A method and system for processing detection data of a helium leak detector for a diaphragm compressor

Abstract

The application relates to the technical field of electronic data processing, and discloses a helium leak detector detection data processing method and system for a diaphragm compressor, which comprises the following steps: acquiring an original helium concentration time sequence signal and a periodic frequency parameter of a stroke of the diaphragm compressor; constructing a wave trap filter group, filtering the original helium concentration time sequence signal to obtain a preliminary signal; carrying out wavelet transform decomposition on the preliminary signal, carrying out threshold processing on wavelet coefficients, and reconstructing to obtain a denoising signal; iteratively fitting the denoising signal to estimate a dynamic background baseline and obtain a background correction signal; calculating a first-order time derivative sequence, a local information entropy sequence and a morphological correlation degree; and when the amplitude of the first-order time derivative sequence is continuously higher than a first threshold value, the local information entropy sequence is lower than a second threshold value, and the morphological correlation degree is higher than a third threshold value, it is determined that a leakage occurs. The application can multi-dimensionally and high-confidence-ly recognize a leakage event, and improves the detection sensitivity, anti-interference ability and reliability of a tiny leakage.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electronic data processing, in particular to a detection data processing method and system for a helium leak detector for a diaphragm compressor. BACKGROUND

[0002] Due to its structural characteristics, the diaphragm compressor has unique advantages when conveying high-purity, flammable, explosive or toxic and harmful gases, and is applied in the fields of chemical industry, nuclear industry, semiconductor and food and medicine. In order to ensure the safe operation and medium purity of the compressor, it is crucial to detect the air tightness of the compressor and its components with high precision. The helium mass spectrometry leak detection method is one of the most sensitive industrial leak detection technologies currently recognized. By using helium as a tracer gas, the helium leak detector detects the trace amount of helium that escapes or penetrates, thereby determining whether there is a leak. However, when the diaphragm compressor is subjected to helium leak detection under online or simulated working conditions, the original helium concentration signal collected by the leak detector often has a very low signal-to-noise ratio. This is mainly because the signal not only contains random environmental noise and instrument thermal noise, but more seriously, the periodic reciprocating motion of the crankshaft connecting rod mechanism of the compressor will produce strong periodic interference, which is highly related to the frequency of the compressor stroke and its harmonic components. This interference often overwhelms the slowly changing helium concentration signal caused by real small leaks, posing a great challenge to accurate identification of leaks.

[0003] In order to extract the weak leakage signal from the strong interference background, the existing data processing methods usually adopt simple filtering or threshold judgment. For example, moving average or low-pass filter is used to smooth the signal to suppress high-frequency random noise. However, such linear filter, while suppressing noise, also tends to blur the edges of the real leakage signal, reducing the instantaneous response capability of the detection, and making it difficult to effectively remove the periodic interference of specific frequency directly related to the operating state of the compressor. Another method is to set a fixed concentration threshold, and when the signal exceeds the threshold, it is determined as a leak. However, due to factors such as changes in ambient temperature, sensor aging, the background signal (baseline) of the helium leak detector will slowly drift, and the fixed threshold strategy cannot adapt to this dynamically changing baseline, which is prone to false positives or false negatives. In addition, some non-leakage instantaneous disturbances may also cause temporary mutations in the signal, and it is difficult to distinguish them from real exponential rising leakage patterns by amplitude judgment alone. Therefore, the existing technology has the problems of poor robustness, low sensitivity and insufficient automatic identification capability when processing leak detection signals of dynamic devices such as diaphragm compressors, and it is difficult to meet the needs of high-reliability continuous monitoring. SUMMARY

[0004] The application provides a helium leak detector detection data processing method and system for a diaphragm compressor to solve the problems of poor robustness, low sensitivity and insufficient automatic identification capability in the prior art, and to meet the high reliability continuous monitoring requirement.

