地震预警系统的P波震相自动识别方法研究

发布时间：2019-05-13 12:18

【摘要】：地震预警系统能在破坏性地震动到达目标地区前为人们提供地震警报信息,从而有效减少地震造成的人员伤亡,减轻地震灾害损失。目前,很多国家都在积极研究和应用地震预警系统,并已取得了诸多成果。初至震相的自动分析识别技术是地震预警系统中最基础但又极为关键的技术环节。准确可靠的震相自动识别结果能够最大程度地保证预警系统产出信息的可靠性,同时也能有效降低预警系统中的震相关联等后续工作的复杂度。本文在学习借鉴国内外震相自动识别相关研究方法的基础上,提出了一套算法简单、可用于实时处理的震相综合识别方法。具体的工作如下:(1)介绍了目前常见的自动震相识别方法,如STA/LTA方法、能量分析方法、偏振分析方法法、短时傅里叶变换方法、小波变换方法、人工神经网络法等,并按时域分析、频域分析、时频分析、综合分析方法四大类进行了综述。(2)阐述了长短时平均结合AIC方法、偏振分析方法、能量分析方法、FFT幅值比方法和峰度方法的原理。分别应用以上五种方法对所选取的福建测震台网观测记录进行P波震相到时的自动捡拾,统计并分析了五种方法的捡拾结果,总结各方法的优缺点。结果表明,单一方法只利用了信号的某一属性特征,其适用范围有一定限制,不能有效避免所有类型的误捡拾。(3)为避免误捡拾,尤其是干扰信号引起的误捡拾,本文充分利用地震信号的幅值特性、频率特性、偏振特性等属性,提出了一套算法简单、可用于实时处理的震相综合识别方法,并应用该方法对所选取的福建测震台网观测记录进行P波震相自动捡拾,统计并分析捡拾结果。结果表明,综合识别方法的抗干扰信号能力显著增强,能有效避免干扰信号引起的误捡拾,有助于提高P波捡拾结果的可靠性。
[Abstract]:The earthquake early warning system can provide people with earthquake warning information before the destructive ground motion reaches the target area, so as to effectively reduce the casualties caused by earthquakes and reduce the loss of earthquake disasters. At present, many countries are actively studying and applying earthquake early warning system, and have made a lot of achievements. The automatic analysis and identification technology of first arrival seismic phase is the most basic but critical technical link in earthquake early warning system. Accurate and reliable seismic phase automatic identification results can ensure the reliability of the output information of the early warning system to the greatest extent, and can also effectively reduce the complexity of the follow-up work such as earthquake correlation in the early warning system. On the basis of learning and drawing lessons from the related research methods of automatic seismic phase recognition at home and abroad, a set of comprehensive seismic phase recognition methods with simple algorithm and can be used for real-time processing is proposed in this paper. The specific work is as follows: (1) the common automatic seismic phase recognition methods, such as STA/LTA method, energy analysis method, polarization analysis method, short-time Fourier transform method, wavelet transform method, artificial neural network method and so on, are introduced. According to time domain analysis, frequency domain analysis, time-frequency analysis and comprehensive analysis methods, four categories of methods are reviewed. (2) the long-time average method combined with AIC method, polarization analysis method and energy analysis method are described. The principle of FFT amplitude ratio method and kurtosis method. The above five methods are used to automatically pick up the P wave seismic phase of the selected Fujian seismic network observation records, the picking results of the five methods are statistically and analyzed, and the advantages and disadvantages of each method are summarized. The results show that the single method only makes use of one attribute feature of the signal, and its scope of application is limited to a certain extent, which can not effectively avoid all types of false picking. (3) in order to avoid false picking, especially the false picking caused by interference signal, In this paper, a set of comprehensive seismic phase identification methods with simple algorithm and can be used for real-time processing is proposed by making full use of the amplitude characteristics, frequency characteristics and polarization characteristics of seismic signals. The P wave seismic phase is automatically picked up by the selected observation records of Fujian seismic network by using this method, and the picking results are counted and analyzed. The results show that the anti-interference signal ability of the comprehensive recognition method is significantly enhanced, which can effectively avoid the false picking up caused by the interference signal, and is helpful to improve the reliability of P wave picking up results.
【学位授予单位】：中国地震局工程力学研究所
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：P315.75

【参考文献】