A Recurrent Neural Network Approach to Improve the Air Quality Index Prediction
Every year, cities all over the world face the problem of the air pollution. In particular seasons, such as the winter, the levels of bad particles coming from the industrial and domestic heating systems increase the risk of pulmonary diseases. Thus, for both the city majors and citizens, it is important to understand and predict the air pollution levels in advance to safe guard the health. Modern forecasting systems are able to alert the population in advance only about the general weather condition, while the air quality information are almost not considered at all. The reasons are manifold and they mostly depend by the difficult that the modern systems have to generalize the problem and correct elaborate the data coming from the sensors. In this paper we address the problem of forecasting the bands of the different air pollutants according to the Air Quality Index in the Apulia region. Using two different Recurrent Neural Network models, we have performed two tests to prove that it is possible to predict the level of the pollutants in a specific area by using the data coming from the surrounding area. By using this approach on both the weather and air stations on the territory it is possible to have alerts many days ahead on the pollution levels.
KeywordsAir Quality Index Recurrent Neural Network Blind prediction
The present study was developed and granted in the framework of the project: “SeVaRA” (European Community, Minister of the Economic Development, Apulia Region, BURP n. 1883 of the 24/10/2018, Id:2NQR592).
- 2.Bevilacqua, V., Cassano, F., Mininno, E., Iacca, G.: Optimizing feed-forward neural network topology by multi-objective evolutionary algorithms: a comparative study on biomedical datasets. In: Italian Workshop on Artificial Life and Evolutionary Computation, pp. 53–64. Springer, Cham (2015)Google Scholar
- 7.Stevens, M., D’Hondt, E.: Crowdsourcing of pollution data using smartphones. In: Workshop on Ubiquitous Crowdsourcing (2010)Google Scholar
- 13.Caivano, D., Cassano, F., Fogli, D., Lanzilotti, R., Piccinno, A.: We@ home: a gamified application for collaboratively managing a smart home. In: International Symposium on Ambient Intelligence, pp. 79–86, Springer, Cham (2017)Google Scholar
- 14.Chung, J., Gulcehre, C., Cho, K., Bengio, Y.: Empirical evaluation of gated recurrent neural networks on sequence modeling. arXiv preprint arXiv:1412.3555 (2014)
- 15.World Health Organization: Air quality guidelines for Europe (2000)Google Scholar