Vertical Oil-in-Water Flow Pattern Identification with Deep CNN-LSTM Network

Oil-in-Water two-phase flows widely exist in various industrial applications. Identifying the flow patterns is of great importance for the optimization of the oil-water two-phase flow control system. In this paper, a Deep CNN-LSTM network is proposed to extract the spatial-temporal features of the vertical oil-in-water two phase flows. Then the flow patterns of the vertical oil-in-water two phase flows are identified with the extracted spatial-temporal features. The testing results on our data set show that the proposed network can effectively identify the typical oil-in-water two-phase flow patterns in vertical pipes with a relatively high accuracy.