A Daily Air Pollutant Concentration Prediction Framework Combining Successive Variational Mode Decomposition and Bidirectional Long Short-Term Memory Network
Abstract
:1. Introduction
- (1)
- During the unsupervised phase, SVMD is employed for feature learning without the need for labeled data. It segments the sequence into multiple intrinsic mode functions (IMF), which capture distinct frequency and magnitude characteristics, thereby representing diverse features of the data.
- (2)
- In the supervision stage, the BO-BiLSTM neural network is superior to other linear networks in extracting time-dependent features efficiently.
- (3)
- Our article created a novel hybrid model based on SVMD-BO-BiLSTM approaches. The proposed hybrid model’s ability to forecast PM2.5 and PM10 concentrations in Tianshui, Gansu and Wuhan, Hubei is examined using datasets from these locations.
- (4)
- We present a scientific and reasonable model evaluation system involving multiple experiments, seven model performance indicators, four air pollutant datasets, and a systematic evaluation of the proposed mixed model using multiple comparison model tests and stability tests. Furthermore, the superior performance of the proposed model indicates that the proposed hybrid model not only provides a new option for proposing air quality regulatory requirements to reduce air pollution, but will also help to guide people’s daily activities, and protect them from harmful air pollutants.
2. Preliminaries
2.1. SVMD Method
- Each mode should exhibit compactness around its central frequency. Therefore, the mode is minimized to meet the following criteria:
- The spectral overlap between the residual signal and the mode should be minimized, meaning that the energy of the residual mode signal is minimized in the frequency band wherein the desired mode is located. To ensure stable implementation of this constraint, an appropriate filter is selected with the following frequency response:
- By minimizing the constraints of and , there may be a challenge in effectively distinguishing the mode and the mode. Therefore, based on the idea of constraint establishment, the frequency response of the filter used is
- During the signal decomposition, the following constraints are established to ensure that the signal can be completely reconstructed.
2.2. BiLSTM Model
2.3. Bayesian Optimization Algorithm
3. SVMD-BO-BiLSTM Prediction Model
3.1. BO-BiLSTM
- (1)
- Obtain the time collection facts to be predicted, and divide the coaching set and test set according to the proportion.
- (2)
- Take the number of neurons in the BiLSTM network, the learning rate of the optimizer, and the wide variety of hidden layers as the optimization object, and initialize the BO algorithm.
- (3)
- Calculate the contemporary function distribution randomly.
- (4)
- Adjust the modern-day function distribution according to the method selected by means of the determination function.
- (5)
- Determine whether the termination prerequisites are met. If yes, the highest quality hyperparameter value is returned. Otherwise, return to Step 4.
- (6)
- Construct the BiLSTM network model with greatest hyperparameters.
3.2. PM2.5 and PM10 Concentration Prediction Framework
4. Experiment
4.1. Data Source
4.2. Parameter Setting
4.3. Comparison of the Proposed Predictor with Other Prediction Methods
4.3.1. Evaluation Indexes
4.3.2. Prediction Result of the Proposed Model
4.3.3. Comparison of Forecasting Results
5. Discussion
6. Conclusions
- (1)
- The SVMD-BO-BiLSTM model considers the impact of implicit time-invariant factors on the prediction results and requires minimal auxiliary data. As a result, it simplifies the model structure, reduces computational complexity, and achieves exceptional accuracy and stability in air quality prediction.
- (2)
- Air quality time series data are processed as signal data using SVMD, allowing unsupervised feature learning to extract frequency features. This method significantly enhances the accuracy of short-term trend prediction, particularly for sudden and abrupt changes in the data. By incorporating unsupervised feature learning, the prediction capability of the supervised LSTM model is enhanced, particularly for high-frequency and irregular fluctuations observed in time series data. The BiLSTM model is more suitable than time series prediction models, such as the GRU, LSTM, and KELM models, for component PM2.5 and PM10 concentration prediction, based on SVMD method, in the experimental case of this study. The new hybrid model has good spatio-temporal generalization and robustness for PM2.5 and PM10 concentration prediction in different moments and regions. It can be used as an effective tool for predicting trends in concentration changes at monitoring points, which has good practical significance.
