抑制地磁秒数据干扰的几种滤波算法比较

Comparison of several filtering algorithms for interference suppression of geomagnetic second data

为了解决地磁台站受干扰的问题,针对不同的干扰特征及实际勘测中的干扰数据,对比了FIR低通滤波、巴特沃斯滤波、中值滤波等5种易于实现的传统滤波算法和基于自主学习的模糊神经网络的滤波算法。以受多种干扰源干扰的九峰野外勘测数据为例进行比较可知传统滤波算法针对性较强,其中中值滤波效果最好。基于人工智能的模糊神经网络对不确定的因素具有理想的鲁棒性且中后期效果优于中值滤波,实现了实时接收数据和调整滤波参数,从而达到了最优的滤波效果,给地磁台站制定干扰抑制系统提供了新的方向。

In order to solve the interference problem of geomagnetic station,the traditional filter algorithms such as FIR low-pass filter,Butterworth filter,median filter and the filter algorithm based on self-learning fuzzy neural network are compared according to different interference characteristics and the interference data in actual survey.Taking the nine peak field survey data disturbed by many kinds of interference sources as an example,the results show that the traditional filtering algorithm has strong pertinence,among which the median filtering effect is the best,the fuzzy neural network based on artificial intelligence has ideal robustness to uncertain factors and the effect in the middle and later period is better than the median filtering effect,which realizes the real-time adjustment of filtering parameters from the received data.It provides a new direction for the geomagnetic station to develop the interference suppression system.