Unified Framework for a Robust Fdi of a Fluid Benchmark System

This paper presents a unified diagnostic framework for a three-tank fluid benchmark system based on Hybrid Bond Graph (HBG) modeling, with a focus on enhancing fault detection and isolation (FDI) capabilities under parameter uncertainties. The proposed HBG model captures both continuous and discrete dynamics, accommodating discrete events such as valve switching. While HBG models enable effective diagnosis by generating analytical redundancy relations (ARRs), they are often sensitive to parameter variations, which can reduce reliability in uncertain environments. To address this problem, we introduce an incremental model derived from the HBG that accounts for deviations in system behavior due to parameter uncertainties, resulting in a more adaptive diagnostic process. By incrementally adjusting model parameters within the HBG framework, the proposed approach ensures that generated ARRs remain valid across varying operating conditions and configurations, regardless of minor parameter fluctuations. Simulation results demonstrate that the unified framework significantly improves robustness in fault detection and isolation, even with variable fluid levels, valve states, and component tolerances. This framework offers a practical FDI solution for three-tank benchmark systems with complex configurations, showing strong applicability in industrial settings where parameter uncertainty is prevalent.