基于支持向量机的水基钻井液配方优化设计  

Optimization of the formula of water base drilling fluid based on support vector machine

作　　者：李红冀 蒲晓林[1] 赵黎丽 王贵[1] 苏俊林[1] 

机构地区：[1]西南石油大学油气藏地质及开发工程国家重点实验室,四川成都610500 [2]新疆油田公司石西油田作业区,新疆维吾尔自治区克拉玛依834000

出　　处：《计算机与应用化学》2014年第7期787-791,共5页Computers and Applied Chemistry

基　　金：国家自然科学基金资助项目(51304169)

摘　　要：合理的钻井液配方是保证钻井工程能够顺利完成的必要条件,因此钻井液配方的优化设计已成为钻井工程中一个十分关注的问题。基于支持向量机建立通过钻井液性能参数反算处理剂加量的模型,即在给定钻井液性能参数的情况下,利用此模型反推各种处理的加量以达到优化钻井液配方的目的。根据某油田常用的强抑制性水基钻井液性能为要求,利用该模型计算出钻井液配方,将这些配方进行实验验证,实验结果表明,支持向量机计算出的7组配方中有5组符合该油田所使用的钻井液性能要求,模型仅经过一次训练正确率达到71.4%。该模型对钻井液的室内评价实验具有指导作用,能缩小实验时处理剂加量范围,减少实验工作量。It was necessary reasonable drilling fluid formulations is to ensure the successful drilling engineering, so optimizing drilling fluid formulations has became important. The essay was to establish a model which could predict the amount of drilling fluid additives by drilling fluid performance parameters based on support vector machine. Under the condition of specified performance parameters of a drilling fluid, performance requirement of a targeted drilling fluid could be attained by back-stepping the amount of drilling fluid additives with this model, which could predict the amount of additives required to add into drilling fluid needed in one certain formation rapidly, making it more sense in engineering and practice. The model was used for an water based drilling fluid case, SVM projected seven groups recommended formula, and then verify whether it is reasonable, The results showed that five of seven groups of additives met the requirement with SVM model, The model only after one training accuracy up to 71.4 %, this model has a guiding role in experimental laboratory, can narrow the scope of additives and reduce experimental workload.

关 键 词：支持向量机 钻井液 配方 优化 模型 

分 类 号：TQ015.9[化学工程] TP391.9[自动化与计算机技术—计算机应用技术]