Volume
OASIcs, Volume 148
37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026)
-
Part of:
Series:
Open Access Series in Informatics (OASIcs)
Conference: International Conference on Principles of Diagnosis and Resilient Systems (DX)
Event
Editors
Publication Details
- published at: 2026-09-28
- Publisher: Schloss Dagstuhl – Leibniz-Zentrum für Informatik
- ISBN: 978-3-95977-455-0
Access Numbers
- Detailed Access Statistics available here
-
Total Accesses (updated on a weekly basis)
0Documents
Documents
No documents found matching your filter selection.
Document
Complete Volume
OASIcs, Volume 148, DX 2026, Complete Volume
Abstract
OASIcs, Volume 148, DX 2026, Complete Volume
Cite as
37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026). Open Access Series in Informatics (OASIcs), Volume 148, pp. 1-342, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2026)
Copy BibTex To Clipboard
@Proceedings{pill_et_al:OASIcs.DX.2026,
title = {{OASIcs, Volume 148, DX 2026, Complete Volume}},
booktitle = {37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026)},
pages = {1--342},
series = {Open Access Series in Informatics (OASIcs)},
ISBN = {978-3-95977-455-0},
ISSN = {2190-6807},
year = {2026},
volume = {148},
editor = {Pill, Ingo and Zanella, Marina and Provan, Gregory},
publisher = {Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
address = {Dagstuhl, Germany},
URL = {https://drops.dagstuhl.de/entities/document/10.4230/OASIcs.DX.2026},
URN = {urn:nbn:de:0030-drops-280802},
doi = {10.4230/OASIcs.DX.2026},
annote = {Keywords: OASIcs, Volume 148, DX 2026, Complete Volume}
}
Document
Front Matter
Front Matter, Table of Contents, Preface, Conference Organization
Abstract
Front Matter, Table of Contents, Preface, Conference Organization
Cite as
37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026). Open Access Series in Informatics (OASIcs), Volume 148, pp. 0:i-0:xii, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2026)
Copy BibTex To Clipboard
@InProceedings{pill_et_al:OASIcs.DX.2026.0,
author = {Pill, Ingo and Zanella, Marina and Provan, Gregory},
title = {{Front Matter, Table of Contents, Preface, Conference Organization}},
booktitle = {37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026)},
pages = {0:i--0:xii},
series = {Open Access Series in Informatics (OASIcs)},
ISBN = {978-3-95977-455-0},
ISSN = {2190-6807},
year = {2026},
volume = {148},
editor = {Pill, Ingo and Zanella, Marina and Provan, Gregory},
publisher = {Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
address = {Dagstuhl, Germany},
URL = {https://drops.dagstuhl.de/entities/document/10.4230/OASIcs.DX.2026.0},
URN = {urn:nbn:de:0030-drops-280799},
doi = {10.4230/OASIcs.DX.2026.0},
annote = {Keywords: Front Matter, Table of Contents, Preface, Conference Organization}
}
Document
A Realization-Theoretic Foundation for Fault Diagnosis, with Diagnosability Equivalence Predictions
Abstract
Fault diagnosis encompasses a wide range of paradigms, including observer-based fault detection and isolation, consistency-based diagnosis, Bayesian diagnosis, geometric diagnosis, and structured machine-learning approaches. Despite their shared objective, these paradigms are formulated in different mathematical languages and are rarely analyzed within a common framework. This paper introduces a realization-theoretic foundation for diagnosis based on the Diagnostic System Specification (DSS), a representation that separates behavioural models, realizations, measurement fields, and inference procedures.
We define diagnostic invariants as symmetry-preserving quantities of realized system structures and show that detectability is fundamentally an orbit separation property in the realization’s symmetry space. Building on this observation, we develop a theory of realizations in which diagnostic models appear as objects equipped with invariant families and nominal symmetry groups. We show that diagnostic paradigms differ not in their underlying architecture but in the realizations they induce and the invariants they expose.
Our main result addresses the question: is there a single mathematical condition of which diagnosis approaches are equivalent in terms of diagnosability? We show that using orbit separation on a paradigm-appropriate realization shows the following: faults are distinguishable iff they lie in different orbits of the realization’s nominal symmetry group. We accomplish this by constructing a minimal sufficient realization associated with a measurement field and characterize it by a universal property. This yields a structural notion of diagnostic equivalence and provides a principled basis for comparing methods across traditionally separate paradigms.
Cite as
Gregory Provan. A Realization-Theoretic Foundation for Fault Diagnosis, with Diagnosability Equivalence Predictions. In 37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026). Open Access Series in Informatics (OASIcs), Volume 148, pp. 1:1-1:20, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2026)
Copy BibTex To Clipboard
@InProceedings{provan:OASIcs.DX.2026.1,
author = {Provan, Gregory},
title = {{A Realization-Theoretic Foundation for Fault Diagnosis, with Diagnosability Equivalence Predictions}},
booktitle = {37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026)},
pages = {1:1--1:20},
series = {Open Access Series in Informatics (OASIcs)},
ISBN = {978-3-95977-455-0},
ISSN = {2190-6807},
year = {2026},
volume = {148},
editor = {Pill, Ingo and Zanella, Marina and Provan, Gregory},
publisher = {Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
address = {Dagstuhl, Germany},
URL = {https://drops.dagstuhl.de/entities/document/10.4230/OASIcs.DX.2026.1},
URN = {urn:nbn:de:0030-drops-278156},
doi = {10.4230/OASIcs.DX.2026.1},
annote = {Keywords: Model-based Diagnosis, Diagnosability, System Realization Theory}
}
Document
An Explainable GNN Framework for Component-Level Anomaly Diagnosis
Abstract
Industrial processes are complex systems composed of multiple interacting sensors that generate multivariate time series (MTS). Detecting anomalies in such systems is critical for reliability and safety, yet understanding their origin is equally important. Existing Graph Neural Network (GNN)-based methods for anomaly detection primarily focus on sensor-level deviations and either attribute anomalies directly to the deviating sensors. When diagnosis is attempted, generally, the most deviated sensor is identified as a root cause of a system fault. However, in many industrial systems, anomalies do not arise from faulty sensors but from disruptions in the influences governing the system dynamics. We propose an explainable GNN-based anomaly detection framework that shifts the perspective from sensor-level anomalies to component-level diagnosis, hypothesizing that anomalous measurements are symptoms of altered inter-sensor influences. Experiments show that the method effectively identifies and prioritizes the true faulty components, providing interpretable insights into system failures.
