MSA-Net: A Precise and Robust Model for Predicting the Carbon Content on an As-Received Basis of Coal
Abstract
:1. Introduction
2. Analysis of Coal Carbon Content as Received
3. Coal Carbon Content as Received Prediction Model
3.1. Attention Module
3.2. MSA-Net Model
3.3. Loss Function
4. Training and Predicting Datasets and Evaluation Indicators
4.1. Complete Dataset Production
4.2. Training and Predicting Datasets Split Methods
4.2.1. Data Partitioning Based on Stratified Sampling
4.2.2. Data Partitioning Based on Odd or Even Months
4.2.3. Data Partitioning Based on Odd or Even Days
4.3. Evaluation Metrics
- Mean Absolute Error (MAE)
- 2.
- Root Mean Square Error (RMSE)
- 3.
- Mean Absolute Percentage Error (MAPE)
- 4.
- Coefficient of Determination (R2)
- 5.
- Pearson Correlation Coefficient (PCC)
- 6.
- Concordance Correlation Coefficient (CCC)
- 7.
- Explained Variance (Evar)
5. Experiments and Analysis
5.1. Implementation Details
5.2. Models Performance Experiments and Analysis
5.3. Ablation Analysis
5.4. Model Testing Experiments on Dividing Datasets on Odd and Even Months
5.5. Model Testing Experiments on Dividing Datasets on Odd and Even Days
6. Discussion
7. Conclusions
Author Contributions
Funding
Institutional Review Board Statement
Informed Consent Statement
Data Availability Statement
Conflicts of Interest
References
- Brown, H. Reducing the impact of climate change. Bull. World Health Organ. 2022, 85, 824–825. [Google Scholar]
- Li, J.; Wang, P.; Ma, S. The impact of different transportation infrastructures on urban carbon emissions: Evidence from China. Energy 2024, 295, 131041. [Google Scholar] [CrossRef]
- Wu, X.; Liu, P.; Yang, L.; Shi, Z.; Lao, Y. Impact of three carbon emission reduction policies on carbon verification behavior: An analysis based on evolutionary game theory. Energy 2024, 295, 130926. [Google Scholar] [CrossRef]
- Qiao, Q.; Eskandari, H.; Saadatmand, H.; Sahraei, M.A. An interpretable multi-stage forecasting framework for energy consumption and CO2 emissions for the transportation sector. Energy 2024, 286, 129499. [Google Scholar] [CrossRef]
- Ditl, P.; Šulc, R. Calculations of CO2 emission and combustion efficiency for various fuels. Energy 2024, 290, 130044. [Google Scholar] [CrossRef]
- International Energy Agency (IEA). CO2 Emissions in 2022. 2022. Available online: https://www.iea.org/reports/co2-emissions-in-2022 (accessed on 15 June 2024).
- Zhang, S.; Zhao, T.; Xie, B.-C.; Gao, J. What drives the GHG emission changes of the electric power industry in China? An empirical analysis using the logarithmic mean divisia index method. Carbon Manag. 2017, 8, 363–377. [Google Scholar] [CrossRef]
- Wang, Y.; Song, J.; Yang, W.; Dong, L.; Duan, H. Unveiling the driving mechanism of air pollutant emissions from thermal power generation in China: A provincial-level spatiotemporal analysis. Resour. Conserv. Recycl. 2019, 151, 104447. [Google Scholar] [CrossRef]
- GB/T 32151.1-2015; Requirements of the Greenhouse Gas Emission Accounting and Reporting—Part 1: Power Generation Enterprise. Standardization Administration of the People’s Republic of China: Beijing, China, 2015.
- Ministry of Ecology and Environment of the People’s Republic of China. Notice on Issuing the Guidelines for Accounting and Reporting of Green-House Gas Emissions by Enterprises—Power Generation Facilities and the Technical Guidelines for Verification of Greenhouse Gas Emissions by Enterprises Power Generation Facilities. 2022. Available online: https://www.mee.gov.cn/xxgk2018/xxgk/xxgk06/202212/t20221221_1008430.html (accessed on 15 June 2024).
