Bond Graph and AI as a hybrid model for fault diagnosis at the scale of a complete machine-tool
Abstract
This article proposes a methodology for fault diagnosis on a whole machine-tool in low data limit environments. The approach involves data selection with dependability analysis and Fault Detection and Isolation (FDI) tools additionally to physic laws for using AI only where it is necessary. Physics laws are based on Bond Graphs for modelling healthy behavior, simulating faults, determining dynamic thresholds and applying Hamming Distance. AI will support the diagnosis by isolating the remaining non-isolable anomalies from faulty signals. This study focuses on a linear axis to validate the methodology.
Fields of Study & Subjects
Fault Detection and Control Systems
Machine Fault Diagnosis Techniques
Engineering and Test Systems
Publication details
- Journal / source
- Procedia CIRP
- Volume
- 138
- Pages
- 192–197
- Publication date
- 1 January 2026
- Language
- EN