A Calculation Method of Signal-to-Noise Ratio of Vibration Signal Based on Noise Adaptive Recognition

Abstract

A method for calculating the signal-to-noise ratio of vibration signals based on noise self-adaptive recognition proposed by the present invention includes: when there is global noise, the noise spectrum sequence Px[f≠F aim ] on all spectral peaks are retained, and all valleys are set to zero to obtain a new noise spectrum sequence P'x[f≠F aim ]; Calculate the corresponding noise range, calculate the effective noise range according to the noise range, and calculate the new noise spectrum sequence P'x[f≠F aim ] the noise level within the effective noise range; when it is local noise, calculate the corresponding noise range, calculate the effective noise range according to the noise range, and calculate the noise spectrum sequence Px[f≠F aim ] the noise level within the effective noise range; according to the effective frequency component Px[F aim ] and the noise level to calculate the signal-to-noise ratio SNR. The present invention enables the finally obtained signal-to-noise ratio to more directly reflect the obviousness of the signal compared with its noise, and is convenient to use the signal-to-noise ratio to judge whether various noise suppression, feature extraction, weak signal amplification, noise removal and other vibration signal processing algorithms are It is effective, and its effect is quantitatively judged and measured.

Description

technical field [0001] The invention relates to the technical field of signal noise, in particular to a method for calculating the signal-to-noise ratio of vibration signals based on noise adaptive recognition. Background technique [0002] For various types of equipment at present, vibration analysis steps such as vibration acquisition, vibration test, vibration detection, and data processing are all effective means to obtain equipment status, and vibration analysis is used for equipment design, optimization, monitoring, and noise reduction. Or vibration suppression, can bring greater reference value and guiding significance. [0003] In the process of vibration analysis, data processing often involves the identification, extraction and amplification of effective components in vibration signals. When measuring whether these algorithms are effective, we need an indicator that can indicate the proportion of active components in the current signal. The signal-to-noise ratio (...

AI Technical Summary

Problems solved by technology

This method can intuitively reflect the proportion of effective components in the mixed signal to a certain extent, but there are also some problems: the average amplitude of the spectrum in the entire frequency band is used as noise, and when the signal is relatively sparse, taking the average will lower the actual level of noise , causing the calculated SNR to be too large, the signal in this case is as figure 1 As shown; when the spectrum of the signal has the characteristics of a ski slope (envelope spectrum, low-pass filter spectrum, etc.), the effective signals in different frequency bands are due to the difference in the noise floor where they are located, and the unified noise calculation mode will result in different frequency bands. There is little difference in the calculation results when the signal amplitudes of the signals are similar, but there is a significant gap in the actual resolution (such as figure 2 , signal 1 and signal 2 are in different frequency bands, the SNR calculation results are not much different, but signal 2 is significantly more significant than signal 1)

[0011] For other types of calculation methods: obtaining the noise amplitude through the maximum component in the spectrum except the effective frequency will lead to unstable calculation of SNR results, and occasional high-energy frequency components will make the calculation invalid; and directly calculate the effective signal and noise signal (reference: Ma Shuling, a SNR calculation method and device, CN104579560B), it is difficult to apply in the field of vibration signal analysis, and the signal and noise cannot be separated to calculate its power in the time domain, and the result is different from the intuitive judgment result bigger

[0012] It can be seen that no matter which method is used above, there are defects. In certain occasions, the calculated results do not match the actual situation.

Images