- Ministry of Ecology and Environment of the People’s Republic of China. Notice on the Key Tasks of Efficient Coordination of Epidemic Prevention and Control and Economic and Social Development Adjustment for 2022 Enterprise Greenhouse Gas Emission Report Management. 2022. Available online: https://www.gov.cn/zhengce/zhengceku/2022-06/12/content_5695325.htm (accessed on 15 June 2024).
- Zhang, F.; Li, H.; Xu, Z.; Chen, W. A novel abrm model for predicting coal moisture content. J. Intell. Robot. Syst. 2022, 104, 30. [Google Scholar] [CrossRef] [PubMed]
- Liu, Y.; Zhang, H.; Zhang, X.; Qing, S.; Zhang, A.; Yang, S. Optimization of Combustion Characteristics of Blended Coals Based on TOPSIS Method. Complexity 2018, 2018, 4057983. [Google Scholar] [CrossRef]
- Shu, P.; Zhang, Y.; Deng, J.; Zhang, Y.; Duan, Z.; Li, L. Study on the effect of mixed particle size on coal heat release characteristics. Int. J. Coal Prep. Util. 2023, 43, 831–846. [Google Scholar] [CrossRef]
- Deng, S.; Hu, Y.; Chen, D.; Ma, Z.; Li, H. Integrated petrophysical log evaluation for coalbed methane in the hancheng area, China. J. Geophys. Eng. 2013, 10, 035009. [Google Scholar] [CrossRef]
- Guan, C.; Wu, T.; Chen, J.; Li, M. Detection of carbon content from pulverized coal using libs coupled with DSC-PLS method. Chemosensors 2022, 10, 490. [Google Scholar] [CrossRef]
- Jo, J.; Lee, D.-G.; Kim, J.; Lee, B.-H.; Jeon, C.-H. Improved ann-based approach using relative impact for the prediction of thermal coal elemental composition using proximate analysis. ACS Omega 2022, 7, 29734–29746. [Google Scholar] [CrossRef] [PubMed]
- Ceylan, Z.; Sungur, B. Estimation of coal elemental composition from proximate analysis using machine learning techniques. Energy Sources Part A Recov. Util. Environ. Eff. 2020, 42, 2576–2592. [Google Scholar] [CrossRef]
- Yin, L.; Liu, G.; Zhou, J.; Liao, Y.; Ma, X. A calculation method for CO2 emission in utility boilers based on bp neural network and carbon balance. Energy Procedia 2017, 105, 3173–3178. [Google Scholar] [CrossRef]
- Ashish, V.; Noam, S.; Niki, P.; Jakob, U.; Llion, J.; Aidan, N.G.; Lukas, K.; Illia, P. Attention is all you need. In Proceedings of the 31st Conference on Neural Information Processing Systems (NIPS 2017), Long Beach, CA, USA, 4–9 December 2017; Curran Associates Inc.: Red Hook, NY, USA, 2017. [Google Scholar]
- Liu, F.; Ren, X.; Zhang, Z.; Sun, X.; Zou, Y. Rethinking skip connection with layer normalization in transformers and resnets. arXiv 2021, arXiv:2105.07205. [Google Scholar]
- Zhu, D. Application of carbon ultimate analysis into greenhouse gas emissions accounting for coal-fired power plants. Power Gener. Technol. 2018, 39, 363–366. [Google Scholar]
- GB/T 30732-2014; Proximate Analysis of Coal—Instrumental Method. General Administration of Quality Supervision, Inspection and Quarantine of the People’s Republic of China. Standardization Administration of the People’s Republic of China: Beijing, China, 2014.
- GB/T 31391-2015; Ultimate Analysis of Coal. General Administration of Quality Supervision, Inspection and Quarantine of the People’s Republic of China. Standardization Administration of the People’s Republic of China: Beijing, China, 2015.
