Inferring Mechanical Fault Models from the Electrical Domain

In the context of Industry 4.0, it is strategic to build a simulable model of an Industrial Cyber-Physical System (CPS) to ensure proper maintenance and early risk assessment to avoid monetary losses. To achieve this, it is necessary to use dedicated techniques for modeling and injecting faults into a simulative model. However, it is generally too complex due to heterogeneous components, e.g., analog and digital parts. Verilog-AMS is a suitable solution to overcome this problem since it allows the covering of different physical descriptions, starting from transistor-level to multi-discipline models (e.g., mechanic, thermic, fluid dynamic). This article proposes a methodology that exploits the specific analogy between mechanical and electrical domains. It starts from a mechanical model, builds the electrical equivalent, and injects electrical faults. The analysis of the injected faults allows building a generic taxonomy for mapping electrical faults onto mechanical ones. The final goal is to support the construction of Failure Mode and Effect Analysis (FMEA) principles in mechanical systems and the prospect of enabling predictive maintenance techniques.