Research And Implementation Of The Oil-Gas Production Forecasting Based On Wavelet Analysis And Neural Networks

In the process of developing the oil fields, oil production forecast is an important element in development. Accurately predicting the oil production depends on the reliable data, and the purpose of the production forecast is to provide the basic decision-making activities. The development of the oil can not be reformed by experiment, and man can not made it repeatedly underground. So the accurate and realistic forecast of oil production is an important guide for the development and production.Wavelet analysis and artificial neural network develop rapidly in recent years, and they are in many areas with potential application. Wavelet analysis has the ability of time threshold, frequency threshold and multi-resolution analysis. At the same time, artificial neural network has a strong ability of nonlinear function approximation, adaptive learning ability, fault tolerance and parallel information processing capacity. In this paper, both the advantages of wavelet analysis and neural network forecasting method of formation of oil and gas in-depth research, to broaden the wavelet analysis and neural network prediction in the oil and gas application, improved the prediction of the oil and gas applications, in order to more effective, rapid and convenient analysis and forecasting of oil and gas production to provide new ideas and methods.This thesis is the background of scientific and technological project to reduce production costs and increase productivity. It focuses on a series of underground oil prediction research. The goal of the research is to establish a practical prediction model output and to find out the rules governing the development of oil production.First of all, this thesis describes the analysis and neural network technology as well as their basic idea in the application. And then, it deeply introduces the analysis of the oil production forecast method and model based on wavelet analysis and artificial neural networks. This model consists of two parts: one is decomposing and reconstructing the data by wavelet analysis in order to eliminate the noise from the data; the other is training neural networks, and then uses the trained network model to forecast. Finally, using the specific example is to test the validity of the model.The results show that the model of the oil production forecast in this thesis is better and could overcome the shortcomings of the data. At the same time, it expands the oil production theory. The model can be applied to other similar production and provides a good foundation to China's oil production forecast.