- GB/T 213-2008; Determination of Calorific Value of Coal. General Administration of Quality Supervision, Inspection and Quarantine of the People’s Republic of China. Standardization Administration of the People’s Republic of China: Beijing, China, 2008.
- GB/T 211-2017; Determination of Moisture in Coal. General Administration of Quality Supervision, Inspection and Quarantine of the People’s Republic of China. Standardization Administration of the People’s Republic of China: Beijing, China, 2017.
- GB/T 212-2008; Proximate Analysis of Coal. General Administration of Quality Supervision, Inspection and Quarantine of the People’s Republic of China. Standardization Administration of the People’s Republic of China: Beijing, China, 2008.
- GB/T 214-2007; Determination of Total Sulfur in Coal. General Administration of Quality Supervision, Inspection and Quarantine of the People’s Republic of China. Standardization Administration of the People’s Republic of China: Beijing, China, 2007.
- GB/T 476-2008; Determination of Carbon and Hydrogen in Coal. General Administration of Quality Supervision, Inspection and Quarantine of the People’s Republic of China. Standardization Administration of the People’s Republic of China: Beijing, China, 2008.
- Liu, J.; Sun, M.; Zhang, W.; Xie, G.; Jing, Y.; Li, X.; Shi, Z. Dae-ner: Dual-channel attention enhancement for chinese named entity recognition. Comput. Speech Lang. 2024, 85, 101581. [Google Scholar] [CrossRef]
- Kingma, D.P.; Ba, J. Adam: A method for stochastic optimization. arXiv 2014, arXiv:1412.6980. [Google Scholar]
- He, K.; Zhang, X.; Ren, S.; Sun, J. Deep residual learning for image recognition. In Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV, USA, 27–30 June 2016; pp. 770–778. [Google Scholar]
- Lipton, Z.C. A critical review of recurrent neural networks for sequence learning. arXiv 2015, arXiv:1506.00019. [Google Scholar]
- Mulás-Tejeda, E.; Gómez-Espinosa, A.; Escobedo Cabello, J.A.; Cantoral-Ceballos, J.A.; Molina-Leal, A. Implementation of a Long Short-Term Memory Neural Network-Based Algorithm for Dynamic Obstacle Avoidance. Sensors 2024, 24, 3004. [Google Scholar] [CrossRef] [PubMed]
- Du, Q.; Wang, Z.; Huang, P.; Zhai, Y.; Yang, X.; Ma, S. Remote Sensing Monitoring of Grassland Locust Density Based on Machine Learning. Sensors 2024, 24, 3121. [Google Scholar] [CrossRef] [PubMed]