Cite as
Sena Ozgunay, Louise Travé-Massuyès, Jean-Michel Loubes, and Raul Sena Ferreira. An Explainable GNN Framework for Component-Level Anomaly Diagnosis. In 37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026). Open Access Series in Informatics (OASIcs), Volume 148, pp. 2:1-2:17, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2026)
Copy BibTex To Clipboard
@InProceedings{ozgunay_et_al:OASIcs.DX.2026.2,
author = {Ozgunay, Sena and Trav\'{e}-Massuy\`{e}s, Louise and Loubes, Jean-Michel and Ferreira, Raul Sena},
title = {{An Explainable GNN Framework for Component-Level Anomaly Diagnosis}},
booktitle = {37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026)},
pages = {2:1--2:17},
series = {Open Access Series in Informatics (OASIcs)},
ISBN = {978-3-95977-455-0},
ISSN = {2190-6807},
year = {2026},
volume = {148},
editor = {Pill, Ingo and Zanella, Marina and Provan, Gregory},
publisher = {Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
address = {Dagstuhl, Germany},
URL = {https://drops.dagstuhl.de/entities/document/10.4230/OASIcs.DX.2026.2},
URN = {urn:nbn:de:0030-drops-278163},
doi = {10.4230/OASIcs.DX.2026.2},
annote = {Keywords: Anomaly Detection, Fault Diagnosis, Graph Neural Networks, Time Series}
}
Document
Dissecting NetDoktor’s Online Questionnaire with Entropy, Tree Simplification & Logical Abduction
Abstract
Medical diagnosis and its accuracy are dependent on the available information. Hence, it is in the interest of both the patient and the doctor to provide the necessary information and ask the right questions in an efficient manner. The same can be said about medical questionnaires, which resemble the conversation during a doctor’s appointment. NetDoktor’s SymptomChecker is a freely accessible, expert-curated questionnaire in the German language. It can be traversed like a large decision tree, linking symptoms to diseases. However, does it find an optimal conversation path to acquire the necessary information from its users? Here, our study surfaces structural deficiencies and ways to optimise the questionnaire paths. In particular, we model the process of acquiring information and refining prognoses as an abductive reasoning problem and provide a corresponding algorithm. In our experiments, we apply simple tree transformations and optimisations, building upon Shannon’s information entropy to expose redundant subtrees and inefficient question order. Our results contribute to a better understanding of medical questionnaires and open the door for future applications of the data provided by NetDoktor. Because SymptomChecker is validated by medical professionals, it is a prime candidate to serve as a reference for benchmarking other systems. For instance, a dataset derived from SymptomChecker can be used for testing systems such as large language models in the task of medical diagnosis.
Cite as
Alexander Perko, Emir Mujić, and Franz Wotawa. Dissecting NetDoktor’s Online Questionnaire with Entropy, Tree Simplification & Logical Abduction. In 37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026). Open Access Series in Informatics (OASIcs), Volume 148, pp. 3:1-3:20, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2026)
Copy BibTex To Clipboard
@InProceedings{perko_et_al:OASIcs.DX.2026.3,
author = {Perko, Alexander and Muji\'{c}, Emir and Wotawa, Franz},
title = {{Dissecting NetDoktor’s Online Questionnaire with Entropy, Tree Simplification \& Logical Abduction}},
booktitle = {37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026)},
pages = {3:1--3:20},
series = {Open Access Series in Informatics (OASIcs)},
ISBN = {978-3-95977-455-0},
ISSN = {2190-6807},
year = {2026},
volume = {148},
editor = {Pill, Ingo and Zanella, Marina and Provan, Gregory},
publisher = {Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
address = {Dagstuhl, Germany},
URL = {https://drops.dagstuhl.de/entities/document/10.4230/OASIcs.DX.2026.3},
URN = {urn:nbn:de:0030-drops-278173},
doi = {10.4230/OASIcs.DX.2026.3},
annote = {Keywords: Medical Questionnaire, Diagnosis, Abductive Reasoning, Information Entropy}
}
Document
Early Failure Prediction from Near-Anomaly Detection: A Proactive Approach
Abstract
Anomaly detection methods often have uncertain behavior with respect to samples near the distribution boundary, limiting their ability to anticipate future anomalies. This work introduces the concept of near-anomalies that, while not yet anomalous, lie close to the boundary and are likely to transition into anomalies in the near future. To address this, we propose an unsupervised method, named Christoffel-based ANomaly Anticipation for eaRly dIscovery (CANARI), which leverages the strong theoretical foundations of the Christoffel function to detect near-anomalies. The method is validated on industrial in-circuit testing data from printed circuit boards, with synthetically generated near-anomaly samples due to the lack of real-world data labeling. Experimental results show that CANARI outperforms the compared baselines that generally use a dual-threshold mechanism (one for anomalies and one for near-anomalies). It therefore provides a proactive solution for anticipating anomalies before they occur, offering a promising approach for resilience, predictive maintenance, and quality control.
