Loading...

Modelling and Prediction Air Polutants Level in Tehran Using Dynamic Neural Networks

Khosravi, Neda | 2014

917 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 47406 (09)
  4. University: Sharif University of Technology
  5. Department: Civil Engineering
  6. Advisor(s): Erhami, Mohammad
  7. Abstract:
  8. In parallel to the growing of population in Tehran metropolitan, air pollution in this city has become to a major problem. From which high concentration of pollutants have adverse effects on public health, accurate estimating and forecasting of concentrations for several days ahead, can provide the possibility to implement the management measures to reduce hazard and risks. Among the air pollution models, application of statistic models based on neural network in comparison to the traditional deterministic models are easier and less costly. In most studies, static models use a classical single MLP to predict one step ahead. For this purpose ANN models are required to estimate next value of time series without feeding back previous predicted values to model’s input regressor. In contrast dynamic models use recursive structure to feedback outputs and can be used to predict multi step ahead of time series. Between dynamic neural networks, NARX (nonlinear autoregressive with exogenous input) has been applied in various research areas such as Mechanics and Economics while the performance of this network for multi-step ahead prediction of hourly time series in air quality forecasting has not been evaluated. In addition in most air quality models the predictors (input variables) consist of routinely available and predictable meteorological parameters but in this work we have used these parameters to calculate several pre-processed data that are not routinely measured such as incoming solar radiation, mixing height, Monin-Obukhov length, friction velocity and stability classes. Then the best combination of inputs between meteorological and pre-processed data and parameters that can properly define the complex nature of pollutant source condition selected, but here we attempt to apply NARX neural network for prediction of 24 and 72 hours ahead of CO, O3, NO2 and PM10. The best model prediction is obtained for O3 and followed by CO, PM10 and NO2. The value of R2 for 1 day ahead prediction of these pollutants are 0.8, 0.79, 0.72 and 0.61 respectively
  9. Keywords:
  10. Dynamic Neural Network ; Ozone ; Nitrogen Dioxide ; Air Pollutants Prediction ; Preprocessing ; Air Particulate Matter ; Pre-Processors ; Nonlinear Autoregressive with Exogenous Input Network (NARx)

 Digital Object List

 Bookmark

No TOC