-
Gaussian Processes with Noisy Regression Inputs for Dynamical Systems
Authors:
Tobias M. Wolff,
Victor G. Lopez,
Matthias A. Müller
Abstract:
This paper is centered around the approximation of dynamical systems by means of Gaussian processes. To this end, trajectories of such systems must be collected to be used as training data. The measurements of these trajectories are typically noisy, which implies that both the regression inputs and outputs are corrupted by noise. However, most of the literature considers only noise in the regressi…
▽ More
This paper is centered around the approximation of dynamical systems by means of Gaussian processes. To this end, trajectories of such systems must be collected to be used as training data. The measurements of these trajectories are typically noisy, which implies that both the regression inputs and outputs are corrupted by noise. However, most of the literature considers only noise in the regression outputs. In this paper, we show how to account for the noise in the regression inputs in an extended Gaussian process framework to approximate scalar and multidimensional systems. We demonstrate the potential of our framework by comparing it to different state-of-the-art methods in several simulation examples.
△ Less
Submitted 1 April, 2025; v1 submitted 16 August, 2024;
originally announced August 2024.
-
Gaussian Process-Based Nonlinear Moving Horizon Estimation
Authors:
Tobias M. Wolff,
Victor G. Lopez,
Matthias A. Müller
Abstract:
In this paper, we propose a novel Gaussian process-based moving horizon estimation (MHE) framework for unknown nonlinear systems. In the proposed scheme, we take advantage of the properties of Gaussian processes. On the one hand, we approximate the system dynamics by the posterior means of the learned Gaussian processes (GPs). On the other hand, we exploit the posterior variances of the Gaussian p…
▽ More
In this paper, we propose a novel Gaussian process-based moving horizon estimation (MHE) framework for unknown nonlinear systems. In the proposed scheme, we take advantage of the properties of Gaussian processes. On the one hand, we approximate the system dynamics by the posterior means of the learned Gaussian processes (GPs). On the other hand, we exploit the posterior variances of the Gaussian processes to design the weighting matrices in the MHE cost function and account for the uncertainty in the learned system dynamics. The data collection and the tuning of the hyperparameters are done offline. We prove robust stability of the GP-based MHE scheme using a Lyapunov-based proof technique. Furthermore, as additional contribution, we analyze under which conditions incremental input/output-to-state stability (a nonlinear detectability notion) is preserved when approximating the system dynamics using, e.g., machine learning techniques. Finally, we illustrate the performance of the GP-based MHE scheme in a simulation case study and show how the chosen weighting matrices can lead to an improved performance compared to standard cost functions.
△ Less
Submitted 7 February, 2024;
originally announced February 2024.
-
Model predictive control for the prescription of antithyroid agents
Authors:
Maylin Menzel,
Tobias M. Wolff,
Johannes W. Dietrich,
Matthias A. Müller
Abstract:
Although hyperthyroidism is a common disease, the pharmaceutical therapy is based on a trial-and-error approach. We extend a mathematical model of the pituitary-thyroid feedback loop such that the intake of one antithyroid agent, namely methimazole (MMI), can be considered and use a model predictive control (MPC) scheme to determine suitable dosages.
Although hyperthyroidism is a common disease, the pharmaceutical therapy is based on a trial-and-error approach. We extend a mathematical model of the pituitary-thyroid feedback loop such that the intake of one antithyroid agent, namely methimazole (MMI), can be considered and use a model predictive control (MPC) scheme to determine suitable dosages.
△ Less
Submitted 31 July, 2023;
originally announced July 2023.
-
Robust Stability of Gaussian Process Based Moving Horizon Estimation
Authors:
Tobias M. Wolff,
Victor G. Lopez,
Matthias A. Müller
Abstract:
In this paper, we introduce a Gaussian process based moving horizon estimation (MHE) framework. The scheme is based on offline collected data and offline hyperparameter optimization. In particular, compared to standard MHE schemes, we replace the mathematical model of the system by the posterior mean of the Gaussian process. To account for the uncertainty of the learned model, we exploit the poste…
▽ More
In this paper, we introduce a Gaussian process based moving horizon estimation (MHE) framework. The scheme is based on offline collected data and offline hyperparameter optimization. In particular, compared to standard MHE schemes, we replace the mathematical model of the system by the posterior mean of the Gaussian process. To account for the uncertainty of the learned model, we exploit the posterior variance of the learned Gaussian process in the weighting matrices of the cost function of the proposed MHE scheme. We prove practical robust exponential stability of the resulting estimator using a recently proposed Lyapunov-based proof technique. Finally, the performance of the Gaussian process based MHE scheme is illustrated via a nonlinear system.
