???????????????????????????

钻井过程中，在井壁上形成的动态滤饼不仅是安全、快速钻进的重要因素，而且动态滤饼的质量与油气层受伤害的程度，测井解释的准确性、油气井的产能等都有重大的影响，因此，动态滤饼一直为国内外众多石油界专家、工程技术人员所重视，对它的研究也未曾中断。渗透率及厚度是动态滤饼的两个重要物性参数，影响因素很多，错综复杂。准确测定动态滤饼的渗透率和厚度是必要的，但也是不易的。以往的实验测定方法，需要人力、物力与时间，费时费事。笔者利用人工神经网络原理，建立了动态滤饼的渗透率及厚度预测的神经网络模型。从ＳＷ－Ⅱ型高温高压动态污染仪测得的实验数据，通过ＢＰ算法，对动态滤饼的渗透率及厚度作出参数预测，其精度在９０％以上，表明该模型的正确性。利用此模型作动态滤饼的参数预测具有省时、省事，在计算机上易于实现，操作简单、快捷的特点，在现场生产及室内研究中，此模型具有广泛的应用前景。

PREDICTION OF THE PETROPHYSICAL PARAMETERS OF DYNAMIC FILTER CAKE BY ARTIFICIAL NERVE NETWORK

??In the process of drilling, the dynamic filter cake formed on the borehole wall is an important factor of influencing safe and rapid drilling,and the quality of the dynamic filter cake is closely related to the degree of reservoir damage,the accuracy of log interpretation and the productivity of oil and gas wells,therefore many experts,engineers and technicians in petroleum circles have paied great attention to the dynamic filter cake and studied it uninterruptedly.The permeability and thickness are two important petrophysical parameters of the dynamic filter cake and many factors can influence them,being intricate.It is necessary to accurately measure the permeability and thickness of the dynamic filter cake,but it is difficulty also.The experimental measure methods used in the past took both time and effort,needing a lot of manpower,material resources and time.Based on the principle of artificial nerve network,a nerve network model of predicting the permeability and thickness of the dynamic filter cake is set up.The models were trained by the experimental data measured by Type-SW-?? high temperature and high pressure dynamic pollutant measuring instrument through a BP algorithm.The accuracy of predicting the permeability and thickness of the dynamic filter cake is more than 90% by use of the model,so that the model is correct,being of the properties of saving time and trouble,realizing in computer easily and operating simply and rapidly and being of a vast range of prospects for application in production and research.