基于支持向量机的地基单站GPS反演大气剖面

Retrieving Atmospheric Profiles Based on Support Vector Machine and Singular Ground-based GPS Receiver

提出基于支持向量机的地基单站GPS遥感大气剖面的反演方法,主要包括经典支持向量机、最小二乘支持向量机、相关向量机3种方法,利用青岛地区的历史数据进行了仿真反演对比研究,并与神经网络反演方法进行比较,结果表明支持向量机能够有效地应用于地基单站GPS大气遥感领域。

The techniques of retrieving atmospheric profiles are proposed herein based on support vector machine （SVM） and singular ground-based GPS, which include classical SVM, least squares SVM, and relevance vector machine （RVM）. The new methods are compared with BP-ANN networks method by simulation based on the historical radiosonde data in Qingdao, showing that the methods based on SVM can be applied to surface-based microwave atmospheric remote sensing effectively.