深部实钻地层可钻性评测新方法

A New Prediction Method of Formation Drillability

目前地层可钻性的各种评测方法均属钻后分析方法 ,即只能对已钻地层的可钻性进行计算。本文从研究可钻性时间序列特征出发 ,应用分形几何方法 ,分析了可钻性时序的可预测性。在此基础上 ,应用人工神经网络技术 ,构造地层可钻性的预测模型 ,并编制了基于 MATL AB神经网络工具箱的可钻性预测软件系统。通过对塔里木油田 Q0 0 1A井地层可钻性的实际预测 ,结果表明地层可钻性时序具有可预测性 ,应用神经网络技术预测可钻性具有预测精度高、方便实用的特点 ,是一种可钻性预测的新方法 ,对指导科学钻井具有重要意义。

At present all kinds of formation drillability evaluating methods belong to after drilling analysis that is, these methods can be used only for calculating the drillability of the drilled stratum. In this paper the predictability of drillability time series was analyzed using fractal method based on the study of drillability time series characters. With this study, a drillability prediction software system is programmed based on th4 MATLAB neural network toolbox. It was proved to be effective when applied to predict the formation drilla bility of Q001A well in Talimu oil field. The practical result shows that the formation drillability is pre dictable and this method is more precise, easier application than the others. It is a new method of drillability prediction and is useful to manage the scientific well drilling.