Eva Miklovičová

Adaptive Control of Time-Delayed Systems with Application for Control of Glucose Concentration in Type 1 Diabetic Patients

Abstract An adaptive controller for glucose control in Type 1 Diabetes Mellitus (T1DM) subject is presented in this paper. The proposed control model of T1DM subject involves a known input time-delay, due to the modeling of a subcutaneous tissues, and a disturbance submodel, where a meal ingestion acts as a measured disturbance. A main MRAC based part of controller for time-delayed systems is supplemented with a heuristic based adaptive disturbance rejection. The controller is verified by means of numerical simulations using an own implementation of T1DM simulator reported in literature.

Model Reference Adaptive Control of Glucose in Type 1 Diabetics: A Simulation Study

Abstract Paper presents the model reference adaptive control applied for the glucose concentration control in Type 1 diabetes mellitus (T1DM) subject. The adaptive controller structure allows to present the commanded insulin infusion by means of the basal infusion rate and the bolus insulin doses. T1DM simulation model is adjusted so that the simulated output corresponds to the particular data logged in a diabetic diary. These facts have allowed to compare the obtained results with the data logged in the diary.

T1DM subject as a time-delay system: Modeling and adaptive control

Type 1 Diabetes Mellitus (T1DM) subject model is presented in this paper. The model consists of two parts. A glucose and insulin plasma kinetics is inferred from a minimal model of glucose kinetics commonly used to analyze the results of an intravenous glucose tolerance test (IVGTT). A subcutaneous insulin and a subcutaneous glucose kinetics are modeled as a time-delay. A meal announcement information is considered as a measured disturbance signal and a disturbance model is also proposed. The presented model serves as a base for the design of an adaptive control algorithm for automatic normoglycemia maintaining in T1DM subject. The adaptive control algorithm is also briefly presented in the paper.

Individualized T1DM simulator for verification of adaptive controller

Individualized type 1 diabetes mellitus (T1DM) subject model is presented in this paper. Insulin-glucose subsystem based on Bergmans minimal model is coupled with subcutaneous insulin absorption and absorption of digested carbohydrates. Identification of model parameters was performed on pharmacokinetics and pharmacodynamics characteristics of administered insulin and data collected from continuous glucose monitoring (CGM) system. The identified model served as a basis for designing a model reference adaptive controller.

REDUCING THE EFFECT OF UNMODELED DYNAMICS BY MRAC CONTROL LAW MODIFICATION

Abstract Solving a tracking problem does not always give desired results even when the adaptive control methods are used. Some difficulties may occur when the apriori assumptions laid down for the problem solution are not satisfied. One of the serious issues is the existence of unmodeled dynamics in the tracking problem. The proposed solutions are mainly based on robustification of the adaptation law. In this paper we propose to reduce the effect of unmodeled dynamics using the MRAC control law modification so that the standard adaptation law ensures the sufficiently small tracking error.

Remarks on models for estimating the carbohydrate to insulin ratio and insulin sensitivity in T1DM

In this paper we estimate linear models for prediction of the interstitial glucose concentration in response to meals and bolus insulin. Parameters of these models can be directly used in simple bolus calculation rules. In contrast to models proposed in the literature, we present a model without an integrator. This model maintains the benefits of the existing empirical models and allows simulation of a longer time period than the post-prandial period, i.e. the couple of hours following a meal. Furthermore, the new model proposed in this paper does not require any re-initialization before meals.

MRAC with reduced state information

Abstract Direct adaptive control algorithms using state variables in control structure are known for their good adaptation capability and tracking performances that result from the fact that they use plant state variables in control law as well as in adaptation law. The implementation of standard model reference adaptive control (MRAC) with state variable control structure requires some a priori knowledge about the plant to be controlled including the plant order. This information is crucial for the proper choice of reference model describing the desired closed loop dynamical behavior and consequently for the adaptive system performances. The aim of our paper is to propose the fuzzy adaptation law for MRAC with state variable structure of control law that is able to ensure the adaptation process convergence and tracking capability even in the presence of unmodelled dynamics.

Model reference fuzzy adaptive learning system

Abstract Fuzzy control has revealed as a practical alternative to several conventional control schemes since it has shown good results in some application areas. However, there are several drawbacks of this approach: (i) the design of fuzzy controllers is usually performed in an ad hoc manner where it is often difficult to choose some of the controller parameters (e.g., the membership functions), and (ii) the fuzzy controller constructed for the nominal plant may later perform inadequately if significant and unpredictable plant parameter variations occur. A “learning system” possesses the capability to improve its performance over time by interacting with its environment. A learning control system is designed so that its “learning controller” has the ability to improve the performance of the closed-loop system by generating command inputs to the plant and utilizing feedback information from the plant. Learning controllers are often designed to mimic the manner in which a human in the control loop would learn how to control a system while it operates. Some characteristics of this human learning process my include: (i) after learning how to control the plant for some operating condition, if the operating conditions change, then the best way to control the system may have to be relearned; (ii) a human with significant amount of experience at controlling the system in one operating region should not forget this experience if the operating condition changes. To mimic these types of human learning behavior, we introduce strategy that can be used to learning controller onto the current operating region of the system.

Model Reference Adaptive Control-Some Development Trends

Abstract The aim of the work is to present the state-of-the-art in this filed of adaptive control. Some development trends based especially on the general stability theory are analyzed. The characteristics of basic model reference adaptive control (MRAC) structures and algorithms are given in order to provide an insight into the development of the theory and to indicate the possibilities of their practical application. As the paper makes only a survey of the subject area, the stability proofs of the presented algorithms are omitted. The plant to be controlled is assumed to be described by a linear model.

Model Reference Adaptive Control of Nonlinear Systems with Canonical Structure

Abstract Model reference adaptive control (MRAC) has been developed in many modifications. The stability proofs have been given under the assumption that the plant model is linear. However, MRAC approaches give often convergent solutions even for nonlinear systems. The aim of this paper is to show that for nonlinear systems with the model in canonical form, the standard MRAC with state feedback structure can be used to obtain the perfect model matching. The stability proof is derived using the general Laypunov stability theory. At the end of paper, an example of adaptive control of third order nonlinear system is presented, where a considerable improvement of control system dynamical behavior has been obtained.