Prediction of oil well production based on the time series model of optimized recursive neural network

The Arps decline model is one of the most commonly used oil field production prediction methods. However, the predictions of Arps decline model are not ideal. In this work, we designed a method that uses long-short term memory (LSTM) network and gate recurrent unit (GRU) to predict oil production. These models use multiple data features for oil well production prediction. The designed method considers the time series and has the ability to deal with nonlinear problems. On the basis of case study of the two numerical models performed on the dataset collected from actual oilfield in China and India, the LSTM and GRU are used to predict the oil production prediction. The final results show that LSTM and GRU have their respective advantages under different input parameters, providing an effective method for dynamic prediction of oil well production. These methods have the ability to be a quick and real-time auxiliary basis for oil well production planning, and have practical value.