CN-122016181-A - Safety valve state abnormity monitoring method and system based on edge calculation

CN122016181ACN 122016181 ACN122016181 ACN 122016181ACN-122016181-A

Abstract

The invention relates to the technical field of anomaly monitoring, in particular to a safety valve state anomaly monitoring method and system based on edge calculation. The method comprises the steps of obtaining and conducting layered denoising to obtain a clean acoustic signal, conducting frequency domain energy value calculation on the clean acoustic signal to obtain energy density data, then calculating energy difference to obtain energy change data, conducting risk analysis and frequency domain feature decomposition on the basis of the energy change data to obtain leakage feature components, calculating energy deviation and conducting threshold comparison to obtain a deviation abnormal sequence, conducting tracking statistics on the deviation abnormal sequence to obtain abnormal duration and conducting interference rejection to obtain refining abnormal duration, conducting confidence calculation and threshold comparison on the basis of the refining abnormal duration to obtain a priority alarm sequence, obtaining historical data of a safety valve, conducting threshold correction to obtain a corrected alarm threshold, and conducting threshold comparison to obtain an instant alarm signal. The invention realizes the accurate identification and alarm of the leakage signal of the safety valve.

Inventors

Yuan Wufei
Lin Zhuangdun
HUANG YU
ZHENG YOUYAN
DENG CONG

Assignees

广东省特种设备检测研究院揭阳检测院
广东省特种设备检测研究院茂名检测院

Dates

Publication Date: 20260512
Application Date: 20260127

Claims (9)

