Loading...

Approximation of titration curves by artificial neural networks and its application to pH control

Pishvaie, M. R ; Sharif University of Technology | 2000

163 Viewed
  1. Type of Document: Article
  2. Publisher: Sharif University of Technology , 2000
  3. Abstract:
  4. Advanced model-based control of pH processes is noticeably a chemical modeling issue, because it can have a profound effect on the attainable control quality. This is especially the case when the pH regulation of streams, consisting of hundreds of constituents with varying concentrations, is encountered. The severe non-linear behavior of pH processes is reflected in the titration curve of the process stream. The performances of all model-based controllers are highly dependent on the accuracy of the model. Considering a great number of parameters such as dissociation constants, solubility products and characteristic concentrations places the designer in a dilemma of choosing between approximate physicochemical models and empirical ones, both having their own merits and shortcomings. Using radial basis function artificial neural networks, a new modeling approach for approximating the titration curve is proposed in this article and some physical interpretations for selecting the neural network parameters are given. Both off-line and on-line training for adaptive control strategy can be used. The effectiveness of the proposed scheme is demonstrated by simulation and experimental studies
  5. Keywords:
  6. Hemical engineering ; pH ; Neural network ; Control system
  7. Source: Scientia Iranica ; Volume 6, Issue 5 , 2000 , Pages 82-91 ; 10263098 (ISSN)
  8. URL: http://scientiairanica.sharif.edu/article_2801.html