Chenjun Ning
Research Subjects
Chenjun Ning has a total of 6 co-authors. Here are the top co-authors by number of shared publications:
- Chen Chen
College of Instrumentation and Electrical Engineering, Key Laboratory of Geophysical Exploration Equipment, Ministry of Education of China, Jilin University, Changchun, China
1 shared publication - Jiale Zheng
State Key Laboratory of Petroleum Pollution Control, Beijing 102206, China
1 shared publication - Lei Li
Department of Electronics and Electrical Engineering, Changchun University of Technology, Changchun 130012, China
1 shared publication - Shanshan Li
Department of Electronics and Electrical Engineering, Changchun University of Technology, Changchun 130012, China
1 shared publication - Wenqi Fan
CNPC Research Institute of Safety and Environmental Technology, Beijing 102206, China
1 shared publication - Xinglei Zhao
State Key Laboratory of Petroleum Pollution Control, Beijing 102206, China
1 shared publication
Latest Publications
- Open Access
Hybrid Deep Learning Neural Networks for Small-Molecule Organic Amine Gas Recognition and Prediction Based on MOS Sensor Array
Carbon capture, utilization and storage (CCUS) is a critical technology for the fossil energy industry to achieve the dual carbon goals. Organic amine-based absorption methods, represented by mono-ethanolamine (MEA), methyl-diethanolamine (MDEA) and 2-amino-2-methyl-1-propanol (AMP), are currently the most widely adopted approaches for carbon dioxide capture. If the concentration of leaked organic amines exceeds the safety threshold, inhalation will cause severe respiratory irritation and even serious illnesses in humans. Accordingly, in situ monitoring of organic amine concentrations in waste gas is of great significance for process optimization, environmental protection, early health warning, energy conservation and emission reduction. In this study, a metal oxide semiconductor (MOS) sensor array was developed to identify categories and concentration variations in small-molecule organic amines. To realize effective gas classification and accurate concentration prediction, single-component gas data and mixed gas data with varying concentrations of the three amines collected in the laboratory were adopted to train a one-dimensional convolutional neural network (1D-CNN) and a bidirectional gated recurrent unit (Bi-GRU) via five-fold cross-validation. The proposed model achieved a classification accuracy of 0.9969 ± 0.0006. For MEA concentration prediction, the optimal determination coefficient (R2), root mean square error (RMSE), and mean absolute error (MAE) were 0.9969, 2.1306 and 1.5503, respectively. The experimental results demonstrate that the proposed method possesses promising practical application prospects in in situ gas monitoring and early hazard warning.
Open AccessHybrid Deep Learning Neural Networks for Small-Molecule Organic Amine Gas Recognition and Prediction Based on MOS Sensor Array
Carbon capture, utilization and storage (CCUS) is a critical technology for the fossil energy industry to achieve the dual carbon goals. Organic amine-based absorption methods, represented by mono-ethanolamine (MEA), methyl-diethanolamine (MDEA) and 2-amino-2-methyl-1-propanol (AMP), are currently the most widely adopted approaches for carbon dioxide capture. If the concentration of leaked organic amines exceeds the safety threshold, inhalation will cause severe respiratory irritation and even serious illnesses in humans. Accordingly, in situ monitoring of organic amine concentrations in waste gas is of great significance for process optimization, environmental protection, early health warning, energy conservation and emission reduction. In this study, a metal oxide semiconductor (MOS) sensor array was developed to identify categories and concentration variations in small-molecule organic amines. To realize effective gas classification and accurate concentration prediction, single-component gas data and mixed gas data with varying concentrations of the three amines collected in the laboratory were adopted to train a one-dimensional convolutional neural network (1D-CNN) and a bidirectional gated recurrent unit (Bi-GRU) via five-fold cross-validation. The proposed model achieved a classification accuracy of 0.9969 ± 0.0006. For MEA concentration prediction, the optimal determination coefficient (R2), root mean square error (RMSE), and mean absolute error (MAE) were 0.9969, 2.1306 and 1.5503, respectively. The experimental results demonstrate that the proposed method possesses promising practical application prospects in in situ gas monitoring and early hazard warning.