1 / 2023-03-28 11:51:16

Radiation signal denoising method of loaded coal-rock based on improved ICEEMDAN algorithm

Improved ICEEMDAN algorithm,Loaded coal-rock,Radiation signal,K-means clustering algorithm,Signal noise reduction

When the coal-rock is loaded, the distribution of the radiation field changes accordingly, resulting in a weak radiation signal. However, such signals are susceptible to external interference, which makes it difficult to reveal the coupling between the radiation signal and the coal-rock state. In this regard, the paper analyzes the composition of various original radiation signals when the coal-rock is loaded, and combines the three Sigma principle with the compound interpolation method to propose the data preprocessing method of original radiation signals, so as to realize the correction of missing values and outliers in the original data. Then, combined with the improved CEEMDAN (ICEEMDAN) algorithm, the K-Means algorithm based on Pearson correlation coefficient (PCCs), DBi index and correlation coefficient, the radiation noise reduction method of loaded coal-rock based on the improved ICEEMDAN is proposed. It achieves noise reduction by decomposing, clustering and reconstructing the raw radiation signal collected directly. Finally, an example of IR radiation signal obtained from a uniaxial loading experiment is taken to verify the noise reduction effect of the proposed algorithm on the radiation signal. The experimental results show that the method based on improved ICEEMDAN has good noise reduction performance, and the improved noise ratio (INNR) of reconstructed radiation signals is up to 0.0655, which can realize the purpose of extracting signals related to the coal-rock deformation. The research results can provide theoretical and technical support for further research on coal and rock dynamic disaster prevention methods based on multi-information fusion.