.png){kind=icon}
Return
Fusing multi-sensor data for bag filter system risk early warning based on deep learning
.png){kind=icon}
PRE
AI
DOI：10.1016/j.psep.2025.107096.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
A risk early warning method for the bag filter system fusing multi-sensor data based on the MTCN-Q-Informer model and the cloud model (CM) is proposed,which is essential in preventing dust explosion accidents. Firstly, based on the multi-sensor data of the bag filter system, which includes monitoring parameters such as the temperature in bag filter box, the temperature in dust hopper, the difference between inlet and outlet pressure, the gas source pressure for filter ash cleaning, and the spray water pressure of the fire extinguishing device, the data feature sequences are extracted through the multi-scale temporal convolutional networks (MTCN). Secondly, the extracted feature sequences are input into the Q-learning optimized Informer model to obtain predicted values for monitoring parameters. Subsequently, the CM is built to integrate predicted data from multiple sensors, enabling collaborative multi-sensor risk early warning for the bag filter system. Finally, in the case study, the result of monitoring parameters prediction shows that the MTCN-Q-Informer model reduces the mean RMSE by 26.9 % and the mean MAE by 41.8 % compared to the Informer model. By combining the CM on this basis, we achieve a functional risk early warning for the bag filter system that lasts for 120 min.
Keywords:
Bag filter system
Risk early warning
Informer
Multi-scale temporal convolutional networks
Q -learning
Multi-sensor data
Journal
IF:
7.8
Papers:
1.0W
Citations:
3.8W
Organization
Cited Papers
Application of Deep Learning Algorithm in Optimization Control of Electrostatic Precipitator in Coal-Fired Power Plants.png)
Processes
OAAI
ProcessesIF0
OAAIShareSave
ShareSaveShareSaveShareSaveShareSaveShareSave
View more