- Yan, Z.; Qin, Z.; Fan, J.; Huang, Y.; Wang, Y.; Zhang, J.; Zhang, L.; Cao, Y. Gas Outburst Warning Method in Driving Faces: Enhanced Methodology through Optuna Optimization, Adaptive Normalization, and Transformer Framework. Sensors 2024, 24, 3150. [Google Scholar] [CrossRef] [PubMed]
- Xiao, P.; Chen, D. Photothermal Radiometry Data Analysis by Using Machine Learning. Sensors 2024, 24, 3015. [Google Scholar] [CrossRef] [PubMed]
Layer Name | Size | |
---|---|---|
Input | bs × N | |
Step-1 | ||
Attention_1 | Linear_A1_1 | bs × N × N |
Linear_A1_2 | bs × N × N | |
Linear_A1_3 | bs × N × N | |
Linear_1_1 | bs × N × M | |
Linear_1_2 | bs × M × N | |
Linear_1_3 | bs × N × N | |
Step-2 | ||
Attention_2 | Linear_A2_1 | bs × N × N |
Linear_A2_2 | bs × N × N | |
Linear_A2_3 | bs × N × N | |
Linear_2_1 | bs × N × M | |
Linear_2_2 | bs × M × N | |
Linear_2_3 | bs × N × 1 | |
Output | bs × 1 |
Input | Method | MAE (×103) | RMSE (×103) | MAPE (%) | R2 | PCC | CCC | Evar |
---|---|---|---|---|---|---|---|---|
four parameters | RF [35] | 6.49 | 8.71 | 1.18 | 0.8504 | 0.9233 | 0.9168 | 0.8516 |
XgbRegressor [36] | 5.31 | 7.40 | 0.97 | 0.8921 | 0.9460 | 0.9409 | 0.8936 | |
SVR [37] | 4.98 | 7.11 | 0.91 | 0.9003 | 0.9552 | 0.9491 | 0.9121 | |
GPR [18] | 4.86 | 6.77 | 0.89 | 0.9097 | 0.9556 | 0.9509 | 0.9114 | |
RNN [33] | 5.20 | 7.07 | 0.95 | 0.9015 | 0.9504 | 0.9477 | 0.9030 | |
LSTM [34] | 5.33 | 7.35 | 0.97 | 0.8934 | 0.9461 | 0.9432 | 0.8947 | |
MLP [19] | 5.05 | 6.91 | 0.92 | 0.9059 | 0.9536 | 0.9502 | 0.9081 | |
MSA-Net | 4.73 | 6.58 | 0.86 | 0.9147 | 0.9567 | 0.9549 | 0.9150 | |
eight parameters | RF [35] | 6.74 | 9.49 | 1.22 | 0.8225 | 0.9186 | 0.8891 | 0.8231 |
XgbRegressor [36] | 5.52 | 7.77 | 1.00 | 0.8809 | 0.9397 | 0.9339 | 0.8812 | |
SVR [37] | 5.10 | 7.04 | 0.93 | 0.9024 | 0.9532 | 0.9504 | 0.9078 | |
GPR [18] | 5.02 | 6.96 | 0.91 | 0.9046 | 0.9540 | 0.9471 | 0.9065 | |
RNN [33] | 4.90 | 6.63 | 0.89 | 0.9133 | 0.9571 | 0.9544 | 0.9156 | |
LSTM [34] | 4.92 | 6.96 | 0.90 | 0.9045 | 0.9532 | 0.9500 | 0.9075 | |
MLP [19] | 4.89 | 6.66 | 0.89 | 0.9125 | 0.9570 | 0.9540 | 0.9142 | |
MSA-Net | 4.55 | 6.20 | 0.83 | 0.9242 | 0.9618 | 0.9604 | 0.9176 |
Input | Method | Statistics | MAE (×103) | RMSE (×103) | MAPE (%) | R2 | PCC | CCC | Evar |
---|---|---|---|---|---|---|---|---|---|
four parameters | RNN [33] | Mean | 5.20 | 7.07 | 0.95 | 0.9015 | 0.9504 | 0.9477 | 0.9030 |