- (3)
- Several evaluation index systems were constructed for model evaluation. The SVMD-BO-BiLSTM model demonstrates remarkable accuracy and stability in predicting PM2.5 and PM10 concentrations across various time scales.
Author Contributions
Funding
Institutional Review Board Statement
Informed Consent Statement
Data Availability Statement
Conflicts of Interest
References
- Duan, J.; Li, Y.-L.; Li, S.; Yang, Y.; Li, F.; Li, Y.; Wang, J.; Dug, P.-Z.; Wu, J.; Wang, W.; et al. Association of in China. Am. J. Kidney Dis. 2022, 22, 638–647. [Google Scholar] [CrossRef]
- Shen, F.; Ge, X.; Hu, J.; Nie, D.; Tian, L.; Chen, M. Air pollution characteristics and health risks in Henan Province, China. Environ. Res. 2017, 156, 625–634. [Google Scholar] [CrossRef]
- Zhang, H.; Wang, Y.; Hu, J.; Ying, Q.; Hu, X.M. Relationships between meteorological parameters and criteria air pollutants in three megacities in China. Environ. Res. 2015, 140, 242–254. [Google Scholar] [CrossRef]
- Biancofiore, F.; Busilacchio, M.; Verdecchia, M.; Tomassetti, B.; Aruffo, E.; Bianco, S.; Di Tommaso, S.; Colangeli, C.; Rosatelli, G.; Di Carlo, P. Recursive neural network model for analysis and forecast of PM10 and PM2.5. Atmos. Pollut. Res. 2017, 8, 652–659. [Google Scholar] [CrossRef]
- Xu, Y.; Du, P.; Wang, J. Research and application of a hybrid model based on dynamic fuzzy synthetic evaluation for establishing air quality forecasting and early warning system: A case study in China. Environ. Pollut. 2017, 223, 435–448. [Google Scholar] [CrossRef]
- Sun, W.; Zhang, H.; Palazoglu, A.; Singh, A.; Zhang, W.; Liu, S. Prediction of 24-hour-average PM2.5 concentrations using a hidden Markov model with different emission distributions in Northern California. Sci. Total Environ. 2013, 443, 93–103. [Google Scholar] [CrossRef]
- Sun, W.; Sun, J. Daily PM2.5 concentration prediction based on principal component analysis and LSSVM optimized by cuckoo search algorithm. J. Environ. Manag. 2017, 188, 144–152. [Google Scholar] [CrossRef] [PubMed]
- Zhang, L.; Lin, J.; Qiu, R.; Hu, X.; Zhang, H.; Chen, Q.; Tan, H.; Lin, D.; Wang, J. Trend analysis and forecast of PM2.5 in Fuzhou, China using the ARIMA model. Ecol. Indic. 2018, 95, 702–710. [Google Scholar] [CrossRef]
- Tien Bui, D.; Moayedi, H.; Gör, M.; Jaafari, A.; Foong, L.K. Predicting slope stability failure through machine learning paradigms. ISPRS Int. J. Geo-Inf. 2019, 8, 395. [Google Scholar] [CrossRef] [Green Version]
- Goudarzi, G.; Birgani, Y.T.; Assarehzadegan, M.A.; Neisi, A.; Dastoorpoor, M.; Sorooshian, A.; Yazdani, M. Prediction of airborne pollen concentrations by artificial neural network and their relationship with meteorological parameters and air pollutants. J. Environ. Health Sci. Eng. 2022, 20, 251–264. [Google Scholar] [CrossRef] [PubMed]
- Goudarzi, G.; Hopke, P.K.; Yazdani, M. Forecasting PM2.5 concentration using artificial neural network and its health effects in Ahvaz, Iran. Chemosphere 2021, 283, 131285. [Google Scholar] [CrossRef] [PubMed]
