Original Article - Open Access.

Idioma principal | Segundo idioma

UMA ABORDAGEM DE APRENDIZAGEM PROFUNDA COM WAVELETS PARA A PREVISÃO DE OZÔNIO TROPOSFÉRICO EM UMA REGIÃO METROPOLITANA TROPICAL

A DEEP LEARNING APPROACH WITH WAVELETS TO FORECASTING TROPOSPHERIC OZONE IN A TROPICAL METROPOLITAN REGION

Carmo Junior, Clovis; Winkler, Ingrid; Nascimento, Erick Giovani Sperandio;

Original Article:

Redes de aprendizado profundo (RAP) têm sido usadas com sucesso em modelos de predição de poluentes do ar. Após treinadas, elas aprendem as relações complexas e não-linearidades das variáveis atmosféricas e acham soluções que requerem menos recursos computacionais que modelos numéricos e analíticos. Este estudo visa desenvolver uma RAP para predição das concentrações de O3 nas próximas 24hs. Testou-se várias arquiteturas de RAP, usando séries temporais multivariáveis como dados de entrada. Integrando as WL na arquitetura, temos melhores resultados, indicando consistência para modelos de predição de poluentes. Estes modelos de aprendizagem profunda demonstram flexibilidade, capacidade de ajuste e mapeamento da complexidade não-linear de dados de predição de poluentes.

Original Article:

Deep neural networks (DNN) have been successfully applied to develop air pollutant forecasting models. Once trained, they can learn the complex relationships and non-linearities present in atmospheric variables, delivering solutions that require less computational resources than numerical and analytical models. This study aims to develop a DNN to forecast concentrations of ozone (O3) for the next 24hs. We tested several DNN using, as input, a multivariate time series dataset. When wavelets (WL) were integrated into the DNN’s architectures, they improved the models’ performance, pointing towards a consistent modeling architecture for air pollution forecasting. These deep learning models showed flexibility, strong nonlinear fitting capabilities and an ability to map nonlinear complexity from the data for air pollution forecasting.

Palavras-chave: Poluição do ar; Predição, Aprendizado profundo; Wavelets, Ozônio,