1. The safety valve state abnormality monitoring method based on edge calculation is characterized by comprising the following steps of: acquiring an original acoustic signal, and carrying out layered denoising on the original acoustic signal to obtain a clean acoustic signal; performing frequency domain energy value calculation on the clean acoustic signal to obtain energy density data, and calculating energy difference according to the energy density data to obtain energy change data; performing risk analysis based on the energy change data to obtain leakage risk frequency, and performing frequency domain feature decomposition according to the leakage risk frequency to obtain leakage feature components; calculating energy deviation according to the leakage characteristic component to obtain an energy deviation sequence, and comparing thresholds based on the energy deviation sequence to obtain a deviation abnormal sequence; Tracking and counting according to the deviation abnormal sequence to obtain abnormal duration, and eliminating interference based on the abnormal duration to obtain refined abnormal duration; Performing confidence calculation based on the refining abnormal time length to obtain accumulated confidence, and performing threshold comparison based on the accumulated confidence to obtain a priority alarm sequence; And acquiring historical data of the safety valve, carrying out threshold correction based on the historical data of the safety valve to obtain a corrected alarm threshold value, and comparing based on the corrected alarm threshold value and the priority alarm sequence to obtain an instant alarm signal.
2. The method for monitoring abnormal state of a safety valve based on edge calculation according to claim 1, wherein the obtaining an original acoustic signal, performing layered denoising on the original acoustic signal to obtain a clean acoustic signal, comprises: Acquiring an original acoustic signal, and carrying out wavelet decomposition on the original acoustic signal to obtain a signal coefficient set; and carrying out soft threshold screening according to the signal coefficient set to obtain screening signal coefficients, and carrying out reverse combination based on the screening signal coefficients to obtain clean acoustic signals.
3. The edge calculation-based safety valve state anomaly monitoring method of claim 1, wherein the performing frequency domain energy value calculation on the clean acoustic signal to obtain energy density data, calculating energy differences according to the energy density data, and obtaining energy variation data, comprises: Performing short-time Fourier transform based on the clean acoustic signals to obtain frequency domain amplitude data, and performing energy value calculation according to the frequency domain amplitude data to obtain energy density data; generating a map based on the energy density data and carrying out logarithmic scale compression to obtain energy spectrum data; And calculating the energy difference of the adjacent frames according to the energy spectrogram data to obtain energy change data.
4. The method for monitoring abnormal state of safety valve based on edge calculation according to claim 1, wherein the performing risk analysis based on the energy variation data to obtain a leakage risk frequency, performing frequency domain feature decomposition according to the leakage risk frequency to obtain a leakage feature component, comprises: comparing the energy change data with a preset change threshold value and marking to obtain a risk time map, and comparing the risk time map with a preset risk model to obtain leakage risk frequency; And carrying out frequency domain feature extraction according to the leakage risk frequency to obtain a preliminary feature component, and carrying out weighted screening on the preliminary feature component to obtain a leakage feature component.
5. The method for monitoring abnormal state of safety valve based on edge calculation according to claim 1, wherein the calculating energy deviation according to the leakage characteristic component to obtain an energy deviation sequence, and comparing threshold values based on the energy deviation sequence to obtain a deviation abnormal sequence comprises: comparing the leakage characteristic component with a preset low-temperature environmental noise baseline, and obtaining an energy deviation value, and calculating a deviation statistical value in a preset sliding time window according to the energy deviation value to obtain deviation window data; And comparing the deviation window data with a preset deviation threshold value, and marking if the deviation window data exceeds the preset deviation threshold value to obtain a deviation abnormal sequence.
6. The method for monitoring the abnormal state of the safety valve based on the edge calculation according to claim 1, wherein the steps of tracking and counting according to the deviation abnormal sequence to obtain an abnormal duration, performing interference rejection based on the abnormal duration to obtain a refined abnormal duration, and include: carrying out adjacent window combination according to the deviation abnormal sequence to obtain a combined abnormal sequence, and carrying out time tracking statistics according to the combined abnormal sequence to obtain an abnormal duration; comparing the abnormal duration with a preset interference duration, and if the abnormal duration is lower than the preset interference duration, judging that the abnormal duration is unstable fluctuation and eliminating the abnormal duration to obtain the refining abnormal duration.
7. The safety valve state anomaly monitoring method based on edge calculation according to claim 1, wherein the calculating of the confidence coefficient based on the refined anomaly duration to obtain an accumulated confidence coefficient, and the comparing of the threshold value based on the accumulated confidence coefficient to obtain a priority alarm sequence comprise: Based on the refining abnormal time length and the leakage characteristic component, carrying out joint construction to obtain a leakage change sequence; performing confidence calculation according to the leakage change sequence to obtain a confidence sequence, and performing time integration according to the confidence sequence to obtain accumulated confidence data; and comparing the accumulated confidence coefficient data with a preset confidence coefficient threshold value, and marking if the accumulated confidence coefficient data exceeds the preset confidence coefficient threshold value to obtain a priority alarm sequence.
8. The method for monitoring abnormal state of safety valve based on edge calculation according to claim 1, wherein the step of obtaining the history data of the safety valve, performing threshold correction based on the history data of the safety valve to obtain a corrected alarm threshold, and comparing based on the corrected alarm threshold and the priority alarm sequence to obtain an instant alarm signal comprises: Acquiring historical data of a safety valve, performing correction calculation based on the historical data of the safety valve to obtain a confidence coefficient correction factor, and performing threshold correction according to the confidence coefficient correction factor to obtain a correction alarm threshold; and comparing the priority alarm sequence with the correction alarm threshold, and if the deviation value of the correction alarm threshold and the priority alarm sequence is smaller than the preset alarm tolerance, generating alarm information to obtain an instant alarm signal.
9. A safety valve condition anomaly monitoring system based on edge calculations, comprising: The signal preprocessing module is used for acquiring an original acoustic signal, and carrying out layered denoising on the original acoustic signal to obtain a clean acoustic signal; The energy analysis module is used for carrying out frequency domain energy value calculation on the clean acoustic signals to obtain energy density data, and calculating energy difference according to the energy density data to obtain energy change data; The feature extraction module is used for carrying out risk analysis based on the energy change data to obtain leakage risk frequency, and carrying out frequency domain feature decomposition according to the leakage risk frequency to obtain leakage feature components; the deviation detection module is used for calculating energy deviation according to the leakage characteristic component to obtain an energy deviation sequence, and performing threshold comparison based on the energy deviation sequence to obtain a deviation abnormal sequence; the duration screening module is used for carrying out tracking statistics according to the deviation abnormal sequence to obtain abnormal duration, and carrying out interference rejection based on the abnormal duration to obtain refined abnormal duration; The confidence evaluation module is used for carrying out confidence calculation based on the refining abnormal time length to obtain accumulated confidence, and carrying out threshold comparison based on the accumulated confidence to obtain a priority alarm sequence; The threshold correction module is used for acquiring historical data of the safety valve, correcting the threshold based on the historical data of the safety valve to obtain a corrected alarm threshold, and comparing the corrected alarm threshold with the priority alarm sequence to obtain an instant alarm signal.