Cite as
Léa Billet, Louise Travé-Massuyès, Elodie Chanthery, and Alexandre Gaffet. Early Failure Prediction from Near-Anomaly Detection: A Proactive Approach. In 37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026). Open Access Series in Informatics (OASIcs), Volume 148, pp. 4:1-4:18, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2026)
Copy BibTex To Clipboard
@InProceedings{billet_et_al:OASIcs.DX.2026.4,
author = {Billet, L\'{e}a and Trav\'{e}-Massuy\`{e}s, Louise and Chanthery, Elodie and Gaffet, Alexandre},
title = {{Early Failure Prediction from Near-Anomaly Detection: A Proactive Approach}},
booktitle = {37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026)},
pages = {4:1--4:18},
series = {Open Access Series in Informatics (OASIcs)},
ISBN = {978-3-95977-455-0},
ISSN = {2190-6807},
year = {2026},
volume = {148},
editor = {Pill, Ingo and Zanella, Marina and Provan, Gregory},
publisher = {Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
address = {Dagstuhl, Germany},
URL = {https://drops.dagstuhl.de/entities/document/10.4230/OASIcs.DX.2026.4},
URN = {urn:nbn:de:0030-drops-278189},
doi = {10.4230/OASIcs.DX.2026.4},
annote = {Keywords: Unsupervised anomaly detection, Christoffel function, Near-anomalies}
}
Document
FELIX: Model Repair in Numeric Planning
Abstract
Automated planning algorithms rely on having a sufficiently accurate domain model to solve planning problems, particularly in complex environments with numeric effects. In real-world settings, however, domain models can become misaligned with the environment over time, i.e., no longer accurately reflect the environment. Even a minor drift in numeric effects can lead to faulty plans and execution failures. In this work, we address the problem of repairing numeric planning domain models by proposing FELIX, a learning-based repair mechanism that restores planning effectiveness by identifying and correcting outdated numeric effects. Rather than re-learning the full environment dynamics from scratch, our approach updates only the affected components of the domain model, enabling efficient adaptation with limited data. FELIX restores planning performance across diverse numeric domains, even under complex model drift. It uniquely handles both nonlinear effect shifts and structural signature changes while maintaining polynomial repair overhead under bounded parameter-type assumptions.
Cite as
Nir Aharoni, Yarin Benyamin, Shiwali Mohan, and Roni Stern. FELIX: Model Repair in Numeric Planning. In 37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026). Open Access Series in Informatics (OASIcs), Volume 148, pp. 5:1-5:18, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2026)
Copy BibTex To Clipboard
@InProceedings{aharoni_et_al:OASIcs.DX.2026.5,
author = {Aharoni, Nir and Benyamin, Yarin and Mohan, Shiwali and Stern, Roni},
title = {{FELIX: Model Repair in Numeric Planning}},
booktitle = {37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026)},
pages = {5:1--5:18},
series = {Open Access Series in Informatics (OASIcs)},
ISBN = {978-3-95977-455-0},
ISSN = {2190-6807},
year = {2026},
volume = {148},
editor = {Pill, Ingo and Zanella, Marina and Provan, Gregory},
publisher = {Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
address = {Dagstuhl, Germany},
URL = {https://drops.dagstuhl.de/entities/document/10.4230/OASIcs.DX.2026.5},
URN = {urn:nbn:de:0030-drops-278197},
doi = {10.4230/OASIcs.DX.2026.5},
annote = {Keywords: Automated diagnosis, learning action models, model repair}
}
Document
Geometric Feature Selection for Interpretable SVM Classifiers in Cyber-Physical Systems
Abstract
Identifying mode-transition conditions in Cyber-Physical Systems (CPS) is fundamental to model-based diagnosis, but learning them from scarce, multivariate time-series data requires boundary expressions that remain interpretable in physical variables. A central challenge is that some features are actively detrimental: their inclusion distorts the orientation of the learned boundary, degrading robustness to future crossings without affecting training accuracy. We introduce SVM-GRFE (Geometric Recursive Feature Elimination for SVM), a feature selection method that operates on polynomial monomials rather than raw variables, jointly assessing accuracy and geometric criteria that capture each monomial’s structural role in boundary orientation; the boundary is back-projected analytically to an explicit polynomial inequality in physical variables. We prove SVM-GRFE returns an interpretable, geometrically stable, and robust boundary. On seven synthetic scenarios, SVM-GRFE achieves 33%-85% monomial reduction, matches or outperforms SVM-RFE in five out of seven scenarios and consistently outperforms SVM-RFE under noise at equal complexity, most notably under heteroscedastic noise.
Cite as
Leonie Hatte, Pauline Ribot, and Elodie Chanthery. Geometric Feature Selection for Interpretable SVM Classifiers in Cyber-Physical Systems. In 37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026). Open Access Series in Informatics (OASIcs), Volume 148, pp. 6:1-6:20, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2026)
Copy BibTex To Clipboard
@InProceedings{hatte_et_al:OASIcs.DX.2026.6,
author = {Hatte, Leonie and Ribot, Pauline and Chanthery, Elodie},
title = {{Geometric Feature Selection for Interpretable SVM Classifiers in Cyber-Physical Systems}},
booktitle = {37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026)},
pages = {6:1--6:20},
series = {Open Access Series in Informatics (OASIcs)},
ISBN = {978-3-95977-455-0},
ISSN = {2190-6807},
year = {2026},
volume = {148},
editor = {Pill, Ingo and Zanella, Marina and Provan, Gregory},
publisher = {Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
address = {Dagstuhl, Germany},
URL = {https://drops.dagstuhl.de/entities/document/10.4230/OASIcs.DX.2026.6},
URN = {urn:nbn:de:0030-drops-278200},
doi = {10.4230/OASIcs.DX.2026.6},
annote = {Keywords: Feature Selection, SVM, Geometric boundary learning, CPS}
}
Document
HSU TwinFlow: A Living Benchmark for Evaluating Anomaly Detection and Diagnosis Methods in Cyber-Physical Production Systems Using Digital-Twin-Generated Data
Abstract
We present a living (evolving) benchmark for evaluating data-driven methods for fault-related tasks, including fault (anomaly) detection and diagnosis, based on datasets generated by a realistic digital twin of a cyber-physical production system (CPPS). The digital twin is designed to capture the structural layout, operational behavior, and material flow of a modern production facility in a realistic manner. Scenario definitions, represented as sequences of high-level events, are used to generate labeled model-training and evaluation data from carefully designed fault scenarios. In addition, the datasets include contextual data, such as scenario metadata, product configurations, material-flow information, together with system knowledge such as the component hierarchy, component-type information, and material-flow graph. The proposed benchmark therefore provides a practical basis for method development and qualitative evaluation beyond simplified or isolated test setups. We provide initial benchmark results by evaluating two representative methods for each of the two considered tasks: anomaly detection and fault diagnosis.
