BP神经网络在雷达测轨数据异常值处理中的应用

Application of Back Propagation Neural Network in Handling of Abnormal Value of Radar Measurement Data

在对雷达跟星校飞试验测量数据进行分析的基础上,利用BP神经网络方法建立了丢失帧数据和野值帧数据的估算模型。通过实例计算、分析、比较,证明BP神经网络方法估算精度要优于三次样条插值。

Based on analysis of measurement data of a calibration flight test by tracking a satellite,this paper establishes estimation models for data of lost frames and exceptional value frames with back propagation(BP) neural network estimation approach.It is proved through calculation, analysis and comparison of data drawn from samples that BP neural network estimation approach has a higher precision in estimating data of lost frames and outlier frames than the cubic spline interpolation approach.