[0005] In a first aspect, the helium leak detector detection data processing method for a diaphragm compressor comprises the following steps:

[0006] An original helium concentration time sequence signal collected by the helium leak detector at a preset sampling frequency is obtained, and a periodic frequency parameter of the stroke of the diaphragm compressor is synchronously obtained; a notch filter set for the fundamental frequency and harmonic frequencies of the periodic frequency parameter is constructed according to the periodic frequency parameter, the original helium concentration time sequence signal is filtered to obtain a preliminary signal; the preliminary signal is decomposed by wavelet transform, and the wavelet coefficient threshold values of each decomposition scale layer are calculated, the wavelet coefficients are processed by threshold value after reconstruction to obtain a denoising signal; the denoising signal is iteratively fitted to estimate a dynamic background baseline of the denoising signal, and the dynamic background baseline is subtracted from the denoising signal to obtain a background corrected signal; based on the background corrected signal, a first-order time derivative sequence and a local information entropy sequence are calculated in a sliding time window, and the background corrected signal and a template signal of a preset typical leakage index rising model are normalized and cross-correlated to obtain a shape correlation degree; when the amplitude of the first-order time derivative sequence is continuously higher than a first threshold value in a preset time, the local information entropy sequence is lower than a second threshold value, and the shape correlation degree is higher than a third threshold value, it is determined that a leakage occurs.

[0007] Preferably, the construction of the notch filter set for the fundamental frequency and harmonic frequencies of the periodic frequency parameter comprises: taking the periodic frequency parameter as the fundamental frequency, and selecting a plurality of harmonic frequencies of the fundamental frequency and a preset order as the notch center frequencies; for each notch center frequency, a digital notch filter with a preset quality factor is designed; and the digital notch filters are cascaded to form the notch filter set.

[0008] Preferably, the wavelet transform decomposition of the preliminary signal, the calculation of the wavelet coefficient threshold values of each decomposition scale layer, the threshold value processing of the wavelet coefficients, and the reconstruction to obtain the denoising signal comprise: selecting a predetermined wavelet base function, decomposing the preliminary signal by multi-layer discrete wavelet transform to obtain the detail coefficients of each layer and the approximation coefficients of the highest layer; independently calculating the denoising threshold values of the detail coefficients of each layer based on the Stein unbiased risk estimation criterion; processing the detail coefficients by a preset threshold function; and performing inverse wavelet transform on the processed detail coefficients and the unprocessed approximation coefficients of the highest layer to reconstruct the denoising signal.

[0009] Preferably, the de-noised signal is iteratively fitted to estimate the dynamic background baseline of the de-noised signal, comprising: using an asymmetric least square smoothing algorithm to iteratively fit the de-noised signal, setting a smoothing parameter and an asymmetric parameter for the asymmetric least square smoothing algorithm; fitting the de-noised signal by iteratively weighting, and stopping iteration when the baselines calculated by two adjacent iterations satisfy a preset convergence criterion, and taking the final fitting result as the dynamic background baseline.

[0010] Preferably, the first-order time derivative sequence and the local information entropy sequence of the background-corrected signal are calculated in a sliding time window, and the background-corrected signal is normalized and cross-correlated with a preset template signal of a typical leakage index rising model to obtain a shape correlation degree, comprising: in the sliding time window, the first-order time derivative sequence of the background-corrected signal is calculated by using a numerical differentiation method; the background-corrected signal in the sliding time window is quantized and its information entropy sequence is calculated as the local information entropy sequence; the background-corrected signal in the sliding time window is normalized and cross-correlated with a preset template signal representing a typical leakage feature to obtain the shape correlation degree.

[0011] Preferably, the leakage is determined to occur when the amplitude of the first-order time derivative sequence is continuously higher than a first preset threshold value for a preset duration, the local information entropy sequence is lower than a second preset threshold value for the same duration, and the shape correlation degree is higher than a third preset threshold value.

[0012] Preferably, the digital notch filter is a second-order IIR digital notch filter.

[0013] Preferably, the de-noised signal is iteratively fitted to estimate the dynamic background baseline of the de-noised signal, comprising: using an asymmetric least square smoothing algorithm to iteratively fit the de-noised signal, setting a smoothing parameter and an asymmetric parameter for the asymmetric least square smoothing algorithm; fitting the de-noised signal by iteratively weighting, and stopping iteration when the baselines calculated by two adjacent iterations satisfy a preset convergence criterion, and taking the final fitting result as the dynamic background baseline.

