A hierarchical object oriented Bayesian network-based fault diagnosis method for building energy systems. (15th January 2022)

• Record Type: Journal Article
• Authors: Li, Tingting; Zhou, Yangze; Zhao, Yang; Zhang, Chaobo; Zhang, Xuejun
• Is Part Of: Applied energy. Volume 306:Part B(2022)
• Publication Date: 2022-01-15
• Subjects: Hierarchical object oriented Bayesian network -- Fault diagnosis -- Building energy systems -- Building energy conversation; Power (Mechanics) -- Periodicals; Energy conservation -- Periodicals; Energy conversion -- Periodicals; 621.042
• DOI: 10.1016/j.apenergy.2021.118088
• Languages: English
• ISSNs: 0306-2619

Abstract

Highlights: Classes are developed to provide reusable Bayesian network fragments. Three levels of classes are proposed to construct a standard class hierarchy. The feature of inheritance is applied to describe the similarities between classes. Hierarchical object oriented Bayesian networks are proposed for fault diagnosis. The proposed method is suitable for fault diagnosis of large-scale complex systems. Abstract: Bayesian network is a powerful algorithm to diagnose the faults in building energy systems based on incomplete and uncertain diagnostic information. In practice, it is very challenging to construct Bayesian networks for large-scale and complex systems. Inspired by the object oriented programming technology, a hierarchical object oriented Bayesian network-based method is proposed in this study. Its basic idea is to reuse the standard Bayesian network fragments predefined in the classes to generate the system-level fault diagnosis models for target systems. Inheritance is adopted to avoid inconsistent modeling of similar classes. It allows sub-classes to inherit the Bayesian network fragments from their super-classes. For a specific building energy system, the fragments are reused and combined to generate a hierarchical object oriented Bayesian network for real-time fault diagnosis. The proposed method is evaluated using the experimental data from an industrial building. The results show that the proposed method can provide customized fault diagnosis solutions for complex building energy systems without tedious and repeated modeling works. Most of the typical faults are successfully isolated.