Std. | 0.15 | 0.09 | 0.03 | 0.0024 | 0.0011 | 0.0015 | 0.0021 | ||
LSTM [34] | Mean | 5.33 | 7.35 | 0.97 | 0.8934 | 0.9461 | 0.9432 | 0.8947 | |
Std. | 0.16 | 0.08 | 0.03 | 0.0024 | 0.0014 | 0.0017 | 0.0027 | ||
MLP [19] | Mean | 5.05 | 6.91 | 0.92 | 0.9059 | 0.9536 | 0.9502 | 0.9081 | |
Std. | 0.28 | 0.17 | 0.05 | 0.0047 | 0.0018 | 0.0029 | 0.0028 | ||
MSA-Net | Mean | 4.73 | 6.58 | 0.86 | 0.9147 | 0.9567 | 0.9549 | 0.9150 | |
Std. | 0.09 | 0.04 | 0.02 | 0.0010 | 0.0007 | 0.0010 | 0.0013 | ||
eight parameters | RNN [33] | Mean | 4.90 | 6.63 | 0.89 | 0.9133 | 0.9571 | 0.9544 | 0.9156 |
Std. | 0.22 | 0.06 | 0.04 | 0.0014 | 0.0010 | 0.0014 | 0.0021 | ||
LSTM [34] | Mean | 4.92 | 6.96 | 0.90 | 0.9045 | 0.9532 | 0.9500 | 0.9075 | |
Std. | 0.15 | 0.11 | 0.03 | 0.0031 | 0.0011 | 0.0020 | 0.0023 | ||
MLP [19] | Mean | 4.89 | 6.66 | 0.89 | 0.9125 | 0.9570 | 0.9540 | 0.9142 | |
Std. | 0.15 | 0.16 | 0.03 | 0.0042 | 0.0023 | 0.0037 | 0.0052 | ||
MSA-Net | Mean | 4.55 | 6.20 | 0.83 | 0.9242 | 0.9618 | 0.9604 | 0.9176 | |
Std. | 0.18 | 0.09 | 0.02 | 0.0021 | 0.0011 | 0.0012 | 0.0020 |
Method | Huber Loss | Attention Module × 1 | Attention Module × 2 | Attention Module × 3 | Skip-Connections | Input Eight Parameters | MAE (×103) |
---|---|---|---|---|---|---|---|
MLP | 5.05 | ||||||
A | √ | 4.96 | |||||
B | √ | √ | 4.91 | ||||
C | √ | √ | 4.87 | ||||
D | √ | √ | 5.21 | ||||
MSA-Net | √ | √ | √ | 4.73 | |||
MSA-Net | √ | √ | √ | √ | 4.55 |
Input | Method | Statistics | MAE (×103) | RMSE (×103) | MAPE (%) | R2 | PCC | CCC | Evar |
---|---|---|---|---|---|---|---|---|---|
four parameters | RNN [33] | Mean | 5.58 | 7.40 | 1.02 | 0.8779 | 0.9420 | 0.9303 | 0.8814 |
Std. | 0.17 | 0.19 | 0.03 | 0.0062 | 0.0028 | 0.0042 | 0.0055 | ||
LSTM [34] | Mean | 5.68 | 7.59 | 1.04 | 0.8713 | 0.9392 | 0.9273 | 0.8761 | |
Std. | 0.10 | 0.17 | 0.02 | 0.0058 | 0.0019 | 0.0047 | 0.0033 | ||
MLP [19] | Mean | 5.22 | 7.00 | 0.95 | 0.8904 | 0.9470 | 0.9410 | 0.8946 | |
Std. | 0.33 | 0.31 | 0.06 | 0.0096 | 0.0042 | .0038 | 0.0060 | ||
Transformer | Mean | 4.94 | 6.70 | 0.90 | 0.8999 | 0.9508 | 0.9458 | 0.9024 | |
Std. | 0.15 | 0.16 | 0.03 | 0.0049 | 0.0021 | 0.0038 | 0.0049 | ||
MSA-Net | Mean | 4.81 | 6.44 | 0.88 | 0.9075 | 0.9537 | 0.9501 | 0.9083 | |
Std. | 0.09 | 0.06 | 0.02 | 0.0018 | 0.0014 | 0.0013 | 0.0023 | ||