Cite as
Nemanja Hranisavljevic, Alexander Diedrich, Lukas Moddemann, Domenic Schaeffer, Frank Marek, Ingo Pill, and Oliver Niggemann. HSU TwinFlow: A Living Benchmark for Evaluating Anomaly Detection and Diagnosis Methods in Cyber-Physical Production Systems Using Digital-Twin-Generated Data. In 37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026). Open Access Series in Informatics (OASIcs), Volume 148, pp. 7:1-7:14, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2026)
Copy BibTex To Clipboard
@InProceedings{hranisavljevic_et_al:OASIcs.DX.2026.7,
author = {Hranisavljevic, Nemanja and Diedrich, Alexander and Moddemann, Lukas and Schaeffer, Domenic and Marek, Frank and Pill, Ingo and Niggemann, Oliver},
title = {{HSU TwinFlow: A Living Benchmark for Evaluating Anomaly Detection and Diagnosis Methods in Cyber-Physical Production Systems Using Digital-Twin-Generated Data}},
booktitle = {37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026)},
pages = {7:1--7:14},
series = {Open Access Series in Informatics (OASIcs)},
ISBN = {978-3-95977-455-0},
ISSN = {2190-6807},
year = {2026},
volume = {148},
editor = {Pill, Ingo and Zanella, Marina and Provan, Gregory},
publisher = {Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
address = {Dagstuhl, Germany},
URL = {https://drops.dagstuhl.de/entities/document/10.4230/OASIcs.DX.2026.7},
URN = {urn:nbn:de:0030-drops-278212},
doi = {10.4230/OASIcs.DX.2026.7},
annote = {Keywords: Benchmark, Digital Twin, Manufacturing, Production Systems, Cyber-Physical Systems, Artificial Intelligence, Anomaly Detection, Diagnosis, System Reconfiguration}
}
Document
Hybrid Fault Detection and Isolation for the Fuel Turbopump Subsystem of the LOX/LNG Expander-Bleed Rocket Engine LUMEN
Abstract
The transition toward reusable and cost-efficient launch systems increases the need for advanced fault detection and isolation (FDI) capabilities to support the reliable and safe operation of liquid rocket engines. Development of these systems typically relies on physics-based simulation models and experimental data. LUMEN (Liquid Upper stage deMonstrator ENgine) is a modular, pump-fed LOX/LNG rocket engine with a nominal thrust of 25 kN, developed and operated by the Institute of Space Propulsion of the German Aerospace Center (DLR). The Rocket Engine Control and Diagnosis Benchmark provides a high-fidelity, experimentally validated EcosimPro/ESPSS simulation model of LUMEN for FDI system development. In this work, we use the fuel turbopump subsystem of this model as a case study and integrate data-driven components into a consistency-based diagnosis (CBD) framework to improve diagnostic performance. The results demonstrate that the hybrid architecture improves isolation accuracy, increasing it from 0.46 to 0.76 compared to the CBD baseline. In addition, we propose a sensor validation network (SVN) that uses a grey-box neural ordinary differential equation surrogate driven by valve-command inputs to decouple sensor faults from component faults. This further increases the isolation accuracy to 0.92, while retaining the ability to detect unknown component faults. These results indicate that incorporating data-driven components helps to compensate for structural non-isolability and improve diagnostic performance while preserving physical interpretability.
Cite as
Eldin Kurudzija, Marcos Quinones-Grueiro, Kai Dresia, Austin Coursey, Gautam Biswas, Jan Deeken, and Günther Waxenegger-Wilfing. Hybrid Fault Detection and Isolation for the Fuel Turbopump Subsystem of the LOX/LNG Expander-Bleed Rocket Engine LUMEN. In 37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026). Open Access Series in Informatics (OASIcs), Volume 148, pp. 8:1-8:20, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2026)
Copy BibTex To Clipboard
@InProceedings{kurudzija_et_al:OASIcs.DX.2026.8,
author = {Kurudzija, Eldin and Quinones-Grueiro, Marcos and Dresia, Kai and Coursey, Austin and Biswas, Gautam and Deeken, Jan and Waxenegger-Wilfing, G\"{u}nther},
title = {{Hybrid Fault Detection and Isolation for the Fuel Turbopump Subsystem of the LOX/LNG Expander-Bleed Rocket Engine LUMEN}},
booktitle = {37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026)},
pages = {8:1--8:20},
series = {Open Access Series in Informatics (OASIcs)},
ISBN = {978-3-95977-455-0},
ISSN = {2190-6807},
year = {2026},
volume = {148},
editor = {Pill, Ingo and Zanella, Marina and Provan, Gregory},
publisher = {Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
address = {Dagstuhl, Germany},
URL = {https://drops.dagstuhl.de/entities/document/10.4230/OASIcs.DX.2026.8},
URN = {urn:nbn:de:0030-drops-278223},
doi = {10.4230/OASIcs.DX.2026.8},
annote = {Keywords: Hybrid, Fault Detection and Isolation, Liquid Rocket Engines, LUMEN}
}
Document
On the Practical Utility of Diagnoses
Abstract
Diagnoses address a universal need for explanations that arises whenever things go wrong or when we observe some unexpected behavior. The ubiquitous need for such explanations has led to an abundance of algorithmic concepts and computation variants for diagnostic processes. Comparisons have mainly focused on highlighting the virtues of individual algorithms or the required computational efforts. In this paper, we focus on evaluating the results of a diagnosis algorithm in the context of an inspection and repair process. We completely ignore the algorithm’s computation concept and whether the observations would justifiably support multiple diagnoses. In particular, we focus on inspecting and discussing the practical utility of diagnoses as a measure of the unnecessary efforts one has to spend when fully repairing a system. We start with assessing the utility of a single diagnosis and then progress to evaluating ambiguity groups, i.e., sets of diagnoses. We propose corresponding metrics that investigate worst case and average performance, contrast them to the utility metric used in the DX Competition 2010, and discuss the metrics' results for several scenarios.
