Singular value decomposition median method-based noise reduction method for gas concentration data

Abstract

The invention discloses a singular value decomposition median method-based noise reduction method for gas concentration data and belongs to the technical field of signal processing. The method comprises the following steps of introducing the concentration data X of the noise-containing gas; detecting whether the concentration data X of the noise-containing gas contain single abnormal data and missing data or not; if yes, processing the single abnormal data by using a mobile mean line method and processing the missing data by using a three-time exponential smoothing method; if not, not processing the data; constructing a hankel matrix for the concentration data of the noise-containing gas; subjecting the hankel matrix to singular value decomposition (SVD); based on the singular value medianfiltering strategy, selecting an effective singular value; conducting the SVD inverse transformation to reconstruct a hankel matrix; and obtaining noise-reduced gas concentration data. According to the invention, by adopting the gas signal noise reduction method provided by the invention, the noise reduction experiment is carried out through actually measuring gas signals. The result shows that the method is good in noise reduction performance and can effectively improve the gas signal analysis precision.

Description

technical field [0001] The invention belongs to the technical field of signal processing, and in particular relates to a noise reduction filtering method for gas concentration data based on a singular value decomposition median method. Background technique [0002] Because the underground environment of coal mines is very harsh, the gas sensors arranged underground are often affected by various disturbances, such as smoke, high temperature, water vapor, etc., and are also affected by electromagnetic interference, resulting in the collected gas concentration data generally containing noise. If the gas concentration data containing noise is directly analyzed and processed, it will not only be unable to accurately predict the amount of gas gushing out and warn of danger in time, but also waste time and do a lot of useless work. Therefore, the gas concentration data must be denoised to restore its real development trend. [0003] At present, the main methods for noise reduction...

AI Technical Summary

Problems solved by technology

[0003] At present, the main methods for noise reduction of gas signals are wavelet transform noise reduction method and support vector regression machine noise reduction method. Since the collected gas signal data often has chaotic characteristics, its spectrum is scattered in the entire frequency space. At this time, wavelet transform is used It is difficult to strictly distinguish the useful signal from the noise spectrum and the kernel function of wavelet threshold and support vector regression machine cannot be selected accurately

Images