Prediction Of Gas Wells In Wells By Radial Basis Function Neural Network

In the later period of gas field development, the effects that effusion to gas production is ignored with increasing number of gas wells. Predicting the fluid status accurately in exploitation is more significant to alleviate the fluid in gas wells and increase the recovery rate of gas reservoir.Scholars have launched a prediction of gas effusion and maked a number of predictive models, but the results of calculations were quite difference, There is no qualitative criteria in model selecting and applying, it caused great inconvenience to actual applicationa new type model of forecasting gas effusion based on previous researchs is presented, mainly on the following:Firstly, the article investigated the carrying liquid gas model (The gas well is divided into straight wells and deviated wells) and summed up under different conditions apply to carrying liquid gas model, then the existing prediction models were analyzed. The factors of gas effusion was produced with some examples.Secondly, the article analysised the factors of affecting gas and the model of gas effusion fluid model, a fluid prediction model based on RBF neural network was established. By contrast with the BP neural network prediction and improved BP neural network prediction results, the RBF model has move advantages in calculation speed and precision.Finally, Y1, Y2wells were predicted respectively by the RBF effusion. By contrast with many production instances, the reliability of the RBF model was testified. This model for the gas production decisions provide an effective theoretical basis and guidance for the production.