eight parameters | RNN [33] | Mean | 5.60 | 7.49 | 1.02 | 0.8746 | 0.9395 | 0.9290 | 0.8774 |
Std. | 0.23 | 0.27 | 0.04 | 0.0089 | 0.0054 | 0.0052 | 0.0091 | ||
LSTM [34] | Mean | 5.95 | 7.83 | 1.09 | 0.8631 | 0.9337 | 0.9217 | 0.8661 | |
Std. | 0.19 | 0.21 | 0.03 | 0.0074 | 0.0032 | 0.0064 | 0.0086 | ||
MLP [19] | Mean | 5.30 | 7.34 | 0.97 | 0.8796 | 0.9419 | 0.9323 | 0.8823 | |
Std. | 0.22 | 0.26 | 0.04 | 0.0086 | 0.0045 | 0.0063 | 0.0087 | ||
Transformer | Mean | 4.91 | 6.56 | 0.90 | 0.9039 | 0.9527 | 0.9481 | 0.9061 | |
Std. | 0.09 | 0.15 | 0.02 | 0.0043 | 0.0024 | 0.0028 | 0.0041 | ||
MSA-Net | Mean | 4.72 | 6.40 | 0.86 | 0.9085 | 0.9556 | 0.9505 | 0.9116 | |
Std. | 0.09 | 0.07 | 0.02 | 0.0019 | 0.0016 | 0.0013 | 0.0030 |
Input | Method | Statistics | MAE (×103) | RMSE (×103) | MAPE (%) | R2 | PCC | CCC | Evar |
---|---|---|---|---|---|---|---|---|---|
four parameters | RNN [33] | Mean | 5.03 | 6.97 | 0.90 | 0.9048 | 0.9535 | 0.9515 | 0.9075 |
Std. | 0.13 | 0.12 | 0.02 | 0.0034 | 0.0018 | 0.0027 | 0.0029 | ||
LSTM [34] | Mean | 5.07 | 6.97 | 0.91 | 0.9048 | 0.9527 | 0.9514 | 0.9057 | |
Std. | 0.17 | 0.17 | 0.03 | 0.0047 | 0.0022 | 0.0020 | 0.0048 | ||
MLP [19] | Mean | 4.95 | 6.98 | 0.89 | 0.9043 | 0.9534 | 0.9511 | 0.9063 | |
Std. | 0.34 | 0.36 | 0.06 | 0.0099 | 0.0043 | 0.0046 | 0.0090 | ||
Transformer | Mean | 4.59 | 6.51 | 0.82 | 0.9169 | 0.9597 | 0.9572 | 0.9208 | |
Std. | 0.27 | 0.30 | 0.05 | 0.0076 | 0.0023 | 0.0037 | 0.0043 | ||
MSA-Net | Mean | 4.31 | 6.19 | 0.78 | 0.9248 | 0.9622 | 0.9613 | 0.9256 | |
Std. | 0.09 | 0.07 | 0.02 | 0.0017 | 0.0007 | 0.0008 | 0.0016 | ||
eight parameters | RNN [33] | Mean | 4.96 | 6.72 | 0.89 | 0.9115 | 0.9562 | 0.9541 | 0.9121 |
Std. | 0.19 | 0.17 | 0.03 | 0.0045 | 0.0020 | 0.0027 | 0.0044 | ||
LSTM [34] | Mean | 5.50 | 7.21 | 0.99 | 0.8981 | 0.9495 | 0.9462 | 0.9001 | |
Std. | 0.13 | 0.12 | 0.02 | 0.0033 | 0.0019 | 0.0032 | 0.0035 | ||
MLP [19] | Mean | 5.30 | 7.08 | 0.95 | 0.9015 | 0.9547 | 0.9492 | 0.9078 | |
Std. | 0.28 | 0.30 | 0.05 | 0.0081 | 0.0043 | 0.0060 | 0.0072 | ||
Transformer | Mean | 4.91 | 6.66 | 0.88 | 0.9132 | 0.9568 | 0.9557 | 0.9143 | |
Std. | 0.23 | 0.16 | 0.04 | 0.0042 | 0.0027 | 0.0031 | 0.0040 | ||
MSA-Net | Mean | 4.34 | 6.20 | 0.78 | 0.9248 | 0.9622 | 0.9613 | 0.9253 | |
Std. | 0.07 | 0.08 | 0.01 | 0.0019 | 0.0011 | 0.0011 | 0.0020 |
Input | Method | Statistics | MAE (×103) | RMSE (×103) | MAPE (%) | R2 | PCC | CCC | Evar |
