Application of a Neural Network in Pressure Drop Prediction

作者：Zabihi R^*; Schaffie M; Ranjbar M

来源：Energy Sources, Part A: Recovery, Utilization, and Environmental Effects , 2014, 36(6): 613-622.

DOI：10.1080/15567036.2011.588679

• Berea sandstone
• calcium sulfate scaling
• multi-layer perceptron
• pressure drop
• radial basis network

Pressure drop is one of the most common problems that occur during waterflooding. In this study, pressure drop due to calcium sulfate scaling in Berea sandstone was predicted by multi-layer perceptron and radial basis function neural networks. To design the optimum multi-layer perceptron model, number of neurons, number of hidden layers, and training function were studied. Also, the spread parameter and normalization method of data were examined for radial basis function model. The multi-layer perceptron model, with two hidden layers, predicted pressure drop with an average error of 1.19%, whereas error of the radial basis function model was 1.76%.

• 出版日期2014-3-19