- Shang, Z.; Deng, T.; He, J.; Duan, X. A novel model for hourly PM2.5 concentration prediction based on CART and EELM. Sci. Total Environ. 2019, 651, 3043–3052. [Google Scholar] [CrossRef] [PubMed]
- Qi, Y.; Li, Q.; Karimian, H.; Liu, D. A hybrid model for spatiotemporal forecasting of PM2.5 based on graph convolutional neural network and long short-term memory. Sci. Total Environ. 2019, 664, 1–10. [Google Scholar] [CrossRef]
- Pak, U.; Ma, J.; Ryu, U.; Ryom, K.; Juhyok, U.; Pak, K.; Pak, C. Deep learning-based PM2.5 prediction considering the spatiotemporal correlations: A case study of Beijing, China. Sci. Total Environ. 2020, 699, 133561. [Google Scholar] [CrossRef] [PubMed]
- Wang, J.; Yang, W.; Du, P.; Li, Y. Research and application of a hybrid forecasting framework based on multi-objective optimization for electrical power system. Energy 2018, 148, 59–78. [Google Scholar] [CrossRef]
- Wang, J.; Liu, F.; Song, Y.; Zhao, J. A novel model: Dynamic choice artificial neural network (DCANN) for an electricity price forecasting system. Appl. Soft Comput. 2016, 48, 281–297. [Google Scholar] [CrossRef]
- Zhang, W.; Qu, Z.; Zhang, K.; Mao, W.; Ma, Y.; Fan, X. A combined model based on CEEMDAN and modified flower pollination algorithm for wind speed forecasting. Energy Convers. Manag. 2017, 136, 439–451. [Google Scholar] [CrossRef]
- Wang, J.; Hu, J. A robust combination approach for short-term wind speed forecasting and analysis–Combination of the ARIMA (Autoregressive Integrated Moving Average), ELM (Extreme Learning Machine), SVM (Support Vector Machine) and LSSVM (Least Square SVM) forecasts using a GPR (Gaussian Process Regression) model. Energy 2015, 93, 41–56. [Google Scholar] [CrossRef]
- Wang, J.; Du, P.; Hao, Y.; Ma, X.; Niu, T.; Yang, W. An innovative hybrid model based on outlier detection and correction algorithm and heuristic intelligent optimization algorithm for daily air quality index forecasting. J. Environ. Manag. 2020, 255, 109855. [Google Scholar] [CrossRef]
- Qiao, W.; Yang, Z.; Kang, Z.; Pan, Z. Short-term natural gas consumption prediction based on Volterra adaptive filter and improved whale optimization algorithm. Eng. Appl. Artif. Intell. 2020, 87, 103323. [Google Scholar] [CrossRef]
- Wang, J.; Yang, W.; Du, P.; Niu, T. Outlier-robust hybrid electricity price forecasting model for electricity market management. J. Clean. Prod. 2020, 249, 119318. [Google Scholar] [CrossRef]
- Hao, Y.; Tian, C.; Wu, C. Modelling of carbon price in two real carbon trading markets. J. Clean. Prod. 2020, 244, 118556. [Google Scholar] [CrossRef]
- Qiao, W.; Lu, H.; Zhou, G.; Azimi, M.; Yang, Q.; Tian, W. A hybrid algorithm for carbon dioxide emissions forecasting based on improved lion swarm optimizer. J. Clean. Prod. 2020, 244, 118612. [Google Scholar] [CrossRef]
- Jiang, P.; Liu, Z. Variable weights combined model based on multi-objective optimization for short-term wind speed forecasting. Appl. Soft Comput. 2019, 82, 105587. [Google Scholar] [CrossRef]