Description

Safety valve state abnormity monitoring method and system based on edge calculation Technical Field The invention relates to the technical field of anomaly monitoring, in particular to a safety valve state anomaly monitoring method and system based on edge calculation. Background In the field of industrial safety, along with the development of edge computing technology, how to effectively capture abnormal signals of equipment and improve the real-time level of state monitoring has become a core problem to be solved in the development of the current safety valve monitoring technology. In the prior art, a safety valve monitoring strategy is mainly judged according to a preset fixed acoustic threshold, a frequency range or an energy upper limit, and an alarm mechanism is triggered when a certain index is detected to exceed the preset range by monitoring real-time acoustic parameters of the safety valve, and an early warning signal is sent or an emergency treatment program is started to ensure production safety. However, the noise environment of the industrial field is increasingly complex, and the fixed strategy is difficult to adapt to the actual characteristics of weak leakage signals under different working conditions, so that the false alarm rate is increased or the false alarm risk is increased. In addition, the traditional scheme often depends on a centralized data processing mode, and a large amount of original data needs to be transmitted to a far end for analysis, so that communication burden is increased, and early warning is not timely possibly caused by network delay. In summary, the prior art lacks the capability of identifying weak leakage signals and background noise of the safety valve, so that the problem of reduced accuracy of potential risk identification is caused. Disclosure of Invention The invention provides a safety valve state abnormity monitoring method and system based on edge calculation, which can dynamically monitor the abnormal state of a safety valve and give an alarm in time, and solve the problem that the identification accuracy is reduced due to the lack of identification capability of weak leakage signals and background noise of the safety valve in the prior art. In order to solve the above technical problems, the present invention provides a safety valve state anomaly monitoring method based on edge calculation, including: acquiring an original acoustic signal, and carrying out layered denoising on the original acoustic signal to obtain a clean acoustic signal; performing frequency domain energy value calculation on the clean acoustic signal to obtain energy density data, and calculating energy difference according to the energy density data to obtain energy change data; performing risk analysis based on the energy change data to obtain leakage risk frequency, and performing frequency domain feature decomposition according to the leakage risk frequency to obtain leakage feature components; calculating energy deviation according to the leakage characteristic component to obtain an energy deviation sequence, and comparing thresholds based on the energy deviation sequence to obtain a deviation abnormal sequence; Tracking and counting according to the deviation abnormal sequence to obtain abnormal duration, and eliminating interference based on the abnormal duration to obtain refined abnormal duration; Performing confidence calculation based on the refining abnormal time length to obtain accumulated confidence, and performing threshold comparison based on the accumulated confidence to obtain a priority alarm sequence; And acquiring historical data of the safety valve, carrying out threshold correction based on the historical data of the safety valve to obtain a corrected alarm threshold value, and comparing based on the corrected alarm threshold value and the priority alarm sequence to obtain an instant alarm signal. In an alternative embodiment, the acquiring the original acoustic signal, performing layered denoising on the original acoustic signal to obtain a clean acoustic signal, and includes: Acquiring an original acoustic signal, and carrying out wavelet decomposition on the original acoustic signal to obtain a signal coefficient set; and carrying out soft threshold screening according to the signal coefficient set to obtain screening signal coefficients, and carrying out reverse combination based on the screening signal coefficients to obtain clean acoustic signals. In an alternative embodiment, the calculating the frequency domain energy value of the clean acoustic signal to obtain energy density data, calculating energy difference according to the energy density data to obtain energy variation data includes: Performing short-time Fourier transform based on the clean acoustic signals to obtain frequency domain amplitude data, and performing energy value calculation according to the frequency domain amplitude data to obtain energy density data; generating a map based on