| 基于DBSCAN与网格搜索的雷电定位算法 |
| |
| 引用本文： | 梁丽,雷勇,张帅弛,李涛,庞文静,王志超.基于DBSCAN与网格搜索的雷电定位算法[J].应用气象学报,2019,30(3):267-278. |
| |
| 作者姓名： | 梁丽 雷勇 张帅弛 李涛 庞文静 王志超 |
| |
| 作者单位： | 1.中国气象局气象探测中心, 北京 100081 |
| |
| 基金项目： | 科技部重大仪器专项（2014YQ110787） |
| |
| 摘 要： | 定位精度是评价雷电定位网络的重要指标之一，定位算法直接影响雷电探测结果的精度。雷电监测系统探测数据误差不可避免，传统定位算法不具备抗误差干扰能力，迭代计算易发散，定位结果精度不高。为了满足实际应用需求，提出一种新的雷电定位算法DG-LLA（DBSCAN and grid-search lighting location algorithm），在定位计算中引入DBSCAN（density-based spatial clustering of applications with noise）方法与网格搜索方法。通过仿真与国家雷电监测网实际定位结果对比分析定位算法性能。结果表明:到达时间差（time difference of arrival，TDOA）法和Taylor级数展开法定位误差较大，仿真区域的均方根误差分别为982 m和668 m；定位中引入DBSCAN方法后，均方根误差明显减小为406 m，引入DBSCAN方法和网格搜索方法后，均方根误差减小为349 m；在相同回击数据条件下，算法DG-LLA与国家雷电监测网相比定位数量更多，回击数据的利用率从43.4%提升到51.5%，新增定位结果周围雷达回波特征较强，定位精度更高。
|
| 关 键 词： | 雷电定位 聚类分析 DBSCAN 网格搜索 |
| 收稿时间： | 2018/12/6 0:00:00 |
| 修稿时间： | 2019/3/14 0:00:00 |
Lightning Location Algorithm Based on DBSCAN and Grid Search |
| |
| Liang Li,Lei Yong,Zhang Shuaichi,Li Tao,Pang Wenjing and Wang Zhichao.Lightning Location Algorithm Based on DBSCAN and Grid Search[J].Quarterly Journal of Applied Meteorology,2019,30(3):267-278. |
| |
| Authors: | Liang Li Lei Yong Zhang Shuaichi Li Tao Pang Wenjing and Wang Zhichao |
| |
| Institution: | 1.Meteorological Observation Center of CMA, Beijing 1000812.Electronic and Information Engineering College, Nanjing University of Information Science & Technology, Nanjing 210044 |
| |
| Abstract: | Lightning location system can monitor the time and location of lightning in real time, supporting disaster early warning and post-disaster treatment in meteorology, power, aerospace, forest fire prevention and other fields. The location algorithm directly affects the accuracy of lightning detection results. Traditional location algorithms may often fall into the local optimum and needs a large amount of calculation. The practical application is limited by the computer capability and the anti-error interference ability is poor. A new lightning location algorithm DG-LLA (DBSCAN and Grid-Search Lighting Location Algorithm) is proposed. The algorithm is verified by a lightning accident example and regional simulation, and then by locating historical data detected in the national lightning monitoring network. The performance of the new algorithm is compared and further analyzed from three aspects:Lightning frequency temporal distribution, lightning spatial distribution and regional spatial distribution.
Simulation results of lightning example show that the location error of TDOA (time difference of arrival) method is the largest, reaching 1314 m. Taylor series expansion method is a classical iterative algorithm with an error of 881 m. The error of proposed DG-LLA algorithm is significantly reduced to 84 m. Examples of artificial lightning initiation show that the new algorithm DG-LLA is more accurate than the national lightning monitoring network, and the average location error is 32.2% lower than operational network. Lightning location algorithm based on adaptive DBSCAN and grid-search optimization can effectively identify noise data and enhance the ability of anti-error interference. Regional simulation result shows that TDOA method and Taylor series expansion method have large positioning errors, with the RMSE (root mean square error) of 982 m and 668 m, respectively. When DBSCAN is added to location, the RMSE is significantly reduced to 406 m. After DBSCAN and grid search are added, the RMSE is further reduced to 349 m. Lightning location algorithm based on adaptive DBSCAN and grid search optimization improves the local searching ability and global searching ability of space and overcomes shortcomings of traditional iterative algorithm, such as easy divergence and local optimum of optimization algorithm, and solve the lightning strike point stably and accurately, and it performances better than operational network. The utilization rate of return data increases from 43.4% to 51.5%. The radar echo around new locations has stronger characteristics and higher locating accuracy. It provides a new method for lightning location. |
| |
| Keywords: | lightning location cluster analysis DBSCAN grid search |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《应用气象学报》浏览原始摘要信息 |
| 点击此处可从《应用气象学报》下载免费的PDF全文 |
|