- Niu, M.; Gan, K.; Sun, S.; Li, F. Application of decomposition-ensemble learning paradigm with phase space reconstruction for day-ahead PM2.5 concentration forecasting. J. Environ. Manag. 2017, 196, 110–118. [Google Scholar] [CrossRef] [PubMed]
- Niu, M.; Wang, Y.; Sun, S.; Li, Y. A novel hybrid decomposition-and-ensemble model based on CEEMD and GWO for short-term PM2.5 concentration forecasting. Atmos. Environ. 2016, 134, 168–180. [Google Scholar] [CrossRef]
- Zhang, H.; Xu, H.R.; Peng, G.; Qian, Y.D.; Zhang, X.X.; Yang, G.L.; Shen, C.; Li, Z.; Yang, J.W.; Wang, Z.Q.; et al. A Prediction model of relativistic electrons at geostationary orbit using the EMD-LSTM network and geomagnetic indices. Space Weather 2022, 20, e2022SW003126. [Google Scholar] [CrossRef]
- Zhang, X.; Chen, H.; Zhu, G.; Zhao, D.; Duan, B. A new groundwater depth prediction model based on EMD-LSTM. Water Supply 2022, 22, 5974–5988. [Google Scholar] [CrossRef]
- Zhang, Y.; Li, C.; Jiang, Y.; Sun, L.; Zhao, R.; Yan, K.; Wang, W. Accurate prediction of water quality in urban drainage network with integrated EMD-LSTM model. J. Clean. Prod. 2022, 354, 131724. [Google Scholar] [CrossRef]
- Ausati, S.; Amanollahi, J. Assessing the accuracy of ANFIS, EEMD-GRNN, PCR, and MLR models in predicting PM2.5. Atmos. Environ. 2016, 142, 465–474. [Google Scholar] [CrossRef]
- Sun, W.; Xu, Z. A hybrid Daily PM2.5 concentration prediction model based on secondary decomposition algorithm, mode recombination technique and deep learning. Stoch. Environ. Res. Risk Assess. 2022, 36, 1143–1162. [Google Scholar] [CrossRef]
- Li, F.; Ma, G.; Chen, S.; Huang, W. An ensemble modeling approach to forecast daily reservoir inflow using bidirectional long-and short-term memory (Bi-LSTM), variational mode decomposition (VMD), and energy entropy method. Water Resour. Manag. 2021, 35, 2941–2963. [Google Scholar] [CrossRef]
- Zhang, G.; Liu, H.; Zhang, J.; Yan, Y.; Zhang, L.; Wu, C.; Hua, X.; Wang, Y. Wind power prediction based on variational mode decomposition multi-frequency combinations. J. Mod. Power Syst. Clean Energy 2019, 7, 281–288. [Google Scholar] [CrossRef] [Green Version]
- Wang, W.; Tong, M.; Yu, M. Blood glucose prediction with VMD and LSTM optimized by improved particle swarm optimization. IEEE Access 2020, 8, 217908–217916. [Google Scholar] [CrossRef]
- Guo, H.; Guo, Y.; Zhang, W.; He, X.; Qu, Z. Research on a Novel Hybrid Decomposition–Ensemble Learning Paradigm Based on VMD and IWOA for PM2.5 Forecasting. Int. J. Environ. Res. Public Health 2021, 18, 1024. [Google Scholar] [CrossRef]
- Ding, G.; Wang, W.; Zhu, T. Remaining Useful Life Prediction for Lithium-Ion Batteries Based on CS-VMD and GRU. IEEE Access 2022, 10, 89402–89413. [Google Scholar] [CrossRef]
- Yang, H.; Liu, Z.; Li, G. A new hybrid optimization prediction model for PM2.5 concentration considering other air pollutants and meteorological conditions. Chemosphere 2022, 307, 135798. [Google Scholar] [CrossRef]
- Tuerxun, W.; Xu, C.; Guo, H.; Guo, L.; Zeng, N.; Cheng, Z. An ultra-short-term wind speed prediction model using LSTM based on modified tuna swarm optimization and successive variational mode decomposition. Energy Sci. 2022, 10, 3001–3022. [Google Scholar] [CrossRef]
