Research On Prediction Bottom Hole Pressure Of Under-balanced Drilling Base Of ANN

Under-balanced drilling technology is an important research topic in petroleum drilling field. It is strategic to discover and protect hydrocarbon reservoir better.The prediction of bottomhole pressure is very important during petroleum exploring and drilling. It can provide the support and guarantee for the safe and effective performance of under-balanced drilling. This paper is aimed at looking for a method using the theories and technology of Artificial Neural Network to predict the bottomhole pressure of under-balanced wells.Firstly,the basic theories of Artificial Neural Network are introduced, including neural cell model, the basic structure of Artificial Neural Network and the study methods. The essential of BP network with function approximation and RBF network and the performance of technology are analyzed. The algorithm of BP and RBF networks are also studied. This paper introduces the development situation, basic concepts and mathematical model of bottomhole pressure about under-balanced drilling technology. The dada acquisition and pressure survey methods during under-balanced drilling are researched, which provide the foundation for obtaining training samples for neural network. On the basis of above work, the neural network model of bottomhole pressure during under-balanced drilling is established, including determining network structure by using empirical method, adjusting the network weight using study algorithm. The neural network prediction software is programmed by using C Language. This software is applied during the real under-balanced drilling. The results show the prediction accuracy of the method above is relatively high. This opens a new way to predict bottomhole pressure of under-balanced wells using the artificial neural networks method.