Oil Field Crude Oil Production Level Prediction Method Based on AHP-PSO-BP

The crude oil production level is an important index to determine the degree of oil field development. There are many factors that affect the crude oil production level, and the amount of data is huge. The complicated relationships between the factor parameters bring great difficulties to the production level prediction. This paper presents a crude oil production level prediction model based on AHP-PSO-BP. When using the BP neural network to predict crude oil production level, we first utilize AHP to select parameters with higher weights to optimize the BP network input, and then we use PSO algorithm to optimize the weights and thresholds of the BP network. With these two optimizations, we have improved the stability and prediction accuracy of the BP network. Experimental results show that when using the AHP-PSO-BP prediction model to predict the crude oil production level, the determination coefficient R
²
we obtained from the model is 95.328%, which is better than the value we obtained from BP network model (82.356%), and GA-BP network model (70.509%). Besides, the AHP-PSO-BP prediction model convergence is faster, the prediction accuracy is higher and the network is more stable.