Research On Dual-domain Gas Signal Noise Reduction Algorithm Based On CEEMDAN Fusion SVD Ratio Method

The gas data under the mine is often interfered by various factors in the process of collection,which leads to the collection of gas data often containing a large amount of noise.If the unprocessed gas data is directly used for feature extraction or prediction,a large number of errors and errors will result.Judging,it is impossible to accurately predict hidden dangers such as gas outburst,and it is impossible to avoid the occurrence of major malignant accidents in coal mines in time.Therefore,gas data noise reduction is very important for gas prevention and control work.Gas noise data with "large number" data and missing data was preprocessed.First,use the PauTu criterion that does not need to look up the table and process the "large number" data for detection and culling;then,use the maximum expected interpolation method to predict the data missing from the filling process.A dual domain gas signal denoising algorithm based on CEEMDAN fusion SVD ratio method is proposed.In the frequency domain,the CEEMDAN method is used to decompose the gas signal from high frequency to low frequency into several IMF components,and the noise-containing high-frequency IMF component is screened out using the permutation entropy algorithm;in the air domain,the SVD decomposition method is used to contain noise.The high-frequency IMF components are decomposed into the "signal subspace" and the "noise subspace" respectively,and the effective order of the singular value matrix is selected by the improved SVD ratio method to delete the "noise subspace",thereby performing noise reduction.The effectiveness of the dual-domain noise reduction algorithm is verified by the gas signal denoising experiment,which improves the SNR and reduces the root mean square error.Finally,this paper designs and implements a gas signal denoising system,providing an intuitive interface display and operational platform for signal analysts.