△ Less
Submitted 14 June, 2023; v1 submitted 13 April, 2023;
originally announced April 2023.
-
Modeling and Predictive Control for the Treatment of Hyperthyroidism
Authors:
Tobias M. Wolff,
Maylin Menzel,
Johannes W. Dietrich,
Matthias A. Müller
Abstract:
In this work, we propose an approach to determine the dosages of antithyroid agents to treat hyperthyroid patients. Instead of relying on a trial-and-error approach as it is commonly done in clinical practice, we suggest to determine the dosages by means of a model predictive control (MPC) scheme. To this end, we first extend a mathematical model of the pituitary-thyroid feedback loop such that th…
▽ More
In this work, we propose an approach to determine the dosages of antithyroid agents to treat hyperthyroid patients. Instead of relying on a trial-and-error approach as it is commonly done in clinical practice, we suggest to determine the dosages by means of a model predictive control (MPC) scheme. To this end, we first extend a mathematical model of the pituitary-thyroid feedback loop such that the intake of methimazole, a common antithyroid agent, can be considered. Second, based on the extended model, we develop an MPC scheme to determine suitable dosages. In numerical simulations, we consider scenarios in which (i) patients are affected by Graves' disease and take the medication orally and (ii) patients suffering from a life-threatening thyrotoxicosis, in which the medication is usually given intravenously. Our conceptual study suggests that determining the medication dosages by means of an MPC scheme could be a promising alternative to the currently applied trial-and-error approach.
△ Less
Submitted 16 January, 2025; v1 submitted 20 December, 2022;
originally announced December 2022.
-
Robust Data-Driven Moving Horizon Estimation for Linear Discrete-Time Systems
Authors:
Tobias M. Wolff,
Victor G. Lopez,
Matthias A. Müller
Abstract:
In this paper, a robust data-driven moving horizon estimation (MHE) scheme for linear time-invariant discrete-time systems is introduced. The scheme solely relies on offline collected data without employing any system identification step. We prove practical robust exponential stability for the setting where both the online measurements and the offline collected data are corrupted by non-vanishing…
▽ More
In this paper, a robust data-driven moving horizon estimation (MHE) scheme for linear time-invariant discrete-time systems is introduced. The scheme solely relies on offline collected data without employing any system identification step. We prove practical robust exponential stability for the setting where both the online measurements and the offline collected data are corrupted by non-vanishing and bounded noise. The behavior of the novel robust data-driven MHE scheme is illustrated by means of simulation examples and compared to a standard model-based MHE scheme, where the model is identified using the same offline data as for the data-driven MHE scheme.
△ Less
Submitted 28 February, 2024; v1 submitted 17 October, 2022;
originally announced October 2022.
-
Data-Based Moving Horizon Estimation for Linear Discrete-Time Systems
Authors:
Tobias M. Wolff,
Victor G. Lopez,
Matthias A. Müller
Abstract:
This paper introduces a data-based moving horizon estimation (MHE) scheme for linear time-invariant discrete-time systems. The scheme solely relies on collected data without employing any system identification step. Robust global exponential stability of the data-based MHE is proven under standard assumptions for the case where the online output measurements are corrupted by some non-vanishing mea…
▽ More
This paper introduces a data-based moving horizon estimation (MHE) scheme for linear time-invariant discrete-time systems. The scheme solely relies on collected data without employing any system identification step. Robust global exponential stability of the data-based MHE is proven under standard assumptions for the case where the online output measurements are corrupted by some non-vanishing measurement noise. A simulation example illustrates the behavior of the data-based MHE scheme.
△ Less
Submitted 28 April, 2022; v1 submitted 9 November, 2021;
originally announced November 2021.