- Nazari, M.; Sakhaei, S.M. Successive variational mode decomposition. Signal Process. 2020, 174, 107610. [Google Scholar] [CrossRef]
- Thoppil, N.M.; Vasu, V.; Rao, C.S.P. Bayesian optimization LSTM/bi-LSTM network with self-optimized structure and hyperparameters for remaining useful life estimation of lathe spindle unit. J. Comput. Inf. Sci. Eng. 2022, 22, 021012. [Google Scholar] [CrossRef]
- Yan, W.; Wang, J.; Cheng, J.; Wan, Z.; Xing, K.; Gao, K. Long Short-Term Memory Networks and Bayesian Optimization for Predicting the Time-Weighted Average Pressure of Shield Supporting Cycles. Geofluids 2021, 2021, 8895844. [Google Scholar] [CrossRef]
- Lu, J. Temporal and Spatial Characteristics of Air Quality and Its Influencing Factors in the Middle and Lower Reaches of the Yangtze River; Wuhan University: Wuhan, China, 2020. [Google Scholar] [CrossRef]
- Teng, M.; Li, S.; Xing, J.; Song, G.; Yang, J.; Dong, J.; Zeng, X.; Qin, Y. 24-Hour prediction of PM2.5 concentrations by combining empirical mode decomposition and bidirectional long short-term memory neural network. Sci. Total Environ. 2022, 821, 153276. [Google Scholar] [CrossRef] [PubMed]
- Ding, P.; Liu, X.; Li, H.; Huang, Z.; Zhang, K.; Shao, L.; Abedinia, O. Useful life prediction based on wavelet packet decomposition and two-dimensional convolutional neural network for lithium-ion batteries. Renew. Sustain. Energy Rev. 2021, 148, 111287. [Google Scholar] [CrossRef]
- Du, P.; Wang, J.; Hao, Y.; Niu, T.; Yang, W. A novel hybrid model based on multi-objective Harris hawks optimization algorithm for daily PM2.5 and PM10 forecasting. Appl. Soft Comput. 2020, 96, 106620. [Google Scholar] [CrossRef]
Number | Minimum (μg/m3) | Maximum (μg/m3) | Average (μg/m3) | Standard Deviation (μg/m3) | ||
---|---|---|---|---|---|---|
PM2.5 | Tianshui | 2057 | 4 | 148 | 29.80 | 20.67 |
Wuhan | 2057 | 4 | 216 | 42.31 | 28.80 | |
PM10 | Tianshui | 2057 | 6 | 395 | 62.50 | 46.27 |
Wuhan | 2057 | 3 | 289 | 68.14 | 41.28 |
Area | Models | MAE (μg/m3) | RMSE (μg/m3) | MAPE (%) | R2 | RA | TIC |
---|---|---|---|---|---|---|---|
Tianshui | EMD-BiLSTM | 1.0694 | 1.9381 | 6.11 | 0.9892 | 0.9669 | 0.1096 |
CEEMDAN-BiLSTM | 1.6332 | 2.4282 | 9.59 | 0.983 | 0.9519 | 0.1123 | |
VMD-BiLSTM | 0.5551 | 1.179 | 2.79 | 0.9953 | 0.9806 | 0.0602 | |
SVMD-BiLSTM | 0.4131 | 1.4278 | 1.29 | 0.9927 | 0.9800 | 0.0237 | |
SVMD-BO-BiLSTM | 0.4414 | 1.0951 | 2.08 | 0.9957 | 0.9841 | 0.0183 | |
Wuhan | EMD-BiLSTM | 5.0676 | 5.4234 | 22.47 | 0.9506 | 0.9024 | 0.1309 |
CEEMDAN-BiLSTM | 1.3756 | 2.2511 | 5.14 | 0.9915 | 0.9663 | 0.1240 | |
VMD-BiLSTM | 0.6975 | 1.9459 | 2.09 | 0.9924 | 0.9806 | 0.0747 | |
SVMD-BiLSTM | 0.4551 | 1.5642 | 1.28 | 0.9945 | 0.9864 | 0.0259 | |
SVMD-BO-BiLSTM | 0.3846 | 0.9689 | 1.05 | 0.9979 | 0.9878 | 0.0178 |
Area | Models | MAE (μg/m3) | RMSE (μg/m3) | MAPE (%) | R2 | RA | TIC |
---|---|---|---|---|---|---|---|
Tianshui | EMD-BiLSTM | 8.4403 | 10.8604 | 19.92 | 0.9258 | 0.8760 | 0.1587 |
CEEMDAN-BiLSTM | 5.4991 | 9.1445 | 12.85 | 0.9474 | 0.9203 | 0.1616 | |
