Neural Network Production Split Model Based on Kriging Interpolation Method and Grey Correlation Analysis Method

Abstract

Oil well production split is an important prerequisite for the efficient development of multi-layered reservoir. Conventional split methods do not consider the synergistic influence of dynamic and static factors related to the oil layer with certain errors. Based on the related dynamic and static parameters of production wells and production data, a neural network split model considering Kriging interpolation and grey correlation method was established. Before the model is established, the involved data is first sorted, and Kriging interpolation is used to complete the data at the appropriate time. Then, the grey correlation analysis method is used to calculate the correlation of the dynamic and static parameters related to reservoir production, and the main parameters of it are extracted. The main parameters are treated as input and substituted into the double hidden layer neural network for modelling, and the resulting model is used to split the production of each layer of the oil well. This method is applied to the production well of the N block of the Y Oilfield. Compared with the conventional method, the accuracy is significantly improved, which is in line with the actual reservoir characteristics.