基于小波网络的遥测视速度异常数据剔除方法

为了提高遥测视速度数据的质量，保证制导工具系统误差分离精度，提出了一种小波网络预测与3σ准则相结合的遥测视速度异常数据检测和剔除方法。将遥测视速度数据加时间窗获得小波网络的训练样本，根据3σ准则对网络输出进行检测判别异常数据，对异常数据用小波网络的一步预测输出代替。试验数据处理结果证明了利用小波网络进行遥测异常数据剔除的可行性。由于小波网络的拟合作用，可以提高输出的信噪比。

Eliminating Abnormal Data in Apparent Velocity of Telemetry based on Wavelet Network

A new method for eliminating the abnormal data in apparent velocity of telemetry based on wavelet network combined with 3σ - rule is proposed in this paper, which can im- prove the quality of the data and ensure the separation precision of the guidance instrument systematic error. The training samples of wavelet network are achieved by adding time window to the apparent velocity data, and the abnormal data can be recognized according to 3 σ- rule and be replaced by one - step predictive output of the network. The results of test data process- ing show the feasibility of this method, and for the fitting effect of wavelet network, the SNR can be increased.