【目的】地震初至的实时自动拾取是地震响应监测中的一个关键问题,也是实现地震预警、发出地震警报并启动应急反应系统的基础和前提。针对博物馆环境复杂、信号非平稳、低信噪比、非周期等特点,提出了一种地震初至自动识别的实时拾取算法。【方法】将博物馆可能存在的振动信号,与不同烈度、不同场地等条件下的地震信号进行对比分析,利用二者之间的差异对地震初至进行实时拾取。首先,将信号进行小波包分解重构,用非递归滤波器阈值降噪法对含噪音地震数据进行降噪处理,提高信号信噪比,然后使用改进后的长短时窗均值比法(STA/LTA)求取地震初至时刻。并通过四川泸定6.8级地震中在甘孜藏族博物馆采集到的振动信号进行拾取验证。【结果】结果分析表明,相较于传统的地震初至拾取无论是在初至到时的拾取还是对于噪音的压制都有较大水平的提高,实现了在地震波达到后的前0.2s内准确识别地震的到来,证明了本文方法的可靠性。

【Purposes】Real-time automatic pickup of earthquake first arrival is a key problem in  earthquake monitoring, and is also the basis and premise for realizing earthquake early warning, send⁃ ing out earthquake alarm, and starting emergency response system. On the basis of characteristics of  complex museum environment, low signal-to-noise ratio, non-stationary, and aperiodic signals, a  real-time picking algorithm for automatic recognition of earthquake first break was proposed. 【Meth⁃ ods】The vibration signals that may exist in the museum were compared with the seismic signals under  different intensities, different sites, and other conditions, and the difference between them is used to  pick up the first arrival of the earthquake in real time. First, the signal was decomposed and recon⁃ structed by wavelet packet, and the noisy seismic data was de-noised by the non-recursive filter  threshold de-noising method to improve the signal-to-noise ratio, and then the short-term average  over long-term average (STA/LTA) was used to calculate the first arrival time of the earthquake.  The vibration signals collected in the Ganzi Tibetan Museum during the Luding 6.8 earthquake in Si⁃ chuan were picked up and verified. 【Findings】 The result analysis shows that compared with tradi⁃ tional seismic first arrival pickup, both the pickup at the first arrival time and the suppression of noise  have a greater level of improvement, which realizes the accurate identification of the arrival of earth⁃ quake within the first 0.2 s after the arrival of seismic wave, and proves the reliability of the method in  this paper.