[0014] Preferably, the information entropy is Shannon information entropy.

[0015] In a second aspect, the detection data processing system of the helium leak detector for diaphragm compressor comprises a memory and a processor, the memory stores computer instructions, and the processor executes the computer instructions to implement the detection data processing method of the helium leak detector for diaphragm compressor.

[0016] The beneficial effects of the present application are: the present application can accurately eliminate the strong periodic interference generated by the mechanical movement of the device itself by specially constructing a notch filter set for the stroke frequency of the compressor and its harmonics, providing a pure signal basis for subsequent processing. On this basis, combined with multi-scale wavelet transform and threshold processing based on specific criteria, the edge details of the real leakage signal can be maximally retained while effectively filtering out random noise, avoiding the fuzzification of signal features. Further, the asymmetric least squares smoothing algorithm is used to estimate and deduct the background baseline of the signal, effectively overcoming the baseline drift problem caused by environmental changes or instrument aging, and avoiding the misjudgment caused by the fixed threshold method. Through comprehensive consideration of the change rate of the first derivative of the signal, the signal complexity reflected by the local information entropy and the high morphological correlation with the typical leakage model, the multi-dimensional and high-confidence recognition of the real leakage event is realized, which significantly improves the detection sensitivity, anti-interference ability and the accuracy and reliability of the judgment of the micro leakage. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 A flowchart of a helium leak detector data processing method for a diaphragm compressor is provided for the embodiments of the present application. DETAILED DESCRIPTION

[0018] The embodiments of the present application will be described in detail below, and examples of the embodiments are shown in the accompanying drawings. The embodiments described below by reference to the drawings are exemplary and are intended to explain the present application, and cannot be understood as a limitation of the present application.

[0019] As shown in Figure 1 , the embodiments of the helium leak detector data processing method for a diaphragm compressor provided by the present application specifically include the following steps:

[0020] S1, obtaining an original helium concentration time series signal collected by a helium leak detector at a preset sampling frequency, and synchronously obtaining a periodic frequency parameter of the stroke of the diaphragm compressor.

[0021] Specifically, the helium concentration data is collected by the data acquisition module built in the helium leak detector at a sampling frequency of 100 Hz to form an original helium concentration time series signal; at the same time, the running speed of the diaphragm compressor is obtained by the speed sensor installed on the crankshaft of the diaphragm compressor, for example, 300 revolutions per minute, and the periodic frequency parameter f0 of the stroke is converted to 5 Hz.

[0022] S2, according to the periodic frequency parameter, directly constructing a notch filter set for the fundamental frequency and at least the first 5 order harmonic frequencies of the periodic frequency parameter, filtering the original helium concentration time series signal to suppress periodic interference, and obtaining a preliminary signal.

[0023] Specifically, taking the acquired periodic frequency parameter f0 as a base frequency, the interference frequency points to be suppressed are determined as f0, 2f0, 3f0, 4f0, 5f0, 6f0. For each interference frequency point, a second-order IIR digital notch filter is designed, and these individual notch filters are cascaded to form a notch filter group. The original helium concentration time series signal is sequentially passed through the notch filter group to obtain a preliminary signal.

[0024] S3, performing multi-scale discrete wavelet transform decomposition on the preliminary signal, and calculating a wavelet coefficient threshold value of each decomposition scale layer based on a Stein unbiased risk estimation criterion, performing threshold value processing on the wavelet coefficient, and reconstructing to obtain a denoised signal.

[0025] Specifically, the Daubechies 4 wavelet basis is selected to perform 5-layer discrete wavelet transform decomposition on the preliminary signal to obtain high-frequency detail coefficients and low-frequency approximation coefficients of each layer. For the high-frequency detail coefficients of each layer, an optimal data-driven threshold value is calculated using the SURE criterion. The high-frequency coefficients are then shrunk using a soft threshold function, the low-frequency approximation coefficients are kept unchanged, and the signal is reconstructed through inverse discrete wavelet transform to obtain a denoised signal.

[0026] S4, using an asymmetric least squares smoothing algorithm to iteratively fit the denoised signal, estimating a dynamic background baseline of the denoised signal, and subtracting the dynamic background baseline from the denoised signal to obtain a background-corrected signal.

