APPLICATION OF ONTOLOGICAL MODELS IN FAULT DIAGNOSIS FOR CYBER-PHYSICAL PRODUCTION SYSTEMS

INTRODUCTION: Advanced manufacturing demands continuous integration between the digital product model and shop-floor manufacturing decisions. Although automated feature recognition already converts CAD model geometry into manufacturing entities, a gap remains between the recognised features and the validation of the manufacturing decisions (machine, process, and tool) made by the engineer, whose incorrect choices propagate errors, costs, and rework. AIMS: This project applied ontological models to diagnose inconsistencies in production systems, developing a knowledge base capable of inferring, for each feature of a sheet metal part, the feasible manufacturing resources, and of confronting the choices declared by the user against that feasibility, detecting and classifying planning inconsistencies before or after execution. MATERIALS AND METHODS: The solution was built as an application ontology in OWL, resting on two reference-level ontologies: a reused materials ontology and a manufacturing processes and resources ontology built for this purpose, with SWRL rules processed by the Pellet reasoner and Python integration to populate individuals from real data produced by automated feature extraction from parts in STEP format. The modelling explicitly separates the user’s choices (chosen properties) from the recommendations inferred by the system (recommended properties), keeping the correct answer visible alongside the detected error. Anomalies are classified by inference into a class hierarchy: wrong machine, wrong process and wrong tool. RESULTS: The knowledge base consolidated 32 SWRL rules organised into seven functional blocks and was demonstrated on a validation part with 25 features, in which 20 choice scenarios exercised the detection rules: incorrect choices of machine, process and tool were automatically flagged with the specific type of error, while keeping the correct machine, process and tool visible. FINAL CONSIDERATIONS: The results demonstrate the feasibility of diagnosing planning inconsistencies through ontological reasoning, with the anomaly type detected in its entirety within the ontology, establishing the prescriptive layer required for future integration with cyber-physical production systems and digital twins.

PALAVRAS-CHAVE: Ontology; Inconsistency diagnosis; OWL; SWRL; Cyber-physical production systems.