Artificial Neural Networks, Sequence-to-Sequence LSTMs, and Exogenous Variables as Analytical Tools for NO<sub>2</sub> (Air Pollution) Forecasting: A Case Study in the Bay of Algeciras (Spain)
This study aims to produce accurate predictions of the NO<sub>2</sub> concentrations at a specific station of a monitoring network located in the Bay of Algeciras (Spain). Artificial neural networks (ANNs) and sequence-to-sequence long short-term memory networks (LSTMs) were used to crea...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-03-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/21/5/1770 |
| _version_ | 1797414574740209664 |
|---|---|
| author | Javier González-Enrique Juan Jesús Ruiz-Aguilar José Antonio Moscoso-López Daniel Urda Lipika Deka Ignacio J. Turias |
| author_facet | Javier González-Enrique Juan Jesús Ruiz-Aguilar José Antonio Moscoso-López Daniel Urda Lipika Deka Ignacio J. Turias |
| author_sort | Javier González-Enrique |
| collection | DOAJ |
| description | This study aims to produce accurate predictions of the NO<sub>2</sub> concentrations at a specific station of a monitoring network located in the Bay of Algeciras (Spain). Artificial neural networks (ANNs) and sequence-to-sequence long short-term memory networks (LSTMs) were used to create the forecasting models. Additionally, a new prediction method was proposed combining LSTMs using a rolling window scheme with a cross-validation procedure for time series (LSTM-CVT). Two different strategies were followed regarding the input variables: using NO<sub>2</sub> from the station or employing NO<sub>2</sub> and other pollutants data from any station of the network plus meteorological variables. The ANN and LSTM-CVT exogenous models used lagged datasets of different window sizes. Several feature ranking methods were used to select the top lagged variables and include them in the final exogenous datasets. Prediction horizons of <i>t</i> + 1, <i>t</i> + 4 and <i>t</i> + 8 were employed. The exogenous variables inclusion enhanced the model’s performance, especially for <i>t</i> + 4 (<i>ρ</i> ≈ 0.68 to <i>ρ</i> ≈ 0.74) and <i>t</i> + 8 (<i>ρ</i> ≈ 0.59 to <i>ρ</i> ≈ 0.66). The proposed LSTM-CVT method delivered promising results as the best performing models per prediction horizon employed this new methodology. Additionally, per each parameter combination, it obtained lower error values than ANNs in 85% of the cases. |
| first_indexed | 2024-03-09T05:35:21Z |
| format | Article |
| id | doaj.art-308bf6ac2d86432287ff9b542a328b19 |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-03-09T05:35:21Z |
| publishDate | 2021-03-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-308bf6ac2d86432287ff9b542a328b192023-12-03T12:29:31ZengMDPI AGSensors1424-82202021-03-01215177010.3390/s21051770Artificial Neural Networks, Sequence-to-Sequence LSTMs, and Exogenous Variables as Analytical Tools for NO<sub>2</sub> (Air Pollution) Forecasting: A Case Study in the Bay of Algeciras (Spain)Javier González-Enrique0Juan Jesús Ruiz-Aguilar1José Antonio Moscoso-López2Daniel Urda3Lipika Deka4Ignacio J. Turias5Intelligent Modelling of Systems Research Group (MIS), Department of Computer Science Engineering, Polytechnic School of Engineering, University of Cádiz, 11204 Algeciras, SpainIntelligent Modelling of Systems Research Group (MIS), Department of Industrial and Civil Engineering, Polytechnic School of Engineering, University of Cádiz, 11204 Algeciras, SpainIntelligent Modelling of Systems Research Group (MIS), Department of Industrial and Civil Engineering, Polytechnic School of Engineering, University of Cádiz, 11204 Algeciras, SpainGrupo de Inteligencia Computacional Aplicada (GICAP), Departamento de Ingeniería Informática, Escuela Politécnica Superior, Universidad de Burgos, Av. Cantabria s/n, 09006 Burgos, SpainThe De Montfort University Interdisciplinary Group in Intelligent Transport Systems (DIGITS), Department of Computer Science and Informatics, De Montfort University, Leicester LE1 9BH, UKIntelligent Modelling of Systems Research Group (MIS), Department of Computer Science Engineering, Polytechnic School of Engineering, University of Cádiz, 11204 Algeciras, SpainThis study aims to produce accurate predictions of the NO<sub>2</sub> concentrations at a specific station of a monitoring network located in