Cite as
Ingo Pill and Johan de Kleer. On the Practical Utility of Diagnoses. In 37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026). Open Access Series in Informatics (OASIcs), Volume 148, pp. 9:1-9:20, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2026)
Copy BibTex To Clipboard
@InProceedings{pill_et_al:OASIcs.DX.2026.9,
author = {Pill, Ingo and de Kleer, Johan},
title = {{On the Practical Utility of Diagnoses}},
booktitle = {37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026)},
pages = {9:1--9:20},
series = {Open Access Series in Informatics (OASIcs)},
ISBN = {978-3-95977-455-0},
ISSN = {2190-6807},
year = {2026},
volume = {148},
editor = {Pill, Ingo and Zanella, Marina and Provan, Gregory},
publisher = {Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
address = {Dagstuhl, Germany},
URL = {https://drops.dagstuhl.de/entities/document/10.4230/OASIcs.DX.2026.9},
URN = {urn:nbn:de:0030-drops-278233},
doi = {10.4230/OASIcs.DX.2026.9},
annote = {Keywords: Model-based Diagnosis, Diagnosis, Algorithms}
}
Document
Planning Domain Repair Using Preferred Plans and Background Knowledge
Abstract
Planning domain repair assists developers with the tedious task of finding and fixing errors in planning domains. We present Domain Repair for PDDL (DrPDDL), a novel automated debugger supporting lifted STRIPS domains with typing. As input, it takes a faulty domain, problem instances with corresponding preferred plans - plans the user desires but which may be invalid in the current domain - and optional State Integrity Axioms (SIAs) that must hold across all reachable states. Unlike prior tools, DrPDDL can also repair domains when the preferred plans are already valid. The debugger uses a pipeline of consecutive algorithms. It first relaxes the domain to fix failing action preconditions and SIA violations, and subsequently constricts it as much as possible while ensuring preferred plan validity. DrPDDL can add or remove positive and negative effects, as well as positive preconditions. We evaluated our debugger by automatically inserting random errors into International Planning Competition domains, generating thousands of test instances. The results demonstrate that DrPDDL successfully detects and repairs a high percentage of these errors.
Cite as
Thomas Eckstein and Gerald Steinbauer-Wagner. Planning Domain Repair Using Preferred Plans and Background Knowledge. In 37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026). Open Access Series in Informatics (OASIcs), Volume 148, pp. 10:1-10:22, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2026)
Copy BibTex To Clipboard
@InProceedings{eckstein_et_al:OASIcs.DX.2026.10,
author = {Eckstein, Thomas and Steinbauer-Wagner, Gerald},
title = {{Planning Domain Repair Using Preferred Plans and Background Knowledge}},
booktitle = {37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026)},
pages = {10:1--10:22},
series = {Open Access Series in Informatics (OASIcs)},
ISBN = {978-3-95977-455-0},
ISSN = {2190-6807},
year = {2026},
volume = {148},
editor = {Pill, Ingo and Zanella, Marina and Provan, Gregory},
publisher = {Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
address = {Dagstuhl, Germany},
URL = {https://drops.dagstuhl.de/entities/document/10.4230/OASIcs.DX.2026.10},
URN = {urn:nbn:de:0030-drops-278246},
doi = {10.4230/OASIcs.DX.2026.10},
annote = {Keywords: Planning, Planning Domain Definition Language (PDDL), Debugging, Preferred Plans, Background Knowledge}
}
Document
QSIM-Guided Weak Supervision for Fault Detection in Cyber-Physical Systems
Abstract
This paper presents an initial investigation into the combination of qualitative simulation and machine learning for fault detection in cyber-physical systems. In recent years, the detection and diagnosis of faults in cyber-physical systems has become increasingly dependent on machine learning. However, supervised machine learning approaches rely on large labelled datasets, which are rarely available during the initial deployment phase. In systems such as smart buildings, faults occur rarely, and even the nominal behaviour of a system can vary across structurally identical installations. We therefore propose a framework that addresses this issue by generating weak supervision labels using qualitative reasoning. Given a qualitative model that captures the system’s nominal behaviour, we use a conformance check to determine whether the system traces are likely to be faulty or not. These binary pseudo-labels are then used to train a downstream classifier. Unlike statistical anomaly detection or self-supervised techniques, the generated supervision signal is based on physical consistency rather than deviations from a learned baseline. We evaluate our approach on four benchmarks relating to electrical, fluid, mechanical and thermal dynamics, comparing our methodology with unsupervised anomaly detection and supervised learning with limited labels. Our initial experiments suggest that, while qualitative conformance checks can provide usable training signals when labelled data is unavailable, their effectiveness depends on the discriminative structure of the qualitative model.
Cite as
Ankita Das, Roxane Koitz-Hristov, and Franz Wotawa. QSIM-Guided Weak Supervision for Fault Detection in Cyber-Physical Systems. In 37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026). Open Access Series in Informatics (OASIcs), Volume 148, pp. 11:1-11:20, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2026)
Copy BibTex To Clipboard
@InProceedings{das_et_al:OASIcs.DX.2026.11,
author = {Das, Ankita and Koitz-Hristov, Roxane and Wotawa, Franz},
title = {{QSIM-Guided Weak Supervision for Fault Detection in Cyber-Physical Systems}},
booktitle = {37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026)},
pages = {11:1--11:20},
series = {Open Access Series in Informatics (OASIcs)},
ISBN = {978-3-95977-455-0},
ISSN = {2190-6807},
year = {2026},
volume = {148},
editor = {Pill, Ingo and Zanella, Marina and Provan, Gregory},
publisher = {Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
address = {Dagstuhl, Germany},
URL = {https://drops.dagstuhl.de/entities/document/10.4230/OASIcs.DX.2026.11},
URN = {urn:nbn:de:0030-drops-278254},
doi = {10.4230/OASIcs.DX.2026.11},
annote = {Keywords: Qualitative Simulation, Fault Detection, Outlier Detection}
}
Document
Reinforcement Learning-Driven Predictive Maintenance Planning via Timed Automata Modeling
Abstract
Predictive maintenance of multi-component systems requires decisions that account for both local component degradation and system-level redundancy. This paper proposes a formal framework for synthesizing interpretable maintenance policies from a system functional architecture. The system is modeled as a network of Dynamic Priced Timed Game Automata (DPTGA): each component automaton encodes degradation dynamics, stochastic failure windows, and repair and replacement costs, while a System Health Monitor automaton is automatically derived from the minimal cut sets of the functional architecture and detects system failure as soon as any cut set becomes unavailable. We execute the DPTGA semantics in an explicit-state simulator and optimize a compact threshold policy with the Cross-Entropy Method (CEM). The policy contains one age-fraction threshold per component and an embedded subsystem-safe predicate that forbids any preventive action that would immediately realize a cut set, making the learned strategy provably safe and directly interpretable as a rule-based maintenance plan. On a seven-component benchmark comprising a parallel subsystem and a 2-out-of-5 redundant block, the CEM policy reduces mean cumulative cost by 15% and median cost by 16% relative to the best rule-based heuristic (corrective, systematic, and condition-based maintenance), while maintaining a 98.7% system survival rate. The approach therefore provides a compact and explainable alternative to deep reinforcement learning for architecture-aware predictive maintenance.
