基于最小矩理论的计算机裂缝自动识别方法

Automatic Identifying Algorithm of Fractures by Computer Based on Minimum Moment Theory

在确定地下裂缝的分布规律时 ,由于钻井的代价一般较高 ,所以应尽可能充分地利用获得的井壁成像信息。如果裂缝面与参考纸面垂直 ,则裂缝井壁图象在纸面内的投影处于一直线上。这时 ,可通过直线拟合确定裂缝位置。文献 [1]中提出了一种交替计算裂缝面相对参考纸面夹角和拟合上述直线的方法 ,但其在对于剔除到裂缝面距离较近的噪音点方面仍有提高余地。为此 ,提出了确定裂缝与参考纸面夹角的最小矩理论 ,进而充分利用直线拟合的特点 ,区分噪音点和实际样本点 ,并比较局部两个样本点连线方向与总体样本点拟合直线方向之间的差异 ,如果这一差异大于某一预定小正数 ,则剔除此连线的一个端点。反复运用这一方法至一定次数 ,去除噪音点 ,再就剩余的样本点进行直线拟合。与以往方法相比 。

Considering about the distribution of underground fractures, the image logging information should be made the best because well drilling is very expensive. If the fracture plane is vertical to the reference plane, the project of the common line of the fracture plane and the well wall on the paper plane would be a line segment and the fracture plane position could be determined through line fitting. In the reference paper[1], an algorithm was proposed, which alternately calculated the angle of the fracture plane to the reference plane and the fitted line. Yet how to get rid of the noise points near to the fracture plane is still a subject to be studied. To this end, it proposed a minimum moment theory determining the angle of the fracture plane to the reference paper plane. Further it took the advantages of line fitting and differentiates noise points from real sample points. Also it found out the difference between the direction of the linking line of two local sample points and that of the fitted line. If this difference is greater than a given small positive number, one of the end point of the linking line would be removed. Repeat this process several times to get rid of the noise points. Afterwards fit out a line with the remained sample points. Compared with previous methods, it could identify the noise points near to the fracture better.