Modeling of CO distribution in Istanbul using Artificial Neural Networks


Sahin U., Ucan O., Soyhan B., Bayat C.

FRESENIUS ENVIRONMENTAL BULLETIN, vol.13, no.9, pp.839-845, 2004 (Peer-Reviewed Journal) identifier

  • Publication Type: Article / Article
  • Volume: 13 Issue: 9
  • Publication Date: 2004
  • Journal Name: FRESENIUS ENVIRONMENTAL BULLETIN
  • Journal Indexes: Science Citation Index Expanded
  • Page Numbers: pp.839-845

Abstract

Artificial Neural Network (ANN) is one of the popular methods in optimization of complex engineering problems compared to the classical statistical methods. ANN approximates non-linear input-output variables and finds an optimum correlation between these variables. Thus the structure of the overall system is simplified. ANN function approximation is achieved by identifying the input-output pattern pairs, using the following steps: (I) Selection of the neural structure (namely the number of layers and that of neurons), (II) Training of ANN using Back-Propagation (BP) algorithms. ANN coefficients can be trained as any system performance characteristics by monitoring test data. (III) Validation of the network to verify generalization capability.