考虑结构面间距的双因素分组方法研究

Double-factor clustering method considering spacing of discontinuities

同一期地质运动中岩体所产生的结构面间距一般认为服从负指数分布。目前方法多根据产状单一因素进行分组,分组完成后同一组结构面的间距分布可能非常不规则,针对该问题,结合目前结构面采集到的数据质量高且具有空间位置等特点,提出一种新的双因素分组方法 (a new double-factor clustering method,NDCM)。分组时可综合考虑结构面产状与间距信息,该方法分组初值由模糊K均值法(fuzzy K-means)获取,通过调整隶属度较低的结构面分组编码达到修正结构面间距,使其满足负指数分布的目的。为获得NDCM的分组精度,进行了数值模型计算,该模型中结构面分组情况已知,利用模糊K均值法和NDCM法分别对此模型进行结构面分组。对比结果表明,NDCM法分组精度明显优于模糊K均值分组方法;重叠率越高组数越多这种优势越明显;相比于模糊K均值法,NDCM分组精度平均可以提升0.06左右。

Spacing of discontinuities formed in the same tectonic stage is believed to obey a negative exponential distribution. Single factor clustering method according to orientation is most commonly used. However, the probability density function of spacing in one set of discontinuities clustered only based on orientation may be irregular, even exhibiting no rules. To overcome the deficiency, this study proposeds a new double-factor clustering method(NDCM), considering spacing distribution and orientation of discontinuities during clustering. The initial clustering result and membership values are obtained based on the fuzzy K-means. Through redistribution of the discontinuities whose membership values are low, this method adjusts the probability density function of spacing to make it more closely obey a negative exponential distribution. To obtain the clustering precision of NDCM, a numerical model calculation which contained two and three discontinuity sets has been made and the correct set code of every discontinuity were known. The model has been calculated by the two methods. Comparison results show that NDCM has an obvious advantage over fuzzy K means in clustering precision. About 6% promotion occurs in precision clustered by NDCM compared with fuzzy K means.