the Bay of Algeciras (Spain). Artificial neural networks (ANNs) and sequence-to-sequence long short-term memory networks (LSTMs) were used to create the forecasting models. Additionally, a new prediction method was proposed combining LSTMs using a rolling window scheme with a cross-validation procedure for time series (LSTM-CVT). Two different strategies were followed regarding the input variables: using NO<sub>2</sub> from the station or employing NO<sub>2</sub> and other pollutants data from any station of the network plus meteorological variables. The ANN and LSTM-CVT exogenous models used lagged datasets of different window sizes. Several feature ranking methods were used to select the top lagged variables and include them in the final exogenous datasets. Prediction horizons of <i>t</i> + 1, <i>t</i> + 4 and <i>t</i> + 8 were employed. The exogenous variables inclusion enhanced the model’s performance, especially for <i>t</i> + 4 (<i>ρ</i> ≈ 0.68 to <i>ρ</i> ≈ 0.74) and <i>t</i> + 8 (<i>ρ</i> ≈ 0.59 to <i>ρ</i> ≈ 0.66). The proposed LSTM-CVT method delivered promising results as the best performing models per prediction horizon employed this new methodology. Additionally, per each parameter combination, it obtained lower error values than ANNs in 85% of the cases.https://www.mdpi.com/1424-8220/21/5/1770forecastingfeature selectionair pollutionnitrogen dioxideartificial neural networksLSTMs |
| spellingShingle | Javier González-Enrique Juan Jesús Ruiz-Aguilar José Antonio Moscoso-López Daniel Urda Lipika Deka Ignacio J. Turias Artificial Neural Networks, Sequence-to-Sequence LSTMs, and Exogenous Variables as Analytical Tools for NO<sub>2</sub> (Air Pollution) Forecasting: A Case Study in the Bay of Algeciras (Spain) Sensors forecasting feature selection air pollution nitrogen dioxide artificial neural networks LSTMs |
| title | Artificial Neural Networks, Sequence-to-Sequence LSTMs, and Exogenous Variables as Analytical Tools for NO<sub>2</sub> (Air Pollution) Forecasting: A Case Study in the Bay of Algeciras (Spain) |
| title_full | Artificial Neural Networks, Sequence-to-Sequence LSTMs, and Exogenous Variables as Analytical Tools for NO<sub>2</sub> (Air Pollution) Forecasting: A Case Study in the Bay of Algeciras (Spain) |
| title_fullStr | Artificial Neural Networks, Sequence-to-Sequence LSTMs, and Exogenous Variables as Analytical Tools for NO<sub>2</sub> (Air Pollution) Forecasting: A Case Study in the Bay of Algeciras (Spain) |
| title_full_unstemmed | Artificial Neural Networks, Sequence-to-Sequence LSTMs, and Exogenous Variables as Analytical Tools for NO<sub>2</sub> (Air Pollution) Forecasting: A Case Study in the Bay of Algeciras (Spain) |
| title_short | Artificial Neural Networks, Sequence-to-Sequence LSTMs, and Exogenous Variables as Analytical Tools for NO<sub>2</sub> (Air Pollution) Forecasting: A Case Study in the Bay of Algeciras (Spain) |
| title_sort | artificial neural networks sequence to sequence lstms and exogenous variables as analytical tools for no sub 2 sub air pollution forecasting a case study in the bay of algeciras spain |
| topic | forecasting feature selection air pollution nitrogen dioxide artificial neural networks LSTMs |
| url | https://www.mdpi.com/1424-8220/21/5/1770 |
| work_keys_str_mv | AT javiergonzalezenrique artificialneuralnetworkssequencetosequencelstmsandexogenousvariablesasanalyticaltoolsfornosub2subairpollutionforecastingacasestudyinthebayofalgecirasspain AT juanjesusruizaguilar artificialneuralnetworkssequencetosequencelstmsandexogenousvariablesasanalyticaltoolsfornosub2subairpollutionforecastingacasestudyinthebayofalgecirasspain AT joseantoniomoscosolopez artificialneuralnetworkssequencetosequencelstmsandexogenousvariablesasanalyticaltoolsfornosub2subairpollutionforecastingacasestudyinthebayofalgecirasspain AT danielurda artificialneuralnetworkssequencetosequencelstmsandexogenousvariablesasanalyticaltoolsfornosub2subairpollutionforecastingacasestudyinthebayofalgecirasspain AT lipikadeka artificialneuralnetworkssequencetosequencelstmsandexogenousvariablesasanalyticaltoolsfornosub2subairpollutionforecastingacasestudyinthebayofalgecirasspain AT ignaciojturias artificialneuralnetworkssequencetosequencelstmsandexogenousvariablesasanalyticaltoolsfornosub2subairpollutionforecastingacasestudyinthebayofalgecirasspain |