Name
Title | ||
|---|---|---|
Fault Root Cause Diagnosis Method Based on Recurrent Neural Network and Granger Causality |
Abstract | ||
|---|---|---|
Data-driven fault diagnosis methods are more and more important in modern industry. At the same time, the research on the cause of fault location when they occur is more and more advanced. The Granger Causality (GC) test method is difficult to analyze nonlinear signals and redundant causality when locating fault root, which limited the application of this method. To solve this problem, a fault root diagnosis method based on Recurrent Neural Network (RNN) and GC is proposed. Firstly, the Principal Component Analysis (PCA) model is established for fault detection, and the contribution plot method is used to select the variables that may fault when the fault occurred. Then, Dynamic Time Warping (DTW) is used to group fault candidate variables to reduce redundant causality. Finally, RNN is integrated with GC technology, which enabled it to deal with nonlinear signals and locate the root cause of faults. The method is applied to the simulation experiment of TE chemical process, and the experimental results show that the method can locate the fault root and identify the fault propagation path accurately. |
Year | DOI | Venue |
|---|---|---|
2021 | 10.1109/SAFEPROCESS52771.2021.9693579 | 2021 CAA Symposium on Fault Detection, Supervision, and Safety for Technical Processes (SAFEPROCESS) |
Keywords | DocType | ISBN |
root cause diagnosis,data drive,Recurrent Neural Network (RNN),Granger Causality (GC) | Conference | 978-1-6654-0116-6 |
Citations | PageRank | References |
0 | 0.34 | 0 |
Authors | ||
4 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Gang Shen | 1 | 0 | 0.68 |
| Peiliang Wang | 2 | 0 | 0.68 |
| Kailiang Hu | 3 | 0 | 0.68 |
| Qiuyang Ye | 4 | 0 | 0.68 |