arterial blood pressure, heart rate,
atrial-ventricular conduction and myocardial contractility have been
considered to be solely governed by the central nervous system. This
convention is now challenged by the recent converging evidence that
cardiac ganglia around the heart may have complex integrative function.
Using neuronal tracing and laser confocal microscopy, we have mapped the
local neural circuitry around the heart and proposed a local reflex
loop. In order to study the functionality of this circuitry, we use a
biophysical model to simulate the behavior of the proposed local reflex
loop. The preliminary results show that this local cardiac reflex loop
predicts behavior that is consistent with the Bainbridge reflex in the
sense that venous infusion increases the heart rate without obvious
change of the system blood pressure
Engineering in Medicine and Biology Society, 1997. Proceedings of the 19th Annual International Conference of the IEEE; 02/1997
-
[show abstract]
[hide abstract]
ABSTRACT: Presents a comparative analysis of various pancreas models,
β-cell models, and glucose control algorithms based on
steady-state, dynamic, and frequency responses. The approach highlights
desirable aspects of individual models as controllers for insulin
delivery devices. Behaviors such as rapid response to hypoglycemia,
derivative response to glucose concentration increases and constrained
insulin delivery are advantages to be integrated into a glucose control
algorithm, whereas large amplitude ratio at high frequency and
over-aggressive insulin delivery are undesirable behaviors to be
avoided. A better understanding of these systems provides the means to
reverse-engineer the biological objective function of the pancreas by
identifying individual model components representative of true
physiological system responses
Engineering in Medicine and Biology Society, 1997. Proceedings of the 19th Annual International Conference of the IEEE; 02/1997
-
[show abstract]
[hide abstract]
ABSTRACT: A controller design strategy for nonlinear systems with two manipulated inputs and one controlled output is proposed. The technique is called “habituating control” as a result of its similarity to control schemes commonly used in biological systems. A nonlinear control law that provides input-output linearization while simultaneously minimizing the “cost” of affecting control is derived. The cost function employed differs according to the relative degrees of the two inputs. Local stability analysis shows that the habituating controller can provide a simple solution to the singularity and nonminimum phase problems. The method is evaluated using a nonlinear chemical reactor model
Control Applications, 1996., Proceedings of the 1996 IEEE International Conference on; 10/1996
-
[show abstract]
[hide abstract]
ABSTRACT: A control synthesis scheme is presented for nonlinear
single-input-single-output systems which have completely unstable
(antistable) zero dynamics. The approach is similar in spirit to linear
approaches for nonminimum phase systems and involves the derivation of
an input-output linearizing controller for a suitably-defined nonlinear
minimum phase approximation to the original system. The linearizing
controller achieves an approximately linear input-output response and
internal stability
IEEE Transactions on Automatic Control 03/1996; · 2.11 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: The authors present an algorithm for controlling blood glucose in
the Type I diabetic patient using model predictive control (MPG) of a
closed-loop insulin infusion pump. The controller is designed from a
detailed 19th order model of the human glucose-insulin system. Input
constraints, based on insulin pump and physiological requirements, are
enforced. The controller is challenged with a hyperglycemic diabetic
patient (setpoint tracking), and a controlled diabetic ingesting a meal
(unmeasured disturbance rejection). To improve both robustness and
performance, a new controller using MPC with state estimation is
synthesized. Compared to the non-state estimating controller, the new
controller results in a 40% improvement in overshoot, to a maximum
hypoglycemic deviation of 7.9 mg/dl, and a 23% decrease in settling
time. These results demonstrate the potential use for this algorithm in
a closed-loop insulin infusion pump
Engineering in Medicine and Biology Society, 1996. Bridging Disciplines for Biomedicine. Proceedings of the 18th Annual International Conference of the IEEE;
-
[show abstract]
[hide abstract]
ABSTRACT: An approach to dynamic scheduling is presented which draws inspiration from a biological control mechanism (the baroreceptor reflex). A simplified model of the reflex is proposed which validates known experimental results, and the nonlinear "scheduling" mechanism is analyzed for applications in automatic control.
American Control Conference, 1994;