Cite as
Ferhat Tamssaouet, Pauline Ribot, and Yannick Pencolé. Reinforcement Learning-Driven Predictive Maintenance Planning via Timed Automata Modeling. In 37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026). Open Access Series in Informatics (OASIcs), Volume 148, pp. 12:1-12:20, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2026)
Copy BibTex To Clipboard
@InProceedings{tamssaouet_et_al:OASIcs.DX.2026.12,
author = {Tamssaouet, Ferhat and Ribot, Pauline and Pencol\'{e}, Yannick},
title = {{Reinforcement Learning-Driven Predictive Maintenance Planning via Timed Automata Modeling}},
booktitle = {37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026)},
pages = {12:1--12:20},
series = {Open Access Series in Informatics (OASIcs)},
ISBN = {978-3-95977-455-0},
ISSN = {2190-6807},
year = {2026},
volume = {148},
editor = {Pill, Ingo and Zanella, Marina and Provan, Gregory},
publisher = {Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
address = {Dagstuhl, Germany},
URL = {https://drops.dagstuhl.de/entities/document/10.4230/OASIcs.DX.2026.12},
URN = {urn:nbn:de:0030-drops-278260},
doi = {10.4230/OASIcs.DX.2026.12},
annote = {Keywords: System Remaining Useful Life, Predictive Maintenance, Timed Automata, Reinforcement Learning}
}
Document
Statistical Foundations of Local Geometric Fault Detection: The Curvature Channel
Abstract
Geometric and topological methods have recently emerged as a promising approach to fault detection in hybrid and cyber-physical systems. Unlike classical residual-based methods, which detect violations of a parametric model, geometric methods monitor changes in the local structure of trajectories through quantities such as intrinsic dimension, smoothness, tangent-cone geometry, and consistency with physical constraints. Despite encouraging empirical results, there is currently no theory characterising when such geometric signatures are detectable, distinguishable, or reliable under noise. This paper develops a statistical theory of local geometric fault detection, with a focus on curvature-based approaches. We study faults detectable by displacement in a curvature signature space, and derive the noise floor of geometric estimators based on finite differences and local neighbourhood averaging. The resulting estimator variance scales as σ_{est} = O(σ/(Δ t²√k)), exposing the fundamental trade-off between sampling rate, noise level, and neighbourhood size. Using this noise model, we derive geometric analogues of the classical detectability and isolability conditions of fault-detection theory, expressed directly in terms of separation in signature space. We further prove an exclusion result showing that a broad class of smooth parametric faults cannot activate curvature-based geometric channels, yielding a form of label-free fault-class identification. These results lead to a practical pre-deployment assessment procedure that determines, from nominal data alone, whether a curvature-based fault detector is expected to achieve a specified false-alarm rate, detection probability, and isolation capability. The theory thereby places local geometric fault detection on the same statistical footing as classical residual-based diagnosis while clarifying its distinctive strengths, limitations, and operating regime.
Cite as
Gregory Provan. Statistical Foundations of Local Geometric Fault Detection: The Curvature Channel. In 37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026). Open Access Series in Informatics (OASIcs), Volume 148, pp. 13:1-13:20, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2026)
Copy BibTex To Clipboard
@InProceedings{provan:OASIcs.DX.2026.13,
author = {Provan, Gregory},
title = {{Statistical Foundations of Local Geometric Fault Detection: The Curvature Channel}},
booktitle = {37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026)},
pages = {13:1--13:20},
series = {Open Access Series in Informatics (OASIcs)},
ISBN = {978-3-95977-455-0},
ISSN = {2190-6807},
year = {2026},
volume = {148},
editor = {Pill, Ingo and Zanella, Marina and Provan, Gregory},
publisher = {Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
address = {Dagstuhl, Germany},
URL = {https://drops.dagstuhl.de/entities/document/10.4230/OASIcs.DX.2026.13},
URN = {urn:nbn:de:0030-drops-278279},
doi = {10.4230/OASIcs.DX.2026.13},
annote = {Keywords: Geometric metrics, Diagnosis methods, Theory}
}
Document
Using SLMs and LLMs for Diagnosis
Abstract
Large Language Models (LLMs) have rapidly transformed how organizations approach complex problem solving, raising the question of whether they can also be applied to technical reasoning tasks. In this paper, we investigate whether LLMs can be used for system diagnosis, i.e., the localization of faults in systems comprising interacting components. In particular, we conducted an initial study using examples ranging from electric circuits to heating systems, evaluating 13 different language models of varying sizes with respect to the correctness and completeness of their diagnosis. The experimental evaluation revealed that language models with more parameters achieve superior performance.
