Development of ANFIS models for air quality forecasting and input optimization for reducing the computational cost and time
| Autor: | Pramila Goyal, Amit Kumar Gorai, Kanchan Prasad |
|---|---|
| Rok vydání: | 2016 |
| Předmět: |
Pollutant
Atmospheric Science Engineering Adaptive neuro fuzzy inference system Meteorology business.industry Air pollution 02 engineering and technology Collinearity 010501 environmental sciences medicine.disease_cause 01 natural sciences Wind speed Dew point 0202 electrical engineering electronic engineering information engineering medicine 020201 artificial intelligence & image processing business Visibility Air quality index 0105 earth and related environmental sciences General Environmental Science |
| Zdroj: | Atmospheric Environment. 128:246-262 |
| ISSN: | 1352-2310 |
| DOI: | 10.1016/j.atmosenv.2016.01.007 |
| Popis: | This study aims to develop adaptive neuro-fuzzy inference system (ANFIS) for forecasting of daily air pollution concentrations of five air pollutants [sulphur dioxide (SO2), nitrogen dioxide (NO2), carbon monoxide (CO), ozone (O3) and particular matters (PM10)] in the atmosphere of a Megacity (Howrah). Air pollution in the city (Howrah) is rising in parallel with the economics and thus observing, forecasting and controlling the air pollution becomes increasingly important due to the health impact. ANFIS serve as a basis for constructing a set of fuzzy IF-THEN rules, with appropriate membership functions to generate the stipulated input–output pairs. The ANFIS model predictor considers the value of meteorological factors (pressure, temperature, relative humidity, dew point, visibility, wind speed, and precipitation) and previous day's pollutant concentration in different combinations as the inputs to predict the 1-day advance and same day air pollution concentration. The concentration value of five air pollutants and seven meteorological parameters of the Howrah city during the period 2009 to 2011 were used for development of the ANFIS model. Collinearity tests were conducted to eliminate the redundant input variables. A forward selection (FS) method is used for selecting the different subsets of input variables. Application of collinearity tests and FS techniques reduces the numbers of input variables and subsets which helps in reducing the computational cost and time. The performances of the models were evaluated on the basis of four statistical indices (coefficient of determination, normalized mean square error, index of agreement, and fractional bias). |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full Text from ScienceDirect