WOA-BP Based Predicting Daily Production Method of Single Wells in Oilfield

The daily production of a single well in an oil field can reflect the changes in oil and water in the reservoir and it is an important basis for formulating single well stimulation measures. However, the factors that affect the daily production of a single well are complex, and there is currently no standard calculation method. In recent years, BP neural networks have been widely used in yield prediction, but they have problems such as slow convergence speed and easy to fall into local optima. In response to the above issues, this paper proposes a backpropagation neural network model WOA-BP based on the whale optimization algorithm. Firstly, the Spearman and Pearson correlation coefficient methods are used to screen feature attributes related to oil production as input parameters of the neural network, with oil production as output parameter; Then, the Whale Optimization Algorithm (WOA) is used to optimize the initial parameters such as learning rate, weight and bias, as well as the number of hidden layer neurons in the BP neural network; Finally, based on the optimized initial network parameters, a single well daily production prediction model is constructed. Train and evaluate the established model using real oilfield data, and compare it with the prediction models of BP, GA-BP, and PSO-BP. The experimental results show that the WOA-BP model has good prediction performance, with a coefficient of determination (R2) of 0.9633 and a mean square error (MSE) of 0.0017. It can effectively predict the daily oil production of a single well and aid with predicting the production of oilfield blocks.