一种基于BP网络的测井曲线自动提取方法

由于测井曲线图纸图像存在大量背景网格与曲线或曲线与曲线交织,致使曲线跟踪中断,需要手工连接,自动曲线提取非常困难.为此,提出一种利用BP神经网络进行曲线自动提取的方法.这种方法先提取曲线段的宽度、走向、分支特征,然后利用BP神经网络识别跟踪曲线交织处的分支,正确预测曲线的下一段.理论分析和实验表明,采用本方法,提高了曲线提取的正确率,抗干扰效果较好,几乎不要手工连接.该方法已经用于实际的石油测井曲线图纸图像的曲线自动提取中.

A method for automatic well log curves recognition using BP neural network

Large amount of background grids and other forms of curve interference in images of well log curves discontinue the well log tracking, so manual connection is needed. To solve the difficult problem of automatic recognition, this article proposes a method for well log curves recognition using BP Neural Network. The proposed method first automatically tracks and records widths, directions and the characteristics of branches, then judges the branches where well logs twist together using BP neural network and predicts the well log's next section accurately. Theoretical analysis and experiment demonstrate that this method improve the precision of automatic tracking and has a good result of anti-interference. Basically, manual connection is not needed. This method has already used in the automatic well log curve tracking of practical well log image.