Abstract The solid tide signal of VP inclinometer is limited by the complex monitoring environment, which contains significant environmental noise. To obtain the real solid tide curve, we propose an improved complete ensemble empirical modal decomposition adaptive noise(ICEEMDAN) based on grey relation analysis for VP inclinometer to suppress signal noise. The method firstly uses ICCEMDAN to decompose the noise-containing signal to obtain several intrinsic mode functions(IMFs), which are sequentially arranged and labeled. Then, based on these IMFs, the evaluation indexes of correlation coefficients, mutual information, R2, Adj-R2, MSE, SSE, RMSE, MAE, MAPE, we compute sample entropy to construct a reliability evaluation index matrix of the IMFs. Finally, grey relation analysis(GRA) is used to calculate the correlation coefficients and degrees between various evaluation indexes and different IMFs. We sort the IMFs based on the correlation degree, and linearly reconstruct the top ranked IMFs to suppress signal noise. Both the simulation and actual denoising experiments show that the GRA-ICEEMDAN model is better than the classical noise reduction models, such as Kalman filter, 70 order low-pass FIR filter, and Savitzky-Golay. The noise component and effective component can be distinguished significantly, and the reconstruction error and original signal loss after decomposition are very small, so it can be extended to the signal noise reduction of other instruments.
PANG Cong,SUN Haiyang,LIU Tianlong et al. A Random Noise Suppression Method for VP Inclinometer Signals Based on ICEEMDAN and Grey Relation Analysis[J]. jgg, 2024, 44(6): 654-660.
PANG Cong,SUN Haiyang,LIU Tianlong et al. A Random Noise Suppression Method for VP Inclinometer Signals Based on ICEEMDAN and Grey Relation Analysis[J]. jgg, 2024, 44(6): 654-660.