---|---|---|---|---|---|---|---|---|---|
four parameters | RNN [33] | Mean | 5.22 | 7.23 | 0.94 | 0.8990 | 0.9489 | 0.9470 | 0.9000 |
Std. | 0.09 | 0.09 | 0.02 | 0.0025 | 0.0013 | 0.0016 | 0.0026 | ||
LSTM [34] | Mean | 5.31 | 7.33 | 0.96 | 0.8963 | 0.9479 | 0.9438 | 0.8971 | |
Std. | 0.09 | 0.09 | 0.02 | 0.0025 | 0.0012 | 0.0024 | 0.0028 | ||
MLP [19] | Mean | 5.36 | 7.46 | 0.96 | 0.8921 | 0.9517 | 0.9423 | 0.9021 | |
Std. | 0.39 | 0.52 | 0.07 | 0.0150 | 0.0047 | 0.0104 | 0.0114 | ||
Transformer | Mean | 4.54 | 6.42 | 0.82 | 0.9203 | 0.9599 | 0.9577 | 0.9208 | |
Std. | 0.12 | 0.09 | 0.02 | 0.0021 | 0.0013 | 0.0012 | 0.0023 | ||
MSA-Net | Mean | 4.37 | 6.23 | 0.79 | 0.9251 | 0.9627 | 0.9605 | 0.9259 | |
Std. | 0.07 | 0.06 | 0.01 | 0.0015 | 0.0009 | 0.0013 | 0.0018 | ||
eight parameters | RNN [33] | Mean | 5.07 | 6.94 | 0.91 | 0.9070 | 0.9545 | 0.9502 | 0.9096 |
Std. | 0.21 | 0.19 | 0.04 | 0.0051 | 0.0020 | 0.0038 | 0.0047 | ||
LSTM [34] | Mean | 5.63 | 7.49 | 1.02 | 0.8916 | 0.9460 | 0.9413 | 0.8939 | |
Std. | 0.15 | 0.17 | 0.03 | 0.0050 | 0.0023 | 0.0026 | 0.0040 | ||
MLP [19] | Mean | 5.48 | 7.45 | 0.99 | 0.8922 | 0.9507 | 0.9409 | 0.8998 | |
Std. | 0.47 | 0.54 | 0.08 | 0.0161 | 0.0056 | 0.0107 | 0.0134 | ||
Transformer | Mean | 4.67 | 6.46 | 0.84 | 0.9195 | 0.9605 | 0.9568 | 0.9214 | |
Std. | 0.08 | 0.11 | 0.02 | 0.0028 | 0.0019 | 0.0022 | 0.0045 | ||
MSA-Net | Mean | 4.22 | 5.96 | 0.76 | 0.9313 | 0.9656 | 0.9642 | 0.9320 | |
Std. | 0.07 | 0.06 | 0.01 | 0.0014 | 0.0007 | 0.0010 | 0.0014 |
Input | Method | Statistics | MAE (×103) | RMSE (×103) | MAPE (%) | R2 | PCC | CCC | Evar |
---|---|---|---|---|---|---|---|---|---|
four parameters | RNN [33] | Mean | 5.62 | 7.56 | 1.02 | 0.8711 | 0.9354 | 0.9297 | 0.8737 |
Std. | 0.11 | 0.09 | 0.02 | 0.0029 | 0.0023 | 0.0027 | 0.0043 | ||
LSTM [34] | Mean | 5.71 | 7.60 | 1.04 | 0.8696 | 0.9331 | 0.9299 | 0.8703 | |
Std. | 0.11 | 0.10 | 0.02 | 0.0036 | 0.0018 | 0.0025 | 0.0033 | ||
MLP [19] | Mean | 5.28 | 7.10 | 0.96 | 0.8862 | 0.9429 | 0.9390 | 0.8877 | |
Std. | 0.18 | 0.21 | 0.03 | 0.0067 | 0.0032 | 0.0049 | 0.0067 | ||
Transformer | Mean | 5.13 | 6.90 | 0.93 | 0.8927 | 0.9467 | 0.9414 | 0.8944 | |
Std. | 0.21 | 0.09 | 0.04 | 0.0028 | 0.0017 | 0.0019 | 0.0031 | ||
MSA-Net | Mean | 4.98 | 6.69 | 0.91 | 0.8990 | 0.9486 | 0.9462 | 0.8995 | |
Std. | 0.10 | 0.05 | 0.02 | 0.0015 | 0.0007 | 0.0015 | 0.0015 | ||
