改进结构的小波神经网络在油田开发指标预测中的应用

Oilfield development index prediction with structure improved wavelet neural network

由于油藏储层的非均质性和决定油田开发指标因素的不确定性,往往很难对油田开发指标进行准确的预测。针对小波神经网络模型及算法预测油田开发指标存在的不足,提出了改进结构的小波神经网络模型。改进结构的小波神经网络模型使输入指标同时在不同时间因子和尺度因子的小波基上展开。实例分析结果表明,改进结构的小波神经网络模型不仅继承了小波神经网络的优点,且具有比小波神经网络预测油田开发指标精度更高、训练速度更快的优势,其预测的平均精度达到97.02%,是预测油田开发指标的一种较实用的方法。

Because of the reservoir heterogeneity and the uncertainty of factors affecting oilfield development index,it is very difficult to predict oilfield development index accurately.However wavelet neural network also has some defects in model and algorithm.WNN model with improved structure was proposed.The model makes the input index spread in various wavelet bases of time factor and scale factor simultaneously.The analysis of examples indicate that WNN model with improved structure not only inherits merits of the WNN,but also has the advantages of higher accuracy and faster training speed for forecasting oilfield development index.It is a practical method for forecasting oilfield development index.