论文部分内容阅读
本文提出了一种自动拾取地震初至的新的算法,它是通过分析分形维沿地震道的变化来检测信号的出现。研究发现“分割法”是最适合于计算分形维的方法。并且发现分形维的变化出现在由噪音向信号加噪音转换即初至的附近。这种维数变化的特征每道都各不相同,但总会出现可检测的变化。已经用信噪比不同的实际数据对该算法进行了试验,并将其结果与应用以前发表过的算法得到的结果进行了比较,通过适当地调谐其参数,证明基于分形的算法比其它所有算法更为精确,尤其是在噪音较强的情况下。试验表明,当噪音强度高达信号的平均振幅80％时,分形方法仍然有效。然而,基于分形的算法比起其它方法来要慢得多,因此,只宜将它用于低信噪比的数据组。
In this paper, we propose a new algorithm to automatically pick up the first arrivals of earthquakes by detecting the occurrence of signals by analyzing the changes of the fractal dimension along the seismic traces. The study found that “segmentation” is the most suitable method for calculating fractal dimensions. And found that the changes in fractal dimension appear in the noise from the signal plus noise conversion near the first arrival. The characteristics of this change in dimensions vary from one lane to another, but there are always detectable changes. The algorithm has been tested with actual data with different signal-to-noise ratios and the results are compared with the results obtained using previously published algorithms. By tuning their parameters appropriately, it proves that the fractal-based algorithm is better than all other algorithms More accurate, especially at noisy conditions. Experiments show that the fractal method is still effective when the noise intensity reaches as high as 80% of the average amplitude of the signal. However, the fractal-based algorithm is much slower than the other methods, so it should only be used for low signal-to-noise data sets.