Tom Oomen

Structuring ambiguity sets in Wasserstein-based distributionally robust optimization (DRO) can improve their statistical properties when the uncertainty consists of multiple independent components. The aim of this paper is to solve stochastic optimization problems with unknown uncertainty when we only have access to a finite set of samples from it. Exploiting strong duality of DRO problems over structured ambiguity sets, we derive tractable reformulations for certain classes of DRO and uncertainty quantification problems. We also derive tractable reformulations for distributionally robust chance-constrained problems. As the complexity of the reformu lations may grow exponentially with the number of independent uncertainty components, we employ clustering strategies to ob tain informative estimators, which yield problems of manageable complexity. We demonstrate the effectiveness of the theoretical results in a numerical simulation example.

Controlled systems often involve disturbances that can be modeled as stochastic processes, with imprecise knowledge of their distributions. The aim of this paper is to develop a distributionally robust model predictive control algorithm for discrete-time linear systems with stochastic disturbances that have an unknown distribution. In the presented method, we infer, from a limited number of samples, structured ambiguity sets that capture the uncertain distribution with high confidence. Exploiting reformulations from distributionally robust optimization, these ambiguity sets yield convex receding horizon control problems that are computationally tractable and recursively feasible. To control the complexity of the problem, the parameterization of each ambiguity set is tailored to the horizon stage. This is achieved by clustering the points that are used to generate the reference distributions around which the ambiguity sets are built.

The standard assumption that exact measurement data is available at equidistant time instances in repetitive control (RC) is not always justified, e.g., when exploiting time-stamped data from incremental encoders or in systems with data dropouts. The aim of this article is to develop an RC framework that is capable of exploiting intermittently sampled measurement signals. Stability conditions for the intermittent RC framework are provided that can be verified using the frequency response function (FRF) data. This condition results in a frequency-domain design procedure for repetitive controllers in this intermittent setting. The intermittent RC framework is validated on an industrial print belt setup for which nonequidistant measurement data are available.

Objective: This study aims to reduce expert annotation effort in detecting patient-ventilator asynchrony (PVA) by introducing a semi-supervised learning framework for time series classification. Methods and procedures: We propose a model-independent framework that integrates hierarchical clustering and dynamic time-warping (DTW) for efficient data selection and label projection. The framework includes five steps: data collection, selection, annotation, projection, and model training. It is validated using a fully labeled dataset from Fondazione I.R.C.C.S. Policlinico San Matteo and applied to an unlabeled dataset from Maasstad Hospital, where annotation consistency and label quality are analyzed. Results: The framework reduces annotation effort by over 75% while closely resembling classification performance. On the San Matteo dataset, the model trained with projected labels achieved performance close to that of a fully supervised model. The method effectively captured rare PVA types and improved macro-averaged F1 scores compared to random sampling. On the Maasstad dataset, despite annotation inconsistencies, the framework demonstrated moderate detection performance (75% micro-averaged $F_{1}$ score) using labels from a single clinical expert. Conclusion: Our semi-supervised framework enables scalable and efficient annotation of clinical time series data, maintaining model accuracy with minimal expert input. It is robust across datasets and adaptable to varying signal quality and annotation consistency.

Subspace predictive control (SPC) is a data-driven control strategy that utilizes input–output measurements to predict future system behavior without requiring explicit model identification. Conventional SPC exhibits vulnerability to an unknown input disturbance, leading to degraded control performance and steady-state errors. To address these limitations, this article proposes a robust SPC method that inherently mitigates the effect of a constant input disturbance by augmenting the state-space representation through the internal model principle (IMP). This augmentation enables the controller to achieve integral action without requiring a separate disturbance observer (DOB) design. The proposed method is implemented in a data-driven framework, where an auxiliary disturbance is introduced into the data-driven algorithm to enhance disturbance rejection. A transfer function analysis verifies that the proposed Robust SPC eliminates a constant disturbance while maintaining the role of a DOB. Experimental validation on a two-inertia system confirms that the proposed method significantly improves reference tracking performance compared to conventional SPC, demonstrating its effectiveness in disturbance rejection without additional modeling complexity.

Free-space optical satellite communication terminals rely on accurate metrology of their pointing mirrors to correctly aim their laser to a counter terminal, while at the same time requiring simple, lightweight and low-cost sensors. The aim of this paper is to develop an automated procedure for the calibration of these sensors in a mass production setting, using a pointing test bench (PTB) to automatically calibrate the angular sensors of many Coarse Pointing Assemblies (CPAs), which position pointing mirrors over a large field-of-regard. The PTB and CPA are aligned using feedback over an external optical position sensor (OPS) and an inverse kinematic model is learned from data, after which Gaussian Process regression models are created to predict and correct sensor errors, taking into account propagation of calibration errors from the PTB to the CPA. Experimental results show that the CPA sensor errors are reduced by two orders of magnitude by this automated calibration approach, even at orientations at which the PTB itself is uncalibrated. The developed framework is generalizable to calibration of arbitrary 2 $^{\circ }$ of freedom (2-DOF) rotary systems and is not limited to specific types of position sensors, thereby enabling significant cost savings and increased accuracy in mass production of satellite communication terminals.

Mechanical ventilators are essential for patients who are unable to breathe independently. The aim of this article is to develop a systematic control design methodology that achieves accurate tracking of both the pressure and flow to ensure comfortable breathing for the patient. A hybrid controller is introduced that ensures improved baseflow tracking performance. The actual controller design leverages frequency-based techniques and is based on static decoupling and the factorized Nyquist criterion. Furthermore, a theoretical stability analysis of the hybrid controller is presented. The presented control strategy is implemented in a real ventilator, and it is demonstrated that the tracking performance is improved by conducting an experimental case study.