ONYXIS: Ontology-Driven Intelligent Fault Diagnosis and Knowledge Graph Reasoning System for Spacecraft

ONYXIS is an open-source project that combines semantic web technology, ontology engineering, and spacecraft telemetry data. Its core is to use knowledge graphs and SWRL rule reasoning to achieve real-time monitoring, fault diagnosis, and autonomous decision-making for satellite subsystems. The project aims to address the limitations of traditional spacecraft monitoring methods and provide an interpretable intelligent monitoring closed loop.

Spacecraft systems are complex engineering systems consisting of dozens of subsystems. Traditional monitoring relies on static thresholds and manual rules, making it difficult to capture complex dependencies between subsystems and unable to provide an interpretable fault reasoning process. ONYXIS proposes introducing semantic web technology, ontology engineering, and knowledge graph reasoning into the field of spacecraft health monitoring to achieve a complete closed loop from telemetry data to fault diagnosis and autonomous response.

The core of ONYXIS is a comprehensive ontology model covering the main subsystems of spacecraft (130+ classes, 700+ instances, 30+ object properties, 45+ data properties), divided into functional domains such as Telemetry, Tracking, and Command (TT&C), Attitude Determination and Control System (ADCS), power management, thermal control, and propulsion. The system uses a dynamic telemetry simulation method to continuously generate cross-subsystem telemetry data and update ontology instances in real time, supporting continuous subsystem evolution, reasoning visibility, and scenario-based demonstrations (normal operation, various fault scenarios).

ONYXIS includes approximately 64 SWRL rules covering telemetry threshold reasoning, subsystem fault inference, operational response, etc. The reasoning process is: Telemetry generation → Scenario change → Ontology instance update → Pellet reasoner execution → SWRL rule execution → Fault inference → Operational response → Visualization. Typical examples:

• Thermal control fault: Battery temperature >50°C → Overheating fault → Trigger safe mode;
• Communication fault: Signal strength 12dBm + Transmit power 3W → TT&C radio fault → Communication degradation;
• Attitude control fault: Tracking accuracy 0.4° + Pointing error 6° → Tracking fault → Antenna pointing fault. The system uses controlled causal dependency reasoning to avoid over-propagation and introduces concepts such as communication degradation and performance degradation.

The core advantage of ONYXIS is interpretability: fault inferences can be traced back to original telemetry data, and response actions have clear rule-based foundations. Its application scenarios include: Education and training (teaching cases for aerospace and semantic web), research demonstration (potential of knowledge graphs in complex system monitoring), prototype development (concept verification for actual systems), and knowledge graph experiments (exploring the boundaries of semantic technology).

Future expansion directions of ONYXIS include: SHACL validation constraints, temporal reasoning, real-time dashboards, advanced causal dependency modeling, stream reasoning, and digital twin integration. The project demonstrates the potential of ontology engineering and knowledge graph reasoning in the field of spacecraft monitoring, providing valuable references for aerospace systems engineering and knowledge graph applications, and will play an important role in future scenarios such as deep space exploration and satellite constellation management.