Adaptive denoising method, system, device and medium for glow spectrum signal

Abstract

The present application relates to the technical field of glow discharge spectrum signal processing, and specifically provides a kind of adaptive denoising method, system, equipment and medium of glow spectrum signal, comprising: the continuous multiple frames of spectrum signal generated in the process of collecting glow discharge;Stability evaluation is carried out to each frame spectrum signal, and stability score is calculated;According to the stability score, the frame number N of the spectrum frame for superposition is dynamically adjusted, and the N frame spectrum signal screened out is weighted and fused to obtain the fusion spectrum signal;The fusion spectrum signal is carried out wavelet decomposition, and the decomposition coefficient of multiple scales is obtained;According to the local feature of each scale signal, the denoising threshold is adaptively determined, and the decomposition coefficient is thresholding processed using the denoising threshold;The decomposition coefficient after thresholding processing is carried out wavelet reconstruction, and the denoising spectrum signal with high signal-to-noise ratio is obtained.The present application effectively suppresses noise while maintaining spectral peak details, significantly improves signal-to-noise ratio and detection accuracy.

Description

Technical Field

[0001] This invention belongs to the field of glow discharge spectral signal processing technology, specifically relating to an adaptive denoising method, system, terminal, and medium for glow discharge spectral signals. Background Technology

[0002] Glow discharge spectroscopy, based on the principle of plasma emission, enables simultaneous detection of multiple elements and has wide applications in fields such as metallic material analysis and environmental monitoring. With the development of portable spectroscopic systems, rapid on-site detection technology based on micro-plasma discharge has become a research hotspot.

[0003] However, glow discharge spectral signal processing faces a fundamental technical challenge: the discharge process has strong nonlinear and non-stationary characteristics, which leads to significant fluctuations in the spectral signals acquired at different time frames. Existing denoising methods are all based on the assumption of stationary signals, which makes it difficult to effectively deal with such dynamic changes.

[0004] Specifically, the traditional multi-frame stacking averaging method uses a fixed number of frames for stacking, without considering the real-time changes in the discharge state. When the discharge is unstable, the introduction of low-quality frames will contaminate the stacking result. The wavelet threshold denoising method uses a fixed threshold to process all scales, which cannot be adaptively adjusted according to the signal energy distribution, and is prone to causing the loss of weak spectral line details or noise residue. The fixed parameter settings of the Savitzky-Golay smoothing method are also difficult to adapt to the local features of complex multi-spectral lines.

[0005] Existing methods lack the ability to sense the dynamic characteristics of glow discharge and the adaptive adjustment mechanism, and cannot achieve a dynamic balance between signal enhancement and detail fidelity, which seriously restricts the further improvement of detection sensitivity and accuracy. Summary of the Invention

[0006] In view of the above-mentioned shortcomings of the prior art, the present invention provides an adaptive denoising method, system, device and medium for glow discharge spectral signals to solve the above-mentioned technical problems.

[0007] In a first aspect, the present invention provides an adaptive denoising method for glow discharge spectral signals, comprising:

[0008] Acquire continuous multi-frame spectral signals generated during the glow discharge process; For each frame of spectral signal, a stability assessment is performed, and a stability score characterizing the quality of that frame of signal is calculated. Based on the stability score, the number N of spectral frames used for superposition is adjusted, and the N spectral signals selected based on the stability score are weighted and fused to obtain a fused spectral signal, where N is an integer greater than 1, and the number of frames N is positively correlated with the stability score. Wavelet decomposition is performed on the fused spectral signal to obtain decomposition coefficients at multiple scales; The denoising threshold is adaptively determined based on the local features of the signal at each scale. The decomposition coefficients are then thresholded using the denoising threshold, and wavelet reconstruction is performed on the thresholded decomposition coefficients to obtain a denoised spectral signal with a high signal-to-noise ratio.

[0009] In an optional implementation, a stability assessment is performed on each frame of the spectral signal, and a stability score characterizing the quality of that frame of signal is calculated, including: Extract multiple stability feature indices from each frame of spectral signal, and construct the feature vector of the frame signal based on the multiple stability feature indices; A weighted Mahalanobis distance stability discrimination algorithm is used to calculate the Mahalanobis distance between the feature vector and the preset stable reference vector; The Mahalanobis distance is mapped to the stability score, and the smaller the Mahalanobis distance, the higher the stability score. The stability characteristics include spectral peak drift, which characterizes the stability of spectral line position; light intensity fluctuation rate, which characterizes the fluctuation of signal intensity; discharge current stability, which characterizes the stability of discharge state; spectral peak centroid drift, which characterizes the degree of spectral line shape retention; baseline slope change, which characterizes background changes; and high-frequency noise energy density, which characterizes noise level.

[0010] In an optional implementation, adjusting the number N of spectral frames used for stacking based on the stability score includes: Establish a mapping relationship between stability score and number of stacked frames, wherein the mapping relationship is monotonically increasing, such that when the stability score increases, the number of stacked frames N increases, and when the stability score decreases, the number of stacked frames N decreases. The mapping relationship is a piecewise linear mapping function that divides the stability score interval into multiple sub-intervals. Each sub-interval corresponds to a number of superimposed frames, and the number of superimposed frames N is within a preset range.

[0011] In an optional implementation, adjusting the number N of spectral frames used for stacking based on the stability score includes: Based on historical stability score sequences, multi-scale trend features containing stability scores in the near future are extracted. The multi-dimensional prediction features are mapped to corresponding fuzzy sets to obtain fuzzy input variables; The corresponding rules in the fuzzy rule base are activated according to the fuzzy input variables, and fuzzy inference aggregation is performed to obtain the output fuzzy set; wherein, the fuzