一种新的雷达数据剔野方法

摘要

基于雷达测量数据是一种特殊的时间序列这一特性,提出了一种针对雷达数据的序列剔野方法.先求得序列的边缘点,计算边缘权重因子,再在特征空间中计算异常因子来识别野值.从子列的角度检测雷达测量数据的野值,弥补了单点检测的局限性.采用某型雷达在某次任务中斜距R的一段测量数据进行实验,并将野值识别情况与用传统方法所得结果进行比较,结果表明,这种针对雷达测量数据的序列剔野方法较传统方法更能有效地剔除雷达数据中的野值.%A novel serial outlier-elimination method designed against radar measurement data is proposed based on the fact that the radar measurement data is time series. Following acquisition of margin points and computation of margin weights, outliers are identified by outlier-factors computed in the characteristic space. The outliers are detected in a series-oriented way so it outperforms conventional point-oriented methods. The range R of the measurement data in a tracking task with a radar is used in a test of the method. Results of comparison with those of conventional methods show that the serial outlier elimination method removes outliers in measurement data more effectively.