EMD-ICA Filter And Its Application In GPS Multipath

GPS surveying,for example: the bridge deck monitoring,monitoring of the dam and the tall building monitoring,etc.This will make the GPS surveying results affected by multipath effect and high frequency random noise,and the results deviating from the true value.However,in GPS surveying,the baseline is usually short and generally using dual-frequency GPS receiver,so the ionosphere and troposphere error can be eliminated using differential method.But the GPS multipath error and random error caused by high frequency noise cannot be eliminated.GPS multipath error,however,there is no spatial correlation and the error due to high frequency random noise does not have regularity,thus eliminate these errors primarily aimed aim at proving the GPS surveying precision of short baseline surveying.This paper studied the high frequency noise elimination and the error of GPS multipath effect correction,the research contents and methods of this paper is as follows:1)Introduced the Independent Component Analysis and Empirical Mode Decomposition,analyze the EMD filter and ICA filter in detail.Based on the above,this paper proposed a filter method called EMD combined ICA Filter,EMD-ICA filter.2)Using vibration table experiment data to explore the performance of the dynamic GPS multipath effect,when GPS antenna in different displacement.Through the spectrum analysis of about 1hour measuring data,analyzing the relation between dynamic GPS multipath effect and static GPS multipath effect.3)Considering the strength of the Fast ICA separate the information ability depends on the accuracy of the given reference information,and the wide band distribution of GPS multipath effect.this paper proposed the EMD-ICA Filter based on mode relevance.Comparing this method with EMD filter based on mode relevance and ICA filter based on the spectral analysis of IMF by experiment.Experiments show that this method can achieve better effect.4)When the value of SNR is too low,this paper proposed a method of slope absolute value.The main ideal of slope absolute value derivative test is that using energy size of every IMF to distinguish useful information and noise.This method is similar to the method of EMD filter combined wavelet threshold,they are all aimed at extracting the low frequency information which drowning in high frequency.Comparing this method with EMD-ICA Filter based on mode relevance and EMD Filter based on mode relevance,the experiments show that this method can extract more useful information and achieve better effect.