基于地震属性技术的井壁稳定随钻预测新方法

基于地震属性和测井数据之间存在非线性关系，提出一种综合利用地震和测井信息随钻预测井壁稳定性的新方法，即在钻前联合运用遗传算法和BP神经网络算法，从所提取的井旁原始地震属性中优选出对测井特征参数最敏感的地震属性组合，并利用已钻井资料建立起反映所研究区块不同地层中地震属性与测井数据之间关系的一系列神经网络映射模型，实钻时通过分析岩屑录井资料选择适当的映射模型，随钻预测出待钻井段的声波和密度测井数据，进一步预测出钻头下方地层的井壁稳定性。该预测方法具有良好的精确度和时效性，在塔西南地区的实际应用中取得了良好效果。

A new method of predicting borehole stability while drilling based on seismic attribute technology

A new method of predicting borehole stability while drilling was presented by using seismic attributes and well logs together. This method is based on the close nonlinear relationships between seismic attributes and well logs. Before drilling the optimal attribute combinations which are sensitive to log properties are selected from original seismic attributes by using genetic algorithm and BP neural network algorithm together. Then a series of mapping models which reflect relationships between seismic attributes and well logs of various formation intervals in given area are constructed through neural network. With analysis of cutting logging data, the proper mapping model can be employed to predict acoustic and density log curves of impending drilling formation while drilling. Based on the analysis, the borehole stability of formation under bit can be predicted. The prediction precision and real-time operation ability of the proposed method are satisfactory, which have been proved in the practical application in south-west area of Tarim Oilfield.