Fault detection and isolation in system of multiple sources of energy using hierarchical Bayesian belief networks
Abstract
Ensuring fault tolerance in systems of multiple sources of energy (SMSE) is crucial for reliable operation. Detecting, localizing, and characterizing faults are essential tasks for system integrity. This paper presents a novel approach utilizing hierarchical Bayesian belief networks (HBBNs) to identify and isolate open-circuit faults in DC–DC power converters commonly employed in SMSE applications. Our method addresses the challenge of fault detection and isolation, significantly enhancing system reliability. In particular, we design a comprehensive system capable of detecting and isolating faults in a system of multiple DC–DC converters, leveraging the interpretability and efficiency of HBBNs. We also utilize measurements from other converters to detect and isolate faults in a single converter, enabling efficient fault management. Our approach utilizes regularly monitored variables of the system, eliminating the need for additional sensors, thereby reducing complexity and cost. Additionally, we generalize HBBNs to be adaptable to any number of converters, providing scalability and flexibility in fault detection and isolation. Notably, the interpretability and simplicity of HBBNs, with a small number of parameters compared to other data-driven methods such as neural networks, contribute to their effectiveness in fault management. Through extensive testing on simulated data generated via a developed state space model, our approach demonstrates its effectiveness in bolstering the robustness of DC–DC power converters against open-circuit faults.