<span>震前深层地下水位短临异常动态监测</span>-华南地震

文章信息/Info

Title:: Dynamic Monitoring of Short-term Impending Anomalies of Deep Groundwater Level before the Earthquake

Author(s):: TANG Shujun; LUO Yinfei; CHEN Jinxian; ZHAO Wenqiang; WANG Yonglei; Qinghai Environmental Geological Exploration Bureau，Qinghai 810000，China

Keywords:: Short-term water level; Dynamic monitoring; Support vector machine; Smoothing processing

摘要:: 针对传统震前地下水短临异常监测方法监测效率低、监测容易出现误差的问题，提出一种新的震前深层地下水位短临异常动态监测方法。通过PCA对地下水位做特征提取处理，搜索获得可逆的线性变换，根据自学习Paute准则对特征进行误差值检测，对符合删除条件的向量进行平滑处理，不符合的保留原值，再通过平滑处理对震前深层地下水位误差值滤波，从而完成误差值校对，随后利用支持向量机算法对震前深层地下水位误差值寻求一种最优质分类超平面，使两种样本的分类间隔最大化，以此构建震前深层地下水位短临异常监测模型，从而完成对震前深层地下水短临异常动态监测。实验证明，此方法在监测震前深层地下水短临异常变化时有着时效性高、监测精准的优点。

Abstract:: In order to solve the problems of low monitoring efficiency and prone to errors in the traditional monitoring method of the groundwater short-term and impending anomaly before the earthquake，a new dynamic monitoring method of short-term imminent anomaly of the groundwater level before the earthquake is proposed. The feature extraction of groundwater level is carried out by PCA，and the reversible linear transformation is obtained. According to the self-learning Paute criterion，the error value is detected on the features，the vectors that meet the deletion conditions are smoothed，and the original values that are not in conformity are retained. Then the error value of deep groundwater level before earthquake is filtered by smoothing processing，so as to complete the error value calibration. The support vector machine algorithm is used to seek a best-quality classified hyperplane for the error value of deep groundwater level before the earthquake，which maximizes the classification interval of the two samples，so as to construct the short-term and impending anomaly monitoring model of the deep groundwater level before the earthquake，and complete the dynamic monitoring of the short-term and impending anomaly of the deep groundwater before the earthquake. The experimental results show that the method has the advantages of high timeliness and accuracy in monitoring the short-term and impending anomalies of deep groundwater before earthquakes.