Trend prediction of pumping well conditions using temporal dynamometer cards

Most of the oil wells in China use sucker rod pump. Traditionally, manual patrol is used to find the working conditions of pumping wells. In the trend of oilfield automation and intelligence, this traditional working mode can not adapt to the increasingly complex working conditions of the oilfield. Therefore, artificial intelligence technology has been widely applied to the automatic diagnosis of pumping unit well conditions. This paper mainly studies the trend prediction of working conditions, and proposes a trend prediction model based on long and short time(LSTM) memory neural network and convolutional neural network. Based on the real oilfield data, this paper trained the trend prediction model. The average accuracy of the model in the test set reached 86%, which can meet the actual needs of the production site, provide scientific decision-making basis for dynamic optimization of pumping well measures, reduce the maintenance workload and improve the operation efficiency of the oilfield.