Cite as
Sophia Plank and Franz Wotawa. Using SLMs and LLMs for Diagnosis. In 37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026). Open Access Series in Informatics (OASIcs), Volume 148, pp. 14:1-14:17, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2026)
Copy BibTex To Clipboard
@InProceedings{plank_et_al:OASIcs.DX.2026.14,
author = {Plank, Sophia and Wotawa, Franz},
title = {{Using SLMs and LLMs for Diagnosis}},
booktitle = {37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026)},
pages = {14:1--14:17},
series = {Open Access Series in Informatics (OASIcs)},
ISBN = {978-3-95977-455-0},
ISSN = {2190-6807},
year = {2026},
volume = {148},
editor = {Pill, Ingo and Zanella, Marina and Provan, Gregory},
publisher = {Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
address = {Dagstuhl, Germany},
URL = {https://drops.dagstuhl.de/entities/document/10.4230/OASIcs.DX.2026.14},
URN = {urn:nbn:de:0030-drops-278284},
doi = {10.4230/OASIcs.DX.2026.14},
annote = {Keywords: LLM diagnosis performance, SLM diagnosis performance, comparative study}
}
Document
DX Competition
Physics-Informed ML Diagnostic Engine: DXC'26 LiU-ICE Track Solution (DX Competition)
Abstract
The paper presents a solution to the DX'26 Competition LiU-ICE Track. The presented approach merges physics insights with data-driven residuals. The applied Grey-Box approach, involving the custom implementation of a recurrent LSTM network with state memory, is well-suited to environments with high inertia. The system does not directly estimate specific values of individual physical quantities, but focuses on determining their rate and direction of change, eliminating the problem of exploding residuals. Furthermore, this approach offers an advantage over classical LSTMs due to its physical representation of states. Structural analysis and the determination of the MSOs enabled the isolation of relationships between the observed variables, whilst simultaneously decoupling the training data from physical quantities unrelated to the variable under investigation. Data decoupling was leveraged to selectively incorporate faulty data into training. The solution implements four virtual sensors, each equipped with its own pair of neural networks. The proposed fault isolation approach is based on a fault diagnosis matrix, fixed alarm thresholds, and scalar product analysis. Additionally, adaptive covariate shift compensation mechanism was introduced to handle changes in external conditions.
The resulting algorithm proved to be highly resilient to measurement noise and capable of accurately analysing the process’s variability, scoring True Detection Rate TDR = 92.26%, False Alarm Rate FAR = 7.86%, and True Isolation Rate TIR = 79.62% on hidden test data.
Cite as
Antoni Rogowski and Anna Sztyber-Betley. Physics-Informed ML Diagnostic Engine: DXC'26 LiU-ICE Track Solution (DX Competition). In 37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026). Open Access Series in Informatics (OASIcs), Volume 148, pp. 15:1-15:20, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2026)
Copy BibTex To Clipboard
@InProceedings{rogowski_et_al:OASIcs.DX.2026.15,
author = {Rogowski, Antoni and Sztyber-Betley, Anna},
title = {{Physics-Informed ML Diagnostic Engine: DXC'26 LiU-ICE Track Solution}},
booktitle = {37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026)},
pages = {15:1--15:20},
series = {Open Access Series in Informatics (OASIcs)},
ISBN = {978-3-95977-455-0},
ISSN = {2190-6807},
year = {2026},
volume = {148},
editor = {Pill, Ingo and Zanella, Marina and Provan, Gregory},
publisher = {Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
address = {Dagstuhl, Germany},
URL = {https://drops.dagstuhl.de/entities/document/10.4230/OASIcs.DX.2026.15},
URN = {urn:nbn:de:0030-drops-278291},
doi = {10.4230/OASIcs.DX.2026.15},
annote = {Keywords: Diagnosis, Algorithms, Evaluation, LiU-ICE Benchmark, Gray-Box, Physics Informed Neural Networks}
}
Document
Short Paper
Counterfactual Fault Responsibility in a Dynamic Structural Causal Model of a DC Motor (Short Paper)
Abstract
Model-based diagnosis can identify fault hypotheses that are compatible with observed abnormal behaviour. Still, it does not necessarily determine which fault is responsible for a particular symptom when several faults are active simultaneously. We propose a counterfactual framework for post-diagnostic fault responsibility analysis in dynamical systems. The physical process is represented as a time-indexed structural causal model, while faults are encoded as interventions on individual system mechanisms. Responsibility is evaluated by comparing the factual system trajectory with counterfactual trajectories in which candidate faults are removed while the operating conditions and control structure are kept fixed. We instantiate the framework on an electro-thermo-mechanical model of a DC motor and evaluate it in two simulated multi-fault scenarios. The results show that counterfactual fault removal can separate the contributions of different faults to individual symptoms and distinguish severity contribution from necessity for a failure event. The proposed approach therefore complements conventional diagnosis with symptom-level causal explanations of faulty system behaviour.
Cite as
Emir Mujić and Franz Wotawa. Counterfactual Fault Responsibility in a Dynamic Structural Causal Model of a DC Motor (Short Paper). In 37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026). Open Access Series in Informatics (OASIcs), Volume 148, pp. 16:1-16:14, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2026)
Copy BibTex To Clipboard
@InProceedings{mujic_et_al:OASIcs.DX.2026.16,
author = {Muji\'{c}, Emir and Wotawa, Franz},
title = {{Counterfactual Fault Responsibility in a Dynamic Structural Causal Model of a DC Motor}},
booktitle = {37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026)},
pages = {16:1--16:14},
series = {Open Access Series in Informatics (OASIcs)},
ISBN = {978-3-95977-455-0},
ISSN = {2190-6807},
year = {2026},
volume = {148},
editor = {Pill, Ingo and Zanella, Marina and Provan, Gregory},
publisher = {Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
address = {Dagstuhl, Germany},
URL = {https://drops.dagstuhl.de/entities/document/10.4230/OASIcs.DX.2026.16},
URN = {urn:nbn:de:0030-drops-278304},
doi = {10.4230/OASIcs.DX.2026.16},
annote = {Keywords: Counterfactual Reasoning, Fault Responsibility, Structural Causal Models, Dynamic Systems, Model-Based Diagnosis}
}
Document
Short Paper
When Dynamic Slicing Helps and Hurts Spectrum-Based Fault Localization: Evidence from an Expanded TCAS Benchmark (Short Paper)
Abstract
Spectrum-based fault localization and dynamic program slicing are two established approaches to automated software debugging. A recent study provided a first experimental evaluation of a hybrid approach that combined both techniques, reporting that the combination improves the wasted effort metric for programs with multiple output variables. However, as stated in the study, further experiments are required for a final judgment of the introduced debugging methodology. In this paper, we review and extend that evaluation through three rounds of experiments, all carried out under a strict single-fault assumption. First, we replicate the third experiment from the original study to validate the obtained findings. Second, we expand the TCAS benchmark from 2 to 11 faulty variants by systematically generating nine new mutants spanning six distinct mutation operator families, and verify their quality through a dedicated kill-ability assessment. Eleven of the twelve mutants pass our quality filter and are used in two independent rounds of extended experiments , each evaluated with three spectrum-based fault localization coefficients (Ochiai, Tarantula, Sarhan-Beszédes) in both the plain and hybrid configurations. Our results confirm that the slicer’s behavior has a decisive impact on the quality of the hybrid approach and that improvements from slicing are far from uniform across mutation types. In particular, bugs that suppress the execution of entire program paths present a fundamental challenge for dynamic slicing, providing concrete evidence for the limitations reported in the original study.
