Specialist system in flow pattern identification using artificial neural Networks

In this work, an application of artificial intelligence in the oils & gas industry is developed to identify flow patterns horizontal and vertical pipes two-phase oil water, normalizing word information converting it numerical values through development neural network, whose input layer composed surface velocities each fluid, velocity mixture, volumetric fraction substances, diameter inclination pipelines viscosity. The Artificial Neural Networks (ANN) has two hidden layers 45 neurons. database with which model was trained, validated, tested 6993 rows corresponding inputs intelligent system particular-ized for annular DO/W pipelines. Notice that obtained after re-engineering presented by 12 18 authors piping, respectively. Finally, mean square error around 1.38%, a maximum coefficient determination 0.79.