A deep feature-based approach for classifying flue gas pollutants and hazard identification using pyramid-dilated temporal convolution networks
Abstract
Nowadays, real-world monitoring and accurate prediction for toxic gas concentration have gained much attention from researchers, as it provides great work in emergency validation situations and rescue works. The Artificial Intelligence (AI)-based gas concentration prediction techniques face more problems in terms of poor convergence speed, equal feature importance, and minimal accuracy. But, choosing the best design and also validating the most suitable safety measures are difficult due to the lack of execution of the important field examinations under emission regulations. So, a deep learning-based flue gas analysis model is implemented for hazard detection. Here, required samples are gathered from benchmark resources and features are extracted through Convolutional-Autoencoder with Spatial–Temporal Attention (CA-STA). The flue gas classification is performed using a Pyramid Dilated-based Adaptive Temporal Convolution Network (PD-ATCN), where the extracted features are given as inputs. The flexibility of the suggested approach is enhanced by fine-tuning the parameters of PD-ATCN using Improved Random Variable-based Cray Fish Optimization (IRV-CFO). Finally, the effectiveness of the proposed framework is evaluated through diverse performance metrics, which show higher robustness. In the accuracy analysis, the developed IRV-CFO-PD-ATCN gained superior typical flue gas pollutants classification and hazards identification outcome as 95.11% for dataset 1 and 92.6% for dataset 2, compared to baseline approaches. Hence, the experimental results showed that the suggested approach is capable of achieving more precise typical flue gas classification and hazards detection outcomes than others under dynamic conditions.
Related
- Spatio-Temporal Assessment of Atmospheric Pollutants and Dispersion Mechanisms over West Bengal (2019–2024): Insights from Sentinel-5P, GEOS-CHEM, and HYSPLIT Models
- Reservoir eutrophication prediction with explainable framework integrating temporal convolutional networks and bidirectional sequence modeling
- A hybrid spatio-temporal deep learning framework for urban travel time prediction and intelligent routing: Integrating recurrent and graph neural networks
- Measurements of Hazardous Air Pollutants in Communities Necessitate Review of Air Quality and Exposure Metrics
- Characteristics and simulation of ozone pollution in Haikou City under combined effects of meteorological factors and pollutants
BibTeX
@article{renganathan2026deep,
title = {{A deep feature-based approach for classifying flue gas pollutants and hazard identification using pyramid-dilated temporal convolution networks}},
author = {K. Renganathan and B. Umamaheswari},
journal = {Scientific Reports},
year = {2026},
month = oct,
doi = {10.1038/s41598-026-70943-2},
url = {https://www.nature.com/articles/s41598-026-70943-2},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - JOUR
TI - A deep feature-based approach for classifying flue gas pollutants and hazard identification using pyramid-dilated temporal convolution networks
AU - K. Renganathan
AU - B. Umamaheswari
PY - 2026
DA - 2026/10/15
JO - Scientific Reports
DO - 10.1038/s41598-026-70943-2
UR - https://www.nature.com/articles/s41598-026-70943-2
AB - Nowadays, real-world monitoring and accurate prediction for toxic gas concentration have gained much attention from researchers, as it provides great work in emergency validation situations and rescue works. The Artificial Intelligence (AI)-based gas concentration prediction techniques face more problems in terms of poor convergence speed, equal feature importance, and minimal accuracy. But, choosing the best design and also validating the most suitable safety measures are difficult due to the lack of execution of the important field examinations under emission regulations. So, a deep learning-based flue gas analysis model is implemented for hazard detection. Here, required samples are gathered from benchmark resources and features are extracted through Convolutional-Autoencoder with Spatial–Temporal Attention (CA-STA). The flue gas classification is performed using a Pyramid Dilated-based Adaptive Temporal Convolution Network (PD-ATCN), where the extracted features are given as inputs. The flexibility of the suggested approach is enhanced by fine-tuning the parameters of PD-ATCN using Improved Random Variable-based Cray Fish Optimization (IRV-CFO). Finally, the effectiveness of the proposed framework is evaluated through diverse performance metrics, which show higher robustness. In the accuracy analysis, the developed IRV-CFO-PD-ATCN gained superior typical flue gas pollutants classification and hazards identification outcome as 95.11% for dataset 1 and 92.6% for dataset 2, compared to baseline approaches. Hence, the experimental results showed that the suggested approach is capable of achieving more precise typical flue gas classification and hazards detection outcomes than others under dynamic conditions.
SN - 2045-2322
ER -