CNN-LSTM Model Optimized by Bayesian Optimization for Predicting Single-Well Production in Water Flooding Reservoir

Figure 12

The prediction results of 6 wells in reservoir A. (a–c) The error distribution of the 3 predicted production of well A\B\C by CNN-LSTM model. “1th-prediction” denotes the next month’s predicted production. Similarly, “2th-prediction” denotes the predicted production of the month after next. And “3th-prediction” means the predicted production of the third month in the future. (e–f) The comparison of real production curves and the 1th predicted production curves of well E/F/G by CNN-LSTM model, LSTM model, GRU model, and CNN model. The absolute error between real production and production predicted by CNN-LSTM is also shown.

(a)

(b)

(c)

(d)

(e)

(f)