SINGLE PHASE FLUID FLOW CLASSIFICATION VIA NEURAL NETWORKS SUPPORT VECTOR MACHINE

摘要

The main objective of this paper is to apply Support Vector Machine (SVM) and Neural Networks (NNs) classifiers to characterize the flow pattern of a non-Newtonian fluid. This approach is achieved by simulating the Reynolds number equation through SVM and NNs. Classification of fluid flow patterns can be seen as a machine learning problem where the inputs are vectors of length 6 with attributes that represent the parameters which determine the fluid flow in the annulus/pipe. SVM classifiers can play an important role in the control/monitor of fluid flow, because the algorithm can be easily implemented in the design of an automated system that will provide real-time control of the fluid flow. The algorithm constructs separating hyperplanes, where the weights of the hyperplane represent as caled level of importance for eachp arameter. Preliminary results show that the most accurate model favors the SVM polynomial and radial basis kernel function model.