eight parameters | RNN [33] | Mean | 5.29 | 6.98 | 0.96 | 0.8901 | 0.9448 | 0.9401 | 0.8914 |
Std. | 0.12 | 0.12 | 0.02 | 0.0037 | 0.0019 | 0.0022 | 0.0035 | ||
LSTM [34] | Mean | 5.73 | 7.43 | 1.04 | 0.8754 | 0.9378 | 0.9310 | 0.8772 | |
Std. | 0.20 | 0.22 | 0.04 | 0.0075 | 0.0028 | 0.0056 | 0.0065 | ||
MLP [19] | Mean | 5.25 | 7.04 | 0.95 | 0.8882 | 0.9447 | 0.9403 | 0.8912 | |
Std. | 0.18 | 0.15 | 0.03 | 0.0048 | 0.0020 | 0.0040 | 0.0043 | ||
Transformer | Mean | 5.16 | 6.94 | 0.94 | 0.8915 | 0.9448 | 0.9416 | 0.8920 | |
Std. | 0.12 | 0.08 | 0.02 | 0.0026 | 0.0016 | 0.0019 | 0.0027 | ||
MSA-Net | Mean | 4.80 | 6.43 | 0.87 | 0.9067 | 0.9528 | 0.9508 | 0.9077 | |
Std. | 0.07 | 0.06 | 0.01 | 0.0017 | 0.0009 | 0.0009 | 0.0017 |
Train Sets | Test Sets | Inputs | MAE | RMSE (×103) | MAPE (%) |
---|---|---|---|---|---|
odd months | even months | Mad, Aad, Vad, Qgr,ad | 4.81 | 6.44 | 0.88 |
Mad, Aad, Vad, Qgr,ad, St,ad | 4.75 | 6.42 | 0.88 | ||
Mt, Mad, Aad, Vad, FCad, Had, St,ad, NCV | 4.72 | 6.40 | 0.86 | ||
even months | odd months | Mad, Aad, Vad, Qgr,ad | 4.31 | 6.19 | 0.78 |
Mad, Aad, Vad, Qgr,ad, St,ad | 4.26 | 6.10 | 0.77 | ||
Mt, Mad, Aad, Vad, FCad, Had, St,ad, NCV | 4.34 | 6.20 | 0.78 | ||
odd days | even days | Mad, Aad, Vad, Qgr,ad | 4.37 | 6.23 | 0.79 |
Mad, Aad, Vad, Qgr,ad, St,ad | 4.30 | 6.15 | 0.78 | ||
Mt, Mad, Aad, Vad, FCad, Had, St,ad, NCV | 4.22 | 5.96 | 0.76 | ||
even days | odd days | Mad, Aad, Vad, Qgr,ad | 4.98 | 6.69 | 0.91 |
Mad, Aad, Vad, Qgr,ad, St,ad | 4.85 | 6.60 | 0.89 | ||
Mt, Mad, Aad, Vad, FCad, Had, St,ad, NCV | 4.80 | 6.43 | 0.87 |
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2024 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Wang, Y.; Liu, Z.; Chen, F.; Xiong, X. MSA-Net: A Precise and Robust Model for Predicting the Carbon Content on an As-Received Basis of Coal. Sensors 2024, 24, 4607. https://doi.org/10.3390/s24144607
Wang Y, Liu Z, Chen F, Xiong X. MSA-Net: A Precise and Robust Model for Predicting the Carbon Content on an As-Received Basis of Coal. Sensors. 2024; 24(14):4607. https://doi.org/10.3390/s24144607
Chicago/Turabian StyleWang, Yinchu, Zilong Liu, Feng Chen, and Xingchuang Xiong. 2024. "MSA-Net: A Precise and Robust Model for Predicting the Carbon Content on an As-Received Basis of Coal" Sensors 24, no. 14: 4607. https://doi.org/10.3390/s24144607