Research into prediction model of water content in crude oil of wellheat metering based on General Regression Neural Network

Abstract

Water content in crude oil is a very important data in oilfield production logging system. It is also an indispensable parameter for the research of its development prospect. During the process of exploitation, storage and transportation of oilfield, high accuracy measuring of water content in crude oil can optimize production parameters and improve oil recovery rate. The GRNN (General Regression Neural Network) has high advantages in approximation ability, classification capacity and learning speed. This paper measured some parameters which have effect on the measurement of the water content of crude oil using the multi-sensor technology and processed these parameters using the K-means clustering, and then proposed a prediction model for water content in crude oil based on GRNN. The result of the simulation in MATLAB shows that the prediction model proposed in this paper has several advantages such as stable prediction result and small error and so on.