Artificial Intelligent Systems and Machine Learning

Nirjhar Bar ¹, Tapan Kumar Ghosh ¹, Manindra Nath Biswas ², Sudip Kumar Das ¹

Affiliations
1 Department of Chemical Engineering, University of Calcutta, 92, A. P. C. Road, Kolkata-700009, IN
2 Government College of Engineering & Leather Technology, LB Block, Sector III, Salt Lake City, Kolkata-700098, IN

Abstract

Experimental investigations have been carried out to determine the flow regime for air-water two-phase flow through 3 mm and 4 mm transparent tubes by visual observation. The feasibility of Artificial Neural Network (ANN) based techniques for the classifications of flow regimes are presented. Total 218 experimental data points are used in the ANN prediction. Five different well known ANN models have been tried to predict the flow regime. The ANN model based on MLP with Levenberg-Marquardt algorithm gives slightly better predictability over the other networks used.

Keywords

Flow Regime, Multilayer Perceptron, Radial Basis Function, Support Vector Machine, Principal Component Analysis.

Full Text