Abstract: Borehole stability is generally studied during drilling or after drilling.By analyzing drilling log,logging data and core data in detail,the leading mechanism of borehole instability is acquired,the safe mud weight program and mud system are selected.This approach is easy and practical for the drilled well,but the result may be impractical for the well to be drilled because of the change of geologic structure.In addition,the stability of wildcat well is generally determined by experience,so the disaster at complex formations is unable to avoid.In order to prevent from borehole instability problems,a new method for predicting borehole stability before drilling is studied closely in this paper.According to neural network theory,a new model for predicting borehole stability by seismic data is established,and is successful in field application.

Key words: borehole stability, predication, seismic data, neural network

摘要： 国内外对井壁稳定研究一般都在钻井过程中或钻井结束后进行,通过钻井、测井、录井和岩心资料的综合分析,确定井壁失稳主导机理,选择安全泥浆密度与泥浆体系,该方法相对容易且符合研究井的实际工程要求,但得到结果的实用性易受地质构造影响.此外,野猫井的井壁稳定性一般凭经验判断,所以复杂地层的井壁失稳事故在所难免.为此,本文在神经网络的基础上,研究了利用地震资料钻前预测井壁稳定性的方法,建立了井壁稳定性钻前预测模型,该模型在现场得到较为成功的应用.至目前国内外在这一领域的研究还是空白.

关键词: 井壁稳定, 预测, 层速度, 神经网络