如何根据钻进过程参数曲线对钻进地层进行可钻性实时分级是仪器钻进系统发展和应用的关键技术。据此,对统计模式识别进行分析,并针对钻进参数样本数据的特点,采用最短距离聚类准则,分别建立类及分类数已知与类及分类数未知条件下的地层判别分类及有效性检验方法;同时,在风化花岗岩金刚石钻进中采用钻进比功曲线对地层进行了分级研究。研究结果表明,在普通风化花岗岩地层中,该聚类分析方法分类性能良好、精度高、误判率低,所获得的分类结果与岩芯地质录孔的分类结果相吻合。该方法为岩土工程地层的自动判别分级提供了定量判别方法。
Classification of rock drillability in real time using the curves of drilling process parameters is a bottleneck of the development and application of the instrumented drilling system.According to analysis of statistical mode recognition and characteristics of drilling parameter data,a discriminant classification method and validity test method have been established respectively for known categories and categorical numbers as well as for unknown categories and categorical numbers by minimum-distance clustering criterion.The methods have been used to classify the weathered granite formation by diamond penetrating specific energy curve.The results show that the clustering approach possesses excellent performance in classification,high precision and low misjudgment ratio and well agrees with the geological logs of rock core in general weathered granite formations,which can provide a quantitative discriminant approach for automatic classification of geotechnical engineering.