| 用逐步代价最小决策法识别地震与爆破 |
| |
| 引用本文： | 张,博,边银菊,王婷婷.用逐步代价最小决策法识别地震与爆破[J].地震学报,2014,36(2):233-243. |
| |
| 作者姓名： | 张 博 边银菊 王婷婷 |
| |
| 作者单位： | 中国北京 100081 中国地震局地球物理研究所 |
| |
| 基金项目： | 地震行业科研专项基金(200808003)资助. |
| |
| 摘 要： | 在动态时间规整法的基础上， 建立了逐步代价最小决策法(SAMC). 该方法中的代价函数可以很好地反映特征归属， 对较差的特征具有一定的“容忍度”、 稳定性好， 还可用全程代价函数评判识别结果的可信度. 用SAMC方法对北京及其周边地区33次地震和29次爆破中提取的5个分类特征量进行识别， 识别率为90%； 从该5个特征量中选择较好的3个特征量进行识别， 识别率为92%； 在上述地区另选13次事件作为检验样本进行U检验， 5个分类特征量和3个分类特征量的识别率分别为92%和100%， 识别效果很好. 这表明SAMC是识别地震与爆破的有效方法.
|
| 关 键 词： | 动态时间规整 逐步代价最小决策法 地震和爆破的识别 识别判据 |
| 收稿时间： | 2012-12-17 |
Discrimination of earthquakes and explosions by SAMC decision method |
| |
| Affiliation: | Institute of Geophysics, China Earthquake Administration, Beijing 100081, China |
| |
| Abstract: | Based on dynamic time warping (DTW) algorithm, this paper proposes a novel recognition algorithm, stepwise accumulating minimal cost (SAMC) decision method. By this method, each cost function can well reflect the tendency of event’s features; moreover, SAMC is unsusceptible to the quality of features. The absolute value of overall cost function can be also served as the reliability of the results. Five features were extracted for recognition from 62 events of earthquakes (33) and explosions (29) which occurred in Beijing and its peripheral regions. The result respectively reached 90% recognition rate for five features and 92% for three features which are better features from the five ones. In another testing by U-test, another 13 events (eight earthquakes and five explosions) were randomly chosen from the same area, and the recognition rate was 92% for five features and 100% for three ones. These suggest that SAMC method can be useful to discriminate earthquakes and explosions effectively. |
| |
| Keywords: | |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《地震学报》浏览原始摘要信息 |
|
点击此处可从《地震学报》下载全文 |
|