Cite as
Iulia Nica and Franz Wotawa. When Dynamic Slicing Helps and Hurts Spectrum-Based Fault Localization: Evidence from an Expanded TCAS Benchmark (Short Paper). In 37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026). Open Access Series in Informatics (OASIcs), Volume 148, pp. 17:1-17:15, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2026)
Copy BibTex To Clipboard
@InProceedings{nica_et_al:OASIcs.DX.2026.17,
author = {Nica, Iulia and Wotawa, Franz},
title = {{When Dynamic Slicing Helps and Hurts Spectrum-Based Fault Localization: Evidence from an Expanded TCAS Benchmark}},
booktitle = {37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026)},
pages = {17:1--17:15},
series = {Open Access Series in Informatics (OASIcs)},
ISBN = {978-3-95977-455-0},
ISSN = {2190-6807},
year = {2026},
volume = {148},
editor = {Pill, Ingo and Zanella, Marina and Provan, Gregory},
publisher = {Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
address = {Dagstuhl, Germany},
URL = {https://drops.dagstuhl.de/entities/document/10.4230/OASIcs.DX.2026.17},
URN = {urn:nbn:de:0030-drops-278314},
doi = {10.4230/OASIcs.DX.2026.17},
annote = {Keywords: software fault localization, dynamic slicing, spectrum-based fault localization, mutation testing, automated debugging, TCAS}
}
Document
Extended Abstract
Summary Of: Modelling Cyber-Physical Systems for Fault Diagnosis (Extended Abstract)
Abstract
This is an extended abstract of the manuscript 'Modelling Cyber-Physical Systems for Fault Diagnosis' [Diedrich et al., 2025] that was published in the journal IEEE Transactions on Systems, Man, and Cybernetics in December 2025.
Cite as
Alexander Diedrich, Mattias Krysander, René Heesch, and Oliver Niggemann. Summary Of: Modelling Cyber-Physical Systems for Fault Diagnosis (Extended Abstract). In 37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026). Open Access Series in Informatics (OASIcs), Volume 148, pp. 18:1-18:6, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2026)
Copy BibTex To Clipboard
@InProceedings{diedrich_et_al:OASIcs.DX.2026.18,
author = {Diedrich, Alexander and Krysander, Mattias and Heesch, Ren\'{e} and Niggemann, Oliver},
title = {{Summary Of: Modelling Cyber-Physical Systems for Fault Diagnosis}},
booktitle = {37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026)},
pages = {18:1--18:6},
series = {Open Access Series in Informatics (OASIcs)},
ISBN = {978-3-95977-455-0},
ISSN = {2190-6807},
year = {2026},
volume = {148},
editor = {Pill, Ingo and Zanella, Marina and Provan, Gregory},
publisher = {Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
address = {Dagstuhl, Germany},
URL = {https://drops.dagstuhl.de/entities/document/10.4230/OASIcs.DX.2026.18},
URN = {urn:nbn:de:0030-drops-278328},
doi = {10.4230/OASIcs.DX.2026.18},
annote = {Keywords: Fault Diagnosis, SMT, Multimode Systems}
}
Document
Extended Abstract
Uncertainty-Aware Fault Diagnosis of Unknown Faults Using Ensemble-Based NODE Residuals (Extended Abstract)
Abstract
This extended abstract presents an uncertainty-aware approach to data-driven fault diagnosis using ensembles of neural ordinary differential equations (NODE). The proposed method leverages physical insights in the design of NODE models for residual generation and employs bootstrap ensembles to quantify prediction uncertainty and adapt detection thresholds. The adaptive thresholds enable robust detection and isolation of faults in scenarios with limited training data and reduces the risk of false alarms due to out-of-distribution data. The methodology is validated on real-world data from the LiU-ICE benchmark, providing insights into how training data quality and residual design impact uncertainty-aware fault diagnosis.
Cite as
Daniel Jung and Theodor Westny. Uncertainty-Aware Fault Diagnosis of Unknown Faults Using Ensemble-Based NODE Residuals (Extended Abstract). In 37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026). Open Access Series in Informatics (OASIcs), Volume 148, pp. 19:1-19:6, Schloss Dagstuhl – Leibniz-Zentrum für Informatik (2026)
Copy BibTex To Clipboard
@InProceedings{jung_et_al:OASIcs.DX.2026.19,
author = {Jung, Daniel and Westny, Theodor},
title = {{Uncertainty-Aware Fault Diagnosis of Unknown Faults Using Ensemble-Based NODE Residuals}},
booktitle = {37th International Conference on Principles of Diagnosis and Resilient Systems (DX 2026)},
pages = {19:1--19:6},
series = {Open Access Series in Informatics (OASIcs)},
ISBN = {978-3-95977-455-0},
ISSN = {2190-6807},
year = {2026},
volume = {148},
editor = {Pill, Ingo and Zanella, Marina and Provan, Gregory},
publisher = {Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
address = {Dagstuhl, Germany},
URL = {https://drops.dagstuhl.de/entities/document/10.4230/OASIcs.DX.2026.19},
URN = {urn:nbn:de:0030-drops-278334},
doi = {10.4230/OASIcs.DX.2026.19},
annote = {Keywords: Data-driven diagnosis, model-based diagnosis, structural methods, neural ordinary differential equations, ensemble models}
}