Palavras-chave: Air pollution; Forecast; Deep learning; Wavelets, Ozone,

DOI: 10.5151/siintec2021-208398

Referências bibliográficas
  • [1] "1 PEDRUZZI, Rizzieri et al. Performance evaluation of a photochemical model using different boundary conditions over the urban and industrialized metropolitan area of Vitória, Brazil. Environmental Science and Pollution Research, [S.L.], v. 26, n. 16, p. 16125-16144, 10 abr. 2019. Springer Science and Business Media LLC.
  • [2] 2 WHO 2013. Review of evidence on health aspects of air pollution-REVIHAAP project: final technical report. World Health Organization Regional Office for Europe. https://www.euro.who.int/en/health-topics/environment-and-health/air-quality/publications/2013/review-of-evidence-on-health-aspects-of-air-pollution-revihaap-project-final-technical-report Accessed May 3, 2021
  • [3] 3 RYBARCZYK, Yves et al. Machine Learning Approaches for Outdoor Air Quality Modelling: a systematic review. Applied Sciences, [S.L.], v. 8, n. 12, p. 2570, 11 dez. 2018. MDPI AG.
  • [4] 4 CONAMA Environmental National Committee, New national air quality standards, Res. 491/ 2018. www2.mma.gov.br/port/conama/legiabre.cfm?codlegi=740. Accessed on: 23 Apr. 2021
  • [5] 5 KITAGAWA, Yasmin Kaore Lago et al. ASSESSMENT OF PRIMARY AIR POLLUTANTS IN A TROPICAL METROPOLITAN REGION BY COMBINING LOCAL AND GLOBAL EMISSIONS INVENTORIES. Air Pollution XXVII, [S.L.], v. 236, p. 99-110, 4 set. 2019. WIT Press.
  • [6] 6 CAI, Jianxian et al. An Air Quality Prediction Model Based on a Noise Reduction Self-Coding Deep Network. Mathematical Problems in Engineering, [S.L.], v. 2020, p. 1-12, 15 maio 2020. Hindawi Limited.
  • [7] 7 RUSSO, Ana et al. Neural network forecast of daily pollution concentration using optimal meteorological data at synoptic and local scales. Atmospheric Pollution Research, [S.L.], v. 6, n. 3, p. 540-549, maio 2015. Elsevier BV.
  • [8] 8 JANARTHANAN, R. et al. A deep learning approach for prediction of air quality index in a metropolitan city. Sustainable Cities and Society, [S.L.], v. 67, p. 102720, abr. 2021. Elsevier BV.
  • [9] 9 LIU, Bingchun et al. Air Pollutant Concentration Forecasting Using Long Short-Term Memory Based on Wavelet Transform and Information Gain: a case study of beijing. Computational Intelligence and Neuroscience, [S.L.], v. 2020, p. 1-12, 30 set. 2020. Hindawi Limited.
  • [10] 10 LIU, Bingchun et al. A Novel Method for Regional NO2 Concentration Prediction Using Discrete Wavelet Transform and an LSTM Network. Computational Intelligence and Neuroscience, [S.L.], v. 2021, p. 1-14, 7 abr. 2021. Hindawi Limited.
  • [11] 11 CABANEROS, Sheen McLean et al. Spatial estimation of outdoor NO2 levels in Central London using deep neural networks and a wavelet decomposition technique. Ecological Modelling, [S.L.], v. 424, p. 109017, maio 2020. Elsevier BV.
  • [12] 12 EBRAHIMI-KHUSFI, Zohre et al. Predicting the ground-level pollutants concentrations and identifying the influencing factors using machine learning, wavelet transformation, and remote sensing techniques. Atmospheric Pollution Research, [S.L.], v. 12, n. 5, p. 101064, maio 2021. Elsevier BV.
  • [13] 13 R. JUNIOR, Amilton S. et al. Assessing Recurrent and Convolutional Neural Networks for Tropospheric Ozone Forecasting in the Region of Vitória, Brazil. Air Pollution XXVIII, [S.L.], v. 244, p. 101-112, 27 jul. 2020. WIT Press.
  • [14] 14 ALBUQUERQUE, Taciana Toledo de Almeida et al. WRF-SMOKE-CMAQ modeling system for air quality evaluation in São Paulo megacity with a 2008 experimental campaign data. Environmental Science and Pollution Research, [S.L.], v. 25, n. 36, p. 36555-36569, 29 out. 2018. Springer Science and Business Media LLC.
  • [15] 15 GIDHAGEN, Lars et al. Experimental and model assessment of PM2.5 and BC emissions and concentrations in a Brazilian city – the Curitiba case study. Atmospheric Chemistry and Physics Discussions, [S.L.], p. 1-37, 10 dez. 2018. Copernicus GmbH.
  • [16] 16 BORREGO, C. et al. Modelling the photochemical pollution over the metropolitan area of Porto Alegre, Brazil. Atmospheric Environment, [S.L.], v. 44, n. 3, p. 370-380, jan. 2010. Elsevier BV.
  • [17] 17 Zucatelli, Pedro Junior et al. Nowcasting prediction of wind speed using computational intelligence and wavelet in Brazil, International Journal for Computational Methods in Engineering Science and Mechanics, 2020.
  • [18] 18 OTTO, S.A. (2019, Jan.,7). How to normalize the RMSE [Blog post]. Retrieved from https://www.marinedatascience.co/blog/2019/01/07/normalizing-the-rmse/. Ass. in Apr 2021.
Como citar:

Carmo Junior, Clovis; Winkler, Ingrid; Nascimento, Erick Giovani Sperandio; "UMA ABORDAGEM DE APRENDIZAGEM PROFUNDA COM WAVELETS PARA A PREVISÃO DE OZÔNIO TROPOSFÉRICO EM UMA REGIÃO METROPOLITANA TROPICAL", p. 361-368 . In: VII International Symposium on Innovation and Technology. São Paulo: Blucher, 2021.
ISSN 2357-7592, DOI 10.5151/siintec2021-208398

últimos 30 dias | último ano | desde a publicação


downloads


visualizações


indexações