A Calculation Method for Bottom Hole Flowing Pressure Based on Radial Basis Function

Abstract

The traditional method using manometers to measure bottom hole flowing pressure (BHFP) is limited by workload and working condition. Meanwhile, the method of forecasting BHFP based on traditional multiphase conduit flow model requires large amount of basic data and the process of calculation is very complicated. Therefore, both of the above methods have their own limitations. This paper has established tripling feed-forward neural network of multiple input but single output by using RBF neural network method in which BHFP serves as output value and the factors influencing the BHFP serve as stimulated input value. After trained by using known samples, the tripling feed-forward neural network has the adequate capacity to be learned and spread. The measured data of QH oilfield and the comparison of data calculated by neural network technique with that calculated by BP method have demonstrated that neural network technique has the characteristics of fast convergence speed and higher accuracy rate in forecasting BHFP and can be applied to these oilfields whose BHFP cannot be directly measured or calculated.