| Attributs | Valeurs |
|---|
| type
| |
| Is Part Of
| |
| Subject
| |
| Title
| - Artificial Neural Network Meta Models To Enhance the Predictionand Consistency of Multiphase Reactor Correlations
|
| has manifestation of work
| |
| related by
| |
| Author
| |
| Abstract
| - To increase confidence in neural network modeling of multiphase reactor characteristics, wehave to take advantage of some a priori knowledge of the physical laws governing these systemsin order to build neural models having phenomenological consistency (PC). A common form ofPC is the monotonicity constraint of a characteristic to be modeled with respect to some importantdimensional variables describing the multiphase system. When the inputs of a neural modelare functions (usually dimensionless) of the variables with respect to which monotonicity isexpected, the monotonicity might not be guaranteed, but such a drawback is only observed afterthe training. A genetic algorithm based methodology was proposed to produce several highlyaccurate and nearly PC networks differing by their inputs and architecture. PC and accuracywere shown to be boosted up meaningfully by combining such networks in a linear meta model.A new optimality criterion for the meta-model parameter identification was proposed, and theresults were compared with classical mean-squared error optimality criterion. The proof of theconcept of the approach was illustrated in modeling the two-phase pressure drop in countercurrently operated randomly packed beds.
|
| article type
| |
| is part of this journal
| |
