Hybrid Artificial Neural Network Coupled with Kalman Filters for Air Quality Forecasting in Guadalajara, Mexico

González-Figueredo, Carlos; Egurrola-Hernández, E.A.; Ramírez-Briseño, R.L.; DeLosReyes-Corona, A.; DeAlba-Martínez, Hugo

View/Open

AirQualityforecastingwithRNAandKalmanAirQuality2018.pdf (1.219Mb)

Date

2018-03

Author

González-Figueredo, Carlos

Egurrola-Hernández, E.A.

Ramírez-Briseño, R.L.

DeLosReyes-Corona, A.

DeAlba-Martínez, Hugo

Metadata

Show full item record

URI

https://hdl.handle.net/20.500.12032/70209

Description

This study aims to develop a novel hybrid scheme of Artificial Neural Networks (ARN) coupled to a non-linear Kalman filter for air quality forecasting in Guadalajara Metropolitan Area, in Mexico. ARN’s are widely used for air quality forecasting, however these schemes need large amounts of data regarding the pollutants concentration levels and meteorological data in order to manage reliable forecasting. To address this issue, we present a scheme consisting of Neural Network models assisted by nonlinear Kalman filter that manage to considerably improve the forecasting performance, adding robustness in case of lack of data, and reducing the need of retraining over time.

Collections

Documentos - ITESO