Gravity signal extracting method based on independent component analysis with multiple reference signals

The measurement data of marine gravity contains substantial noises, the low frequency part of which have similar frequencies with gravity signal, and it's difficult to inhibit the noise of the measurement data and extract the gravity signal by using classical algorithms. In order to effectively eliminate the noise of the measurement gravity and improve the accuracy, a novel method of extracting the gravity signals is proposed based on the theory of independent component analysis (ICA) with multiple reference signals. The measurement gravity signal is decomposed into intrinsic mode functions(IMFs) by empirical mode decomposition(EMD) algorithm, and processed by Kalman filter and wavelet translation at the same time. The signal reconstructed by part of IMFs and the result of the Kalman filter and wavelet translation are used as the reference signals of the ICA algorithm. The gravity signal is estimated by the FastICA algorithm based on the negative entropy. The de-noising experiment has been simulated based on the real gravity data. The results of theoretical analysis and simulation experiments indicate that the proposed method can effectively eliminate the noise of the measurement gravity and recovery the wave form of gravity signal, and the accuracy of the signal can be approximately increased 30% compared with classical algorithms.