Large-Scale Decentralized Fault Diagnosis for Multi-Group Data with Auxiliary Information via Distributed Multiple Testing
In the era of big data, efficiently diagnosing faults in high-dimensional data streams (HDS) is critical for numerous industrial applications. This paper addresses a novel decentralized fault diagnosis problem involving multi-group HDS with multi-sequence auxiliary information (MAI). Traditional diagnostic methods, which are designed for a single group of single-sequence HDS, struggle with the complexity and volume of such data, often leading to suboptimal diagnostic performance. To overcome this challenge, we propose a distributed fault diagnosis framework that leverages advanced multiple testing techniques and data fusion strategies to analyze multi-group HDS with MAI. Under this framework, we introduce a generalized multi-sequence local index of significance for the data streams in each group, based on a Cartesian hidden Markov model, to effectively fuse information from auxiliary sequences. This is then integrated into a distributed multiple testing procedure for group-wise diagnosis of the target data sequence. The proposed procedure minimizes the group-wise expected number of false positives in the target sequence while controlling the overall group-wise missed discovery rate at a specified level. Numerical studies demonstrate that the proposed method outperforms state-of-the-art diagnostic techniques, providing more reliable and effective fault diagnostics.
Authors
- Fugee Tsung (ORCID: https://orcid.org/0000-0002-0575-8254)
- Wendong Li (ORCID: https://orcid.org/0000-0001-5517-7856)
- Dongdong Xiang (ORCID: https://orcid.org/0000-0002-4609-7392)
- ZHANG Zhihan
Institutions
- Hong Kong University of Science and Technology (HK)
- East China Normal University (CN)
Publication Details
- Journal
- Technometrics
- Published
- 2026-11-12
- DOI
- https://doi.org/10.1080/00401706.2026.2763809
- Primary Topic
- Fault Detection and Control Systems
- Type
- article
- Field-Weighted Citation Impact
- 0.00