Data Driven Diagnosis for Large Cyber-Physical-Systems with Minimal Prior Information

2025-06-12

AI Summary

Root cause diagnosis in large-scale cyber-physical systems (CPS) remains challenging under conditions of incomplete system models and scarce labeled fault data. Method: This paper proposes a lightweight graph-based diagnostic framework that relies solely on the basic inter-subsystem connectivity structure and normal-operation data. It integrates an unsupervised neural symptom generator for anomaly representation learning with a sparse causal graph–driven, interpretable inference algorithm—enabling high-confidence root cause localization and search-space compression under minimal causal priors. Contribution/Results: The framework eliminates dependence on comprehensive system models or labeled fault datasets, establishing a novel diagnostic paradigm characterized by low prior knowledge requirements, high interpretability, and strong generalization. Evaluated on synthetic data, it correctly identifies true root causes in 82% of cases and significantly narrows the diagnostic scope in 73% of scenarios. Validation on the real-world SWaT water treatment testbed confirms its practical efficacy and operational feasibility.

Technology Category

• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning
• Reasoning under Uncertainty: Causality
• Machine Learning: Causal Learning

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Diagnostic processes for complex cyber-physical systems often require extensive prior knowledge in the form of detailed system models or comprehensive training data. However, obtaining such information poses a significant challenge. To address this issue, we present a new diagnostic approach that operates with minimal prior knowledge, requiring only a basic understanding of subsystem relationships and data from nominal operations. Our method combines a neural network-based symptom generator, which employs subsystem-level anomaly detection, with a new graph diagnosis algorithm that leverages minimal causal relationship information between subsystems-information that is typically available in practice. Our experiments with fully controllable simulated datasets show that our method includes the true causal component in its diagnosis set for 82 p.c. of all cases while effectively reducing the search space in 73 p.c. of the scenarios. Additional tests on the real-world Secure Water Treatment dataset showcase the approach's potential for practical scenarios. Our results thus highlight our approach's potential for practical applications with large and complex cyber-physical systems where limited prior knowledge is available.

Problem

• Diagnosing large cyber-physical systems with minimal prior knowledge
• Reducing dependency on detailed system models or training data
• Improving accuracy and efficiency in fault detection for complex systems

Innovation

• Neural network-based symptom generator for anomaly detection
• Graph diagnosis algorithm using minimal causal relationships
• Operates with minimal prior knowledge and nominal data