The Neural Network Modeling of Suspended Particulate Matter with Autoregressive Structure

BAYRAKTAR H., AKTAN M.

Ekoloji, cilt.19, ss.32-37, 2010 (SCI-Expanded) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 19
  • Basım Tarihi: 2010
  • Dergi Adı: Ekoloji
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, TR DİZİN (ULAKBİM)
  • Sayfa Sayıları: ss.32-37
  • Anahtar Kelimeler: Autoregressive, Erzurum, meteorological parameters, neural network, particulate matter, AIR-POLLUTION, ERZURUM-CITY
  • Atatürk Üniversitesi Adresli: Evet

Özet

Pollution Sources and emissions, and their interactions with terrain and the atmosphere arc necessary in developing appropriate air pollution management plans and action strategies. In this study, we will investigate the relationship between the total Suspended particulate matter (TSP) concentration and meteorological parameters such as wind speed and direction, temperature, air pressure, precipitation and relative humidity The TSP measurements of the past two days, sunshine duration, and sunshine amount in the winter season (November through March) in the city of Erzurum between the years 1990 and 2007 were investigated. The artificial neural network (ANN) models were constructed using a mixed autoregressive relationship to realize the stochastic nature of the TSP levels for each month of the winter season. The impact of wind direction on TSP concentration was introduced to the model by defining two variables. Linear regression models were also constructed to check the performance of the neural networks. The most significant factors affecting the TSP concentration are found to be the TSP level of the previous day (TSP(t-1)), expected temperature (temp), wind speed (w), air pressure (p), and precipitation (pc).