Download - ESANN2006 - A Cyclostationary Neural Network model for the prediction of the NO2 concentration
A Cyclostationary Neural Network Model for the Prediction
of the NO2 Concentration
Monica Bianchini, Ernesto Di Iorio, Marco Maggini,Chiara Mocenni, Augusto Pucci
Dipartimento di Ingegneria dell’InformazioneVia Roma 56, 53100 Siena (ITALY)
Air Pollution ProblemAir Pollution Problem
Nitrogen oxide (NONitrogen oxide (NOxx = NO + NO = NO + NO22) emissions are ) emissions are
among the most important factors affecting the among the most important factors affecting the air quality in urban areasair quality in urban areas
Traffic is the main problem on a local urban Traffic is the main problem on a local urban scalescale
Modeling efforts to predict and control the NOModeling efforts to predict and control the NOx x
concentrationsconcentrations
Development of tools for pollution managementDevelopment of tools for pollution management
Project GoalsProject Goals
Build an efficient Build an efficient neural based model for the prediction of the NO2 concentration
First prediction approximation for an early warning
Independence from exogenous data
Modeling the NO2 time series only based on the past concentrations of NO and NO2
The Cyclostationary NeuralNetwork Model
Correlation of past NO and current NO2 (daily periodicity)NO2(t) follows a cyclostationary dynamics (period T = 24)CNN model composed by 24 MLP blocks one for each random variable of the cyclostationary process
where T = 24 and fk with k = (t mod T) + 1, represents the k–th approximation function realized by the k–th MLP block
Experimental SetupExperimental Setup
Data gathered by ARPA Lombardia (northern Italy)ARPA supplies a real–time air pollution monitoring system composed by mobile and fixed stationsDataset made up by NO and NO2 concentrations detected hourly by the station number 649 (Brescia–Broletto)Performance measures: mean absolute error Performance measures: mean absolute error and mean square errorand mean square error
Future WorksFuture Works
CNN hardware implementation on NOCNN hardware implementation on NOxx
sensorssensors
Management of multiple data from Management of multiple data from different sensorsdifferent sensors
Testing on other urban area datasetsTesting on other urban area datasets
Testing on wider datasetsTesting on wider datasets