论文部分内容阅读
提出了一种用于地震早期预警的S波震相实时自动识别方法.该方法不对原始信号进行任何滤波处理,直接对三分向记录进行计算分析.首先根据P波前0.5s数据的卓越频率计算适用于该三分向记录的窗长,采用由偏斜角和水平能量与总能量比值的平方积作为确定S波识别区间的特征函数,将特征函数已有数据的5倍均值和5倍方差之和作为识别区间的触发阈值;然后采用VAR-AIC方法对两个水平分向识别区间的数据分别计算分析,对两个识别结果进行判断,最终确定S波初动时刻.经过对118个三分向记录的实际应用验证,通过自动识别结果与人机交互震相识别结果相比,本文方法对于S波相对P波尾波信噪比大于5dB的地震记录,其识别误差小于0.1s的概率高达89.39%.
A real-time S-wave automatic phase identification method is proposed for early warning of earthquake.The method does not perform any filtering on the original signal and directly calculates and analyzes tristimulus records.According to the excellent frequency The window length suitable for the tris record is calculated and the square product of the skew angle and the ratio of the horizontal energy to the total energy is used as the characteristic function for identifying the S-wave recognition interval. The five-times mean value of the existing data of the eigenfunction function is multiplied by 5 times And the sum of variance as the triggering threshold of recognition interval. Then, the data of two horizontal directional recognition intervals are calculated and analyzed respectively by VAR-AIC method, the two recognition results are judged, and the initial moment of S wave is finally determined. Compared with the recognition results of human-machine interaction phase, the method proposed in this paper is suitable for the seismograms with the S-wave relative P-wave coda signal-to-noise ratio greater than 5dB, and the recognition error is less than 0.1s The probability is as high as 89.39%.