VMD-BiLSTM | 1.4000 | 2.8760 | 4.15 | 0.9918 | 0.9762 | 0.0931 | |
SVMD-BiLSTM | 0.8818 | 2.7702 | 7.88 | 0.9924 | 0.9814 | 0.0217 | |
SVMD-BO-BiLSTM | 0.8360 | 2.6950 | 1.45 | 0.9936 | 0.9837 | 0.0227 | |
Wuhan | EMD-BiLSTM | 4.7955 | 5.3231 | 11.21 | 0.9735 | 0.9333 | 0.1262 |
CEEMDAN-BiLSTM | 2.2813 | 3.3890 | 5.18 | 0.9893 | 0.9666 | 0.1266 | |
VMD-BiLSTM | 0.9735 | 2.6349 | 1.55 | 0.9898 | 0.9813 | 0.0738 | |
SVMD-BiLSTM | 0.7912 | 1.6438 | 1.47 | 0.9967 | 0.9861 | 0.0209 | |
SVMD-BO-BiLSTM | 0.4677 | 1.2177 | 0.82 | 0.9978 | 0.9915 | 0.0096 |
PM2.5 | Tianshui | Wuhan | ||||||||
---|---|---|---|---|---|---|---|---|---|---|
MAE (μg/m3) | RMSE (μg/m3) | MAPE (%) | R2 | RA | MAE (μg/m3) | RMSE (μg/m3) | MAPE (%) | R2 | RA | |
SVMD-KELM | 2.5244 | 3.9106 | 10.15 | 0.9449 | 0.8996 | 4.3520 | 6.2202 | 14.56 | 0.9125 | 0.8803 |
SVMD-GRU | 0.4822 | 1.7069 | 1.89 | 0.9895 | 0.9786 | 0.6286 | 1.5367 | 2.43 | 0.9947 | 0.9851 |
SVMD-BiGRU | 0.7049 | 2.5878 | 2.22 | 0.9759 | 0.9646 | 1.3756 | 2.2511 | 5.14 | 0.9915 | 0.9663 |
SVMD-LSTM | 1.6333 | 2.4282 | 9.60 | 0.9831 | 0.9520 | 0.7250 | 2.0995 | 1.91 | 0.9900 | 0.9771 |
SVMD-BiLSTM | 0.4131 | 1.4278 | 1.29 | 0.9927 | 0.9800 | 0.4551 | 1.5642 | 1.28 | 0.9945 | 0.9864 |
Proposed Method | 0.4414 | 1.0951 | 2.08 | 0.9957 | 0.9841 | 0.3846 | 0.9689 | 1.05 | 0.9979 | 0.9878 |
PM10 | Tianshui | Wuhan | ||||||||
---|---|---|---|---|---|---|---|---|---|---|
MAE (μg/m3) | RMSE (μg/m3) | MAPE (%) | R2 | RA | MAE (μg/m3) | RMSE (μg/m3) | MAPE (%) | R2 | RA | |
SVMD-KELM | 6.7510 | 10.8193 | 34.28 | 0.8844 | 0.8793 | 6.5128 | 8.8460 | 12.08 | 0.8847 | 0.8896 |
SVMD-GRU | 1.2227 | 3.0350 | 3.83 | 0.9909 | 0.9775 | 0.9672 | 1.6762 | 1.76 | 0.9959 | 0.9836 |
SVMD-BiGRU | 0.9777 | 3.1997 | 6.01 | 0.9899 | 0.9815 | 1.0565 | 3.0781 | 1.69 | 0.9860 | 0.9799 |
SVMD-LSTM | 2.0968 | 4.1838 | 3.51 | 0.9845 | 0.9590 | 0.9735 | 2.6349 | 1.55 | 0.9898 | 0.9813 |
SVMD-BiLSTM | 0.8818 | 2.7702 | 7.88 | 0.9924 | 0.9814 | 0.7912 | 1.6438 | 1.47 | 0.9967 | 0.9861 |
Proposed Method | 0.8360 | 2.6950 | 1.45 | 0.9936 | 0.9837 | 0.4677 | 1.2177 | 0.82 | 0.9978 | 0.9915 |
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. |
© 2023 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
Huang, Z.; Li, L.; Ding, G. A Daily Air Pollutant Concentration Prediction Framework Combining Successive Variational Mode Decomposition and Bidirectional Long Short-Term Memory Network. Sustainability 2023, 15, 10660. https://doi.org/10.3390/su151310660
Huang Z, Li L, Ding G. A Daily Air Pollutant Concentration Prediction Framework Combining Successive Variational Mode Decomposition and Bidirectional Long Short-Term Memory Network. Sustainability. 2023; 15(13):10660. https://doi.org/10.3390/su151310660
Chicago/Turabian StyleHuang, Zhong, Linna Li, and Guorong Ding. 2023. "A Daily Air Pollutant Concentration Prediction Framework Combining Successive Variational Mode Decomposition and Bidirectional Long Short-Term Memory Network" Sustainability 15, no. 13: 10660. https://doi.org/10.3390/su151310660