[0027] Specifically, the asymmetric least squares smoothing algorithm is used by setting an asymmetric factor p, the factor taking a value between 0.001 and 0.1, and a smoothing factor lambda. In the iteration process, a smaller weight is given to the signal points higher than the current fitting baseline, and a larger weight is given to the points lower than the baseline. A weighted penalty least squares problem is solved through multiple iterations to obtain a dynamic background baseline that closely follows the bottom of the signal. The denoised signal is subtracted from the dynamic background baseline to obtain a background-corrected signal.

[0028] S5, based on the background-corrected signal, calculating a first-order time derivative sequence and a local information entropy sequence in a sliding time window, and performing normalized cross-correlation operation on the background-corrected signal and a template signal of a pre-set typical leakage index rising model to obtain a shape correlation degree.

[0029] Specifically, a sliding time window with length N is set, and N is exemplarily equal to 50 data points. In each window, the first-order derivative of the background corrected signal is calculated using the difference method. At the same time, the amplitude of the signal in the window is histogramed, the probability distribution is calculated, and the local information entropy is calculated according to the Shannon entropy formula. In addition, an exponential rising function in the form of A multiplied by 1 minus e to the negative t divided by τ power is generated as a template signal, and a normalized cross-correlation calculation is performed between the template signal and the background corrected signal in the window to obtain a morphology correlation value ranging from -1 to 1.

[0030] S6, when the amplitude of the first-order time derivative sequence is continuously higher than the first threshold value in the preset time, the local information entropy sequence is lower than the second threshold value, and the morphology correlation degree is higher than the third threshold value, it is determined that leakage occurs.

[0031] Specifically, the first threshold value is set to 0.1, the second threshold value is set to 2.5, and the third threshold value is set to 0.85. The preset time is set to 0.5 seconds. During data processing, the three indicators are continuously monitored. If the amplitude of the first-order time derivative sequence is greater than 0.1 for 0.5 seconds, and at the same time, the value of the local information entropy sequence is less than 2.5, and the value of the morphology correlation degree is also greater than 0.85, a leakage alarm is triggered.

[0032] In an optional embodiment, a notch filter set for the fundamental frequency and at least the first 5 harmonic frequencies of the periodic frequency parameter is directly constructed, including: taking the periodic frequency parameter as the fundamental frequency, and selecting the fundamental frequency and a plurality of harmonic frequencies of the preset order of the fundamental frequency as the notch center frequencies; for each notch center frequency, a digital notch filter with a preset quality factor is designed; and the digital notch filters are cascaded to form the notch filter set.

[0033] Specifically, in industrial environment, the pipeline vibration signal is often seriously affected by the power frequency interference of 50 Hz and its harmonics. Therefore, the embodiment selects 50 Hz as the periodic frequency parameter, i.e. the fundamental frequency. In order to completely eliminate this interference, the second, third, fourth and fifth harmonics of the fundamental frequency are selected to be filtered out, i.e. the notch center frequencies are set to 50 Hz, 100 Hz, 150 Hz, 200 Hz and 250 Hz. The main power frequency interference frequency band can be comprehensively covered. For each of the above determined notch center frequencies, a second order IIR digital notch filter is designed. In order to accurately filter out the interference frequency while maximizing the retention of effective signals in the adjacent frequency band, the quality factor Q value of the filter is set to a high value, for example 35. Subsequently, the five digital notch filters designed for 50 Hz, 100 Hz, 150 Hz, 200 Hz and 250 Hz respectively are cascaded. After the signal passes through the notch filter bank, the power frequency and its harmonic components are effectively suppressed, thereby obtaining a pure preliminary signal.

[0034] In an optional embodiment, the preliminary signal is subjected to multi-scale discrete wavelet transform decomposition, and the wavelet coefficient threshold value of each decomposition scale layer is calculated based on the Stein unbiased risk estimation criterion. After threshold processing of the wavelet coefficients, a denoising signal is reconstructed, comprising: selecting a predetermined wavelet basis function, subjecting the preliminary signal to multi-layer discrete wavelet transform decomposition to obtain detail coefficients of each layer and approximation coefficients of the highest layer; independently calculating the denoising threshold value of each layer of detail coefficients based on the Stein unbiased risk estimation criterion; processing the detail coefficients using a preset threshold function; and performing wavelet inverse transform on the processed detail coefficients and the unprocessed approximation coefficients of the highest layer to reconstruct the denoising signal.

[0035] Specifically, in order to effectively separate the noise in the preliminary signal, the embodiment selects the Daubechies 5 wavelet basis function, i.e. db5 wavelet, which has good time-frequency localization characteristics. The preliminary signal subjected to notch filtering is subjected to 5-layer discrete wavelet transform decomposition. This decomposition produces five layers of detail coefficients, which represent the signal details at different frequency scales, and an approximation coefficient of the highest layer, which represents the low frequency profile of the signal. For example, the first layer of detail coefficients corresponds to the highest frequency part of the signal, and the fifth layer of approximation coefficients corresponds to the most gentle component of the signal.

[0036] Specifically, after decomposition, instead of using a uniform threshold, Stein's unbiased risk estimator is applied to calculate the threshold for each decomposition scale layer (i.e. wavelet coefficient threshold) independently. For example, the threshold for the first layer of detail coefficients is calculated to be 0.8, and the threshold for the second layer is calculated to be 0.5, and so on. Subsequently, soft thresholding is applied to each layer of detail coefficients. Soft thresholding can set coefficients below the threshold to zero and shrink coefficients above the threshold towards zero, which can make the reconstructed signal smoother than hard thresholding. Finally, the five layers of thresholded detail coefficients and the highest layer of approximation coefficients without any processing are used to reconstruct the signal through inverse wavelet transform, thereby obtaining a denoised signal with random noise effectively removed.

[0037] In an optional embodiment, an asymmetric least squares smoothing algorithm is used to iteratively fit the denoised signal to estimate the dynamic background baseline, including: setting a smoothing parameter and an asymmetry parameter for the asymmetric least squares smoothing algorithm; fitting the denoised signal by iteratively weighting, and stopping iteration when the baselines calculated by two adjacent iterations satisfy a preset convergence criterion, and taking the final fitting result as the dynamic background baseline.

[0038] Specifically, in order to accurately extract the slowly changing dynamic background baseline in the signal while ignoring the upward pulse caused by leakage, the asymmetric least squares smoothing algorithm in this embodiment is set with two key parameters. The smoothing parameter lambda is set to 10 to the power of 6, and a larger value ensures that the fitted baseline is sufficiently smooth. The asymmetry parameter p is set to 0.01, which is much smaller than 0.5. This makes the asymmetric least squares smoothing algorithm apply much smaller weights to signal points above the current baseline than to signal points below the baseline when fitting, so that the fitting result tends to be close to the lower envelope of the signal.

[0039] The baseline fitting is performed iteratively. In each iteration, the weight is dynamically adjusted using the asymmetry parameter p according to the relative position of the current data point to the baseline of the previous round, and the weighted least squares fitting curve is recalculated. The above process is repeated until the root mean square error between the baselines calculated by two consecutive iterations is less than a preset convergence criterion, for example, less than 10 to the power of -6. When this condition is met, the iteration stops, and the final fitting curve obtained at this time is considered to accurately reflect the dynamic background baseline reflecting the change in working conditions.

[0040] In an optional embodiment, based on the background correction signal, a first-order time derivative sequence and a local information entropy sequence are calculated in a sliding time window, and the background correction signal is normalized and cross-correlated with a preset template signal of a typical leakage index rising model to obtain a shape correlation degree, including: in the sliding time window, the first-order time derivative sequence of the background correction signal is calculated by using a numerical differentiation method; the background correction signal in the sliding time window is quantized and its information entropy sequence is calculated as the local information entropy sequence; the background correction signal in the sliding time window is normalized and cross-correlated with a preset template signal representing a typical leakage feature to obtain the shape correlation degree.

[0041] Specifically, a sliding time window with a length of 2 seconds is used to analyze the background correction signal. In each time window, the first-order time derivative of the signal is calculated by using the central difference method, and the magnitude of the derivative value reflects the rate of change of the signal amplitude, which will increase sharply when leakage occurs. At the same time, the amplitude range of the signal in the sliding time window is evenly divided into 64 quantization levels, and the Shannon information entropy of the signal is calculated based on this. Under normal working conditions, the signal randomness is strong, and the information entropy is high, while when leakage occurs, the signal shape tends to be regular, and the information entropy will decrease significantly.

[0042] In order to evaluate the similarity between the shape of the background correction signal and the leakage feature, a template signal of a typical leakage index rising model is constructed in advance. The template signal can be obtained by processing the monitoring data of a known and typical small leakage event. In the sliding time window, the background correction signal in the window is normalized and cross-correlated with the preset template signal. The calculation result is the shape correlation degree, which is between -1 and 1. A value close to 1, such as 0.9, indicates that the waveform of the current background correction signal is highly consistent with the waveform of the typical leakage feature.

[0043] In an optional embodiment, determining that leakage occurs includes: when the amplitude of the first-order time derivative sequence is continuously higher than a first preset threshold value for a preset duration, and the local information entropy sequence is lower than a second preset threshold value for the same duration, and the shape correlation degree is higher than a third preset threshold value, it is determined that leakage occurs.

[0044] Specifically, to ensure the accuracy and reliability of the leakage determination, three independent criteria and a duration condition are provided in the embodiment. The first preset threshold is 5.0 units per second for the first-order time derivative, which is used to capture the mutation of the signal. The second preset threshold is 3.5 for the local information entropy, which is used to identify the transition of the signal from random to regular. The third preset threshold is 0.85 for the shape correlation degree, which is used to confirm the matching degree of the signal waveform and the leakage characteristics. These thresholds are obtained through statistical analysis of a large amount of historical normal operation data and simulated leakage data. Only when the above three conditions are met at the same time, the leakage alarm can be triggered. This multi-feature fusion and duration constraint determination strategy can effectively filter out false positives caused by single impact or temporary interference. Only when the signal shows a sustained behavior that meets all the key characteristics of the leakage, the leakage is finally determined to have occurred.

[0045] The implementation principle of the detection data processing method for the helium leak detector for diaphragm compressors in the embodiment of the present application is as follows: by constructing a notch filter set for the stroke frequency of the compressor and its harmonics, the strong periodic interference generated by the mechanical movement of the equipment can be accurately eliminated, providing a pure signal basis for subsequent processing. Moreover, combined with multi-scale wavelet transform and threshold processing, the edge details of the real leakage signal can be maximally retained while effectively filtering out random noise, avoiding the blurring of signal characteristics. In addition, the asymmetric least squares smoothing algorithm is used to estimate and deduct the background baseline of the signal, effectively overcoming the baseline drift problem caused by environmental changes or instrument aging, avoiding false judgments caused by fixed threshold methods. Moreover, by comprehensively investigating the change rate of the first-order derivative of the signal, the signal complexity reflected by the local information entropy, and the high shape correlation with the typical leakage model, the real leakage event can be identified in multiple dimensions and with high confidence, significantly improving the detection sensitivity, anti-interference ability, and accuracy and reliability of the determination of the micro leakage.

[0046] The embodiment of the detection data processing system for the helium leak detector for diaphragm compressors provided by the present application includes a memory and a processor, and the memory stores computer instructions. When the processor executes the computer instructions, the detection data processing method for the helium leak detector for diaphragm compressors in the above embodiment is realized.

[0047] Although the embodiments of the present application have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the present application. Those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1. A method for processing data detected by a helium leak detector for a diaphragm compressor, characterized in that, The process includes the following steps: acquiring the original helium concentration time series signal collected by a helium leak detector at a preset sampling frequency, and simultaneously acquiring the periodic frequency parameters of the diaphragm compressor stroke; constructing a notch filter bank for the fundamental frequency and harmonic frequencies of the periodic frequency parameters to filter the original helium concentration time series signal to obtain a preliminary signal; performing wavelet transform decomposition on the preliminary signal, calculating the wavelet coefficient thresholds for each decomposition scale, reconstructing the wavelet coefficients after threshold processing to obtain a denoised signal; iteratively fitting the denoised signal to estimate the dynamic background baseline of the denoised signal, and subtracting the dynamic background baseline from the denoised signal to obtain a background correction signal; Based on the background correction signal, its first-order time derivative sequence and local information entropy sequence are calculated within a sliding time window. At the same time, the background correction signal is normalized and cross-correlated with the template signal of a pre-set typical leakage exponential rise model to obtain the morphological correlation. When the amplitude of the first-order time derivative sequence is continuously higher than the first preset threshold for a preset duration, and the local information entropy sequence is lower than the second preset threshold for the same duration, and the morphological correlation is higher than the third preset threshold, leakage is determined to have occurred.

2. The data processing method for a helium leak detector used in a diaphragm compressor according to claim 1, characterized in that, The construction of a notch filter bank for the fundamental frequency and harmonic frequencies of the periodic frequency parameter includes: using the periodic frequency parameter as the fundamental frequency, and selecting the fundamental frequency and a plurality of harmonic frequencies of a preset order as notch center frequencies; designing a digital notch filter with a preset quality factor for each notch center frequency; and cascading the digital notch filters to form the notch filter bank.

3. The data processing method for a helium leak detector used in a diaphragm compressor according to claim 1, characterized in that, The process of performing wavelet transform decomposition on the initial signal, calculating wavelet coefficient thresholds for each decomposition scale, and reconstructing the wavelet coefficients after threshold processing to obtain a denoised signal includes: selecting a predetermined wavelet basis function, performing multi-level discrete wavelet transform decomposition on the initial signal to obtain detail coefficients for each level and approximation coefficients for the highest level; independently calculating the denoising thresholds for the detail coefficients at each level based on the Stein unbiased risk estimation criterion; processing the detail coefficients using a preset threshold function; and performing inverse wavelet transform on the processed detail coefficients and the unprocessed approximation coefficients of the highest level to reconstruct the denoised signal.

4. The data processing method for a helium leak detector used in a diaphragm compressor according to claim 1, characterized in that, The step of iteratively fitting the denoised signal to estimate the dynamic background baseline of the denoised signal includes: iteratively fitting the denoised signal using an asymmetric least squares smoothing algorithm, setting a smoothness parameter and an asymmetric parameter for the asymmetric least squares smoothing algorithm; fitting the denoised signal by iterative weighting, stopping the iteration when the baseline calculated in two adjacent iterations meets the preset convergence criterion, and using the final fitting result as the dynamic background baseline.

5. The data processing method for a helium leak detector used in a diaphragm compressor according to claim 1, characterized in that, The method involves calculating the first-order time derivative sequence and the local information entropy sequence of the background correction signal within a sliding time window, and simultaneously performing a normalized cross-correlation operation on the background correction signal with a pre-defined template signal representing a typical leakage exponential increase model to obtain the morphological correlation. This includes: calculating the first-order time derivative sequence of the background correction signal using a numerical differentiation method within the sliding time window; quantizing the background correction signal within the sliding time window and calculating its information entropy sequence as the local information entropy sequence; and performing a normalized cross-correlation operation on the background correction signal within the sliding time window with a pre-defined template signal representing typical leakage characteristics to obtain the morphological correlation.

6. The data processing method for a helium leak detector used in a diaphragm compressor according to claim 2, characterized in that, The digital notch filter is a second-order IIR digital notch filter.

7. The data processing method for a helium leak detector used in a diaphragm compressor according to claim 3, characterized in that, The step of selecting a predetermined wavelet basis function to perform multi-level discrete wavelet transform decomposition on the preliminary signal includes: using the Daubechies 4 wavelet basis to perform 5-level discrete wavelet transform decomposition on the preliminary signal.

8. The data processing method for a helium leak detector used in a diaphragm compressor according to claim 5, characterized in that, The information entropy mentioned is Shannon information entropy.

9. A data processing system for a helium leak detector used in a diaphragm compressor, characterized in that, It includes a memory and a processor. The memory stores computer instructions. When the processor executes the computer instructions, it implements the data processing method for a helium leak detector for a diaphragm compressor as described in any one of claims 1-8.

Images