Loading...

Artificial Neural Network Based Prediction of Heat of Adsorption of Alkanes on Various Zeolites

Zhiyani, Mehrzad | 2014

506 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 45837 (03)
  4. University: Sharif University of Technology
  5. Department: Chemistry
  6. Advisor(s): Gobal, Ferydon
  7. Abstract:
  8. Generally, predicting the adsorption and Catalytic characteristics of the solids base on the primary principles is impossible; this is only possible for small molecules and single crystal surfaces. On the other hand, phenomenological approaches, which are based on experimental data, are efficient approaches in many cases. The best predictions and designs are done by those who have an extensive information “resources” about the “Reactive Substances- Catalytic-conditions” and are able to “analyze” the data based on the “chemical physical Models”. A Neural network is an extremely simplified model of the human brain that predicts a complex characteristic(s) from a series of primary characteristics to a nonlinear relation. In this case, the connection topology of units and the weight of connections forms the network learning system for a specific purpose; e.g., personal experience and the ability for prediction. In this research, an artificial neural network model, called ANN-LM, is proposed to predict the alkane adsorption enthalpy on three zeolites, i.e., ZSM5, ZSM22, and Beta. This nonlinear model is capable to predict 98.4%, 99.5%, and 99.4% of alkane adsorption enthalpy variation for ZSM5, ZSM22, and Beta, respectively. The simulations of the proposed model demonstrate the promising results for the prediction of alkane adsorption enthalpy even for a limited network learning set of 16 molecules
  9. Keywords:
  10. Alkane ; Zeolite ; Artificial Neural Network ; Adsorption ; Adsorption Enthalpy

 Digital Object List

 Bookmark

No TOC