Radiation Signal Denoising Method of Loaded Coal-Rock Based on ICEEMDAN-PCK-Means-IP

When the coal-rock is loaded, the radiation field changes to produce weak radiation, which is easily perturbed by the environment. In this article, the preprocessing method combining the three-sigma principle and the compound interpolation method was studied, based on the analysis of the composition of the radiation collected directly from the loaded coal-rock. Combined with the improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) algorithm, the K-Means algorithm based on Pearson distance, the DBi index, and improved Pearson coefficient, a radiation denoising method based on ICEEMDAN-PCK-Means-IP was proposed. Then, the effect of each link in the method on the noise reduction of the simulated loaded coal-rock radiation signal was analyzed. Finally, the infrared radiation (IR) signal directly obtained from uniaxial loading experiments on coal-rock was used as an example to compare and verify the denoising effect of the proposed method on the radiation of loaded coal-rock. The results show that the radiation from the directly collected loaded coal-rock will be disturbed by multiple sources, which will drown out the features due to the deformation. The preprocessing method in this article can effectively eliminate the “bad points” in the original data, and its results are highly similar to the data without “bad points.” Moreover, each link of the ICEEMDAN-PCK-Means-IP method affects the denoising effect. Its denoising effect on simulated and real radiation is obviously better than that of other “EMD-like” denoising methods or its combined method, and can effectively extract the deformation-related signals from the original radiation.