This paper proposes a data-driven fault detection approach for nuclear power plant. The approach starts from input and output (I/O) data obtained from operating data of industrial process. Due to the model is not explicitly appeared, the proposed approach is named as implicit model approach (IMA). Residual generator is obtained directly from I/O data rather than from the mechanism, based which the algorithm of IMA-based fault detection is proposed. The main advantage of IMA-based fault detection is that it can circumvent complicated model identification. The approach generates parameterized matrices of residual signal inspired by subspace relevant technology without any prior knowledge about mechanisms of the plant. Fault information has been injected to a simulating platform of a compact reactor in the simulation part, by which we verify the effectiveness of IMA-based fault detection.
- Nuclear Engineering Division
Data-Driven Fault Diagnosis for Nuclear Power Plant: The Implicit Model Approach
- Views Icon Views
- Share Icon Share
- Search Site
Chen, Z, Li, X, Ke, Z, Tao, M, & Feng, Y. "Data-Driven Fault Diagnosis for Nuclear Power Plant: The Implicit Model Approach." Proceedings of the 2018 26th International Conference on Nuclear Engineering. Volume 4: Nuclear Safety, Security, and Cyber Security; Computer Code Verification and Validation. London, England. July 22–26, 2018. V004T06A043. ASME. https://doi.org/10.1115/ICONE26-82473
Download citation file: