A novel deep learning framework with variational auto-encoder for indoor air quality prediction

Highlights

● PLS-VAER is proposed for modeling of PM_2.5 concentration.

● Data are decomposed by PLS to capture nonlinear feature.

● VAER can improve the predictive performance by variational inference.

● The proposed model provides a novel method for monitoring indoor air quality.

Abstract

Exposure to poor indoor air conditions poses significant risks to human health, increasing morbidity and mortality rates. Soft measurement modeling is suitable for stable and accurate monitoring of air pollutants and improving air quality. Based on partial least squares (PLS), we propose an indoor air quality prediction model that utilizes variational auto-encoder regression (VAER) algorithm. To reduce the negative effects of noise, latent variables in the original data are extracted by PLS in the first step. Then, the extracted variables are used as inputs to VAER, which improve the accuracy and robustness of the model. Through comparative analysis with traditional methods, we demonstrate the superior performance of our PLS-VAER model, which exhibits improved prediction performance and stability. The root mean square error (RMSE) of PLS-VAER is reduced by 14.71%, 26.47%, and 12.50% compared to single VAER, PLS-SVR, and PLS-ANN, respectively. Additionally, the coefficient of determination (R²) of PLS-VAER improves by 13.70%, 30.09%, and 11.25% compared to single VAER, PLS-SVR, and PLS-ANN, respectively. This research offers an innovative and environmentally-friendly approach to monitor and improve indoor air quality.

Graphical abstract

Keywords

Indoor air quality / PM_2.5 concentration / Variational auto-encoder / Latent variable / Soft measurement modeling

References

[1] Aljunaid M , Tao Y , Shi H . (2021). A novel mutual information and partial least squares approach for quality-related and quality-unrelated fault detection. Processes (Basel, Switzerland), 9(1): 166

[2] Alsenan S A , Al-Turaiki I M , Hafez A M . (2020). Feature extraction methods in quantitative structure activity relationship modeling: a comparative study. IEEE Access: Practical Innovations, Open Solutions, 8: 78737–78752

[3] Ángel de Miguel M , Armingol J M , García F . (2022). Vehicles trajectory prediction using recurrent VAE network. IEEE Access: Practical Innovations, Open Solutions, 10: 32742–32749

[4] Apsemidis A , Psarakis S , Moguerza J M . (2020). A review of machine learning kernel methods in statistical process monitoring. Computers & Industrial Engineering, 142: 106376

[5] Challoner A , Pilla F , Gill L . (2015). Prediction of indoor air exposure from outdoor air quality using an artificial neural network model for inner city commercial buildings. International Journal of Environmental Research and Public Health, 12(12): 15233–15253

[6] Chen R Q , Shi G H , Zhao W L , Liang C H . (2021). A joint model for IT operation series prediction and anomaly detection. Neurocomputing, 448: 130–139

[7] Chen Y Y , Sung F C , Chen M L , Mao I F , Lu C Y . (2016a). Indoor air quality in the metro system in north Taiwan, China. International Journal of Environmental Research and Public Health, 13(12): 1200

[8] Chen Z , Ding S X , Zhang K , Li Z , Hu Z . (2016b). Canonical correlation analysis-based fault detection methods with application to alumina evaporation process. Control Engineering Practice, 46: 51–58

[9] Chen Z , Zhang K , Ding S X , Shardt Y A W , Hu Z . (2016c). Improved canonical correlation analysis-based fault detection methods for industrial processes. Journal of Process Control, 41: 26–34

[10] Correia C , Martins V , Cunha-Lopes I , Faria T , Diapouli E , Eleftheriadis K , Almeida S M . (2020). Particle exposure and inhaled dose while commuting in Lisbon. Environmental Pollution, 257: 113547

[11] Diao M , Holloway T , Choi S , O’Neill S M , Al-Hamdan M Z , Van Donkelaar A , Martin R V , Jin X , Fiore A M , Henze D K . . (2019). Methods, availability, and applications of PM_2.5 exposure estimates derived from ground measurements, satellite, and atmospheric models. Journal of the Air & Waste Management Association, 69(12): 1391–1414

[12] Feng S , Gao D , Liao F , Zhou F , Wang X . (2016). The health effects of ambient PM_2.5 and potential mechanisms. Ecotoxicology and Environmental Safety, 128: 67–74

[13] Han K , Wen H , Shi J , Lu K H , Zhang Y , Fu D , Liu Z . (2019). Variational autoencoder: an unsupervised model for encoding and decoding fMRI activity in visual cortex. NeuroImage, 198: 125–136

[14] Hong Y , Hwang U , Yoo J , Yoon S . (2019). How generative adversarial networks and their variants work: an overview. ACM Computing Surveys, 52(1): 3301282

[15] Ji W , Liu C , Liu Z , Wang C , Li X . (2021). Concentration, composition, and exposure contributions of fine particulate matter on subway concourses in China. Environmental Pollution, 275: 116627

[16] Jin X B , Gong W T , Kong J L , Bai Y T , Su T L . (2022). PFVAE: a planar flow-based variational auto-encoder prediction model for time series data. Mathematics, 10(4): 610

[17] Kim M H , Kim Y S , Lim J , Kim J T , Sung S W , Yoo C . (2010). Data-driven prediction model of indoor air quality in an underground space. Korean Journal of Chemical Engineering, 27(6): 1675–1680

[18] Längkvist M , Karlsson L , Loutfi A . (2014). A review of unsupervised feature learning and deep learning for time-series modeling. Pattern Recognition Letters, 42: 11–24

[19] Lee S H , Choi S . (2007). Two-dimensional canonical correlation analysis. IEEE Signal Processing Letters, 14(10): 735–738

[20] Liu H , Yang C , Huang M , Yoo C . (2020). Multivariate statistical monitoring of subway indoor air quality using dynamic concurrent partial least squares. Environmental Science and Pollution Research International, 27(4): 4159–4169

[21] Loy-Benitez J , Heo S , Yoo C . (2020). Soft sensor validation for monitoring and resilient control of sequential subway indoor air quality through memory-gated recurrent neural networks-based autoencoders. Control Engineering Practice, 97: 104330

[22] Makarenkov V , Legendre P . (2002). Nonlinear redundancy analysis and canonical correspondence analysis based on polynomial regression. Ecology, 83(4): 1146–1161

[23] Mannan M , Al-Ghamdi S G . (2021). Indoor air quality in buildings: a comprehensive review on the factors influencing air pollution in residential and commercial structure. International Journal of Environmental Research and Public Health, 18(6): 3276

[24] Mehmood T , Liland K H , Snipen L , Saebo S . (2012). A review of variable selection methods in partial least squares regression. Chemometrics and Intelligent Laboratory Systems, 118: 62–69

[25] Melaku Y A , Gill T K , Taylor A W , Adams R , Shi Z . (2018). A comparison of principal component analysis, partial least-squares and reduced-rank regressions in the identification of dietary patterns associated with bone mass in ageing Australians. European Journal of Nutrition, 57(5): 1969–1983

[26] Memarzadeh M , Matthews B , Avrekh I . (2020). Unsupervised anomaly detection in flight data using convolutional variational auto-encoder. Aerospace (Basel, Switzerland), 7(8): 115

[27] Ooi S K , Tanny D , Chen J , Wang K . (2021). Developing semi-supervised variational autoencoder-generative adversarial network models to enhance quality prediction performance. Chemometrics and Intelligent Laboratory Systems, 217: 104385

[28] Pu Z , Yan J , Chen L , Li Z , Tian W , Tao T , Xin K . (2023). A hybrid Wavelet-CNN-LSTM deep learning model for short- term urban water demand forecasting. Frontiers of Environmental Science & Engineering, 17(2): 22

[29] Qian J , Song Z , Yao Y , Zhu Z , Zhang X . (2022). A review on autoencoder based representation learning for fault detection and diagnosis in industrial processes. Chemometrics and Intelligent Laboratory Systems, 231: 104711

[30] Qin Y , Lou Z , Wang Y , Lu S , Sun P . (2022). An analytical partial least squares method for process monitoring. Control Engineering Practice, 124: 105182

[31] Ran X , Chen W , Yvert B , Zhang S . (2022). A hybrid autoencoder framework of dimensionality reduction for brain-computer interface decoding. Computers in Biology and Medicine, 148: 105871

[32] RasiwasiaNMahajanDMahadevanVAggarwalG (2014). Cluster canonical correlation analysis. Reykjavik, ICELAND, 823–831

[33] Reche C , Moreno T , Martins V , Minguillon M C , Jones T , de Miguel E , Capdevila M , Centelles S , Querol X . (2017). Factors controlling particle number concentration and size at metro stations. Atmospheric Environment, 156: 169–181

[34] San Martin G , Lopez Droguett E , Meruane V , das Chagas Moura M . (2019). Deep variational auto-encoders: a promising tool for dimensionality reduction and ball bearing elements fault diagnosis. Structural Health Monitoring, 18(4): 1092–1128

[35] Shu X , Bao T , Li Y , Gong J , Zhang K . (2022). VAE-TALSTM: a temporal attention and variational autoencoder-based long short-term memory framework for dam displacement prediction. Engineering with Computers, 38(4): 3497–3512

[36] Śmiełowska M , Marc M , Zabiegala B . (2017). Indoor air quality in public utility environments: a review. Environmental Science and Pollution Research International, 24(12): 11166–11176

[37] SouzaFAraujoRMendesJ (2016). Review of soft sensor methods or regression applications. Chemometrics and Intelligent Laboratory Systems 152: 69-79 doi: 10.1016/j.chemolab.2015.12.011

[38] Su X , Sutarlie L , Loh X J . (2020). Sensors and analytical technologies for air quality: particulate matters and bioaerosols. Chemistry, an Asian Journal, 15(24): 4241–4255

[39] Sun J , Wang X , Xiong N , Shao J . (2018). Learning sparse representation with variational auto-encoder for anomaly detection. IEEE Access: Practical Innovations, Open Solutions, 6: 33353–33361

[40] Vallejo M , de La Espriella C , Gómez-Santamaría J , Ramírez-Barrera A F , Delgado-Trejos E . (2020). Soft metrology based on machine learning: a review. Measurement Science & Technology, 31(3): 032001

[41] Wang B , Li Z , Dai Z , Lawrence N , Yan X . (2020). Data-driven mode identification and unsupervised fault detection for nonlinear multimode processes. IEEE Transactions on Industrial Informatics, 16(6): 3651–3661

[42] Wang J , Lu Y , Xin C , Yoo C , Liu H . (2022). Kernel PLS with AdaBoost ensemble learning for particulate matters forecasting in subway environment. Measurement, 204: 111974

[43] Wei W , Ramalho O , Malingre L , Sivanantham S , Little J C , Mandin C . (2019). Machine learning and statistical models for predicting indoor air quality. Indoor Air, 29(5): 704–726

[44] Xie W , You J , Zhi C , Li L . (2021). The toxicity of ambient fine particulate matter (PM_2.5) to vascular endothelial cells. Journal of Applied Toxicology, 41(5): 713–723

[45] Xu B , Hao J L . (2017). Air quality inside subway metro indoor environment worldwide: a review. Environment International, 107: 33–46

[46] Xu Q S , Liang Y Z , Shen H L . (2001). Generalized PLS regression. Journal of Chemometrics, 15(3): 135–148

[47] Yan X , Xu Y , She D , Zhang W . (2022). Reliable fault diagnosis of bearings using an optimized stacked variational denoising auto-encoder. Entropy (Basel, Switzerland), 24(1): 24010036

[48] Zhang K , Yang J , Sha J , Liu H . (2022). Dynamic slow feature analysis and random forest for subway indoor air quality modeling. Building and Environment, 213: 108876

[49] Zhang M H , Xu Q S , Massart D L . (2004). Averaged and weighted average partial least squares. Analytica Chimica Acta, 504(2): 279–289

[50] Zhang Y , Li F , Ni C , Gao S , Zhang S , Xue J , Ning Z , Wei C , Fang F , Nie Y . . (2023). Prediction and cause investigation of ozone based on a double-stage attention mechanism recurrent neural network. Frontiers of Environmental Science & Engineering, 17(2): 21

[51] Zhu J , Shi H , Song B , Tao Y , Tan S . (2020). Information concentrated variational auto-encoder for quality-related nonlinear process monitoring. Journal of Process Control, 94: 12–25

Acknowledgements

This work was supported by the Opening Project of Guangxi Key Laboratory of Clean Pulp & Papermaking and Pollution Control, China (No. 2021KF11), the Shandong Provincial Natural Science Foundation, China (No. ZR2021MF135), the National Natural Science Foundation of China (No. 52170001) and the Natural Science Foundation of Jiangsu Provincial Universities, China (No. 22KJA530003).

Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Electronic Supplementary Material

Supplementary material is available in the online version of this article at https://doi.org/10.1007/s11783-024-1768-7 and is accessible for authorized users.

Data Accessibility Statement

The data and code that support the findings of this study are available from the corresponding author, Hongbin Liu, upon reasonable request.

RIGHTS & PERMISSIONS

2024 Higher Education Press