A Bond Graph Approach to Detect and Isolate Fault on Hybrid Multi-Source System

The emergence of hybrid multi-source platforms presents a promising solution to address the escalating demand for electricity while aligning with sustainable practices. Given the reliance on renewable energy resources, ensuring the reliability of such systems is paramount. This paper proposes a methodology employing bond graph modeling for the design of fault detection and isolation algorithms. The approach utilizes Analytical Redundancy Equations (ARRs) as residual signal generators, embodying the energy conversion principles dictated by the bond graph representation of the system. To generate accurate ARRs, the model must exhibit derivative causality to obviate the need for explicit consideration of unknown initial conditions. This framework enables the detection of diverse fault types and mitigates false alarms through miss detection prevention. The proposed approach is validated using the 20sim software, a well-established platform for bond graph modeling. This methodology offers notable advantages, including expeditious graphical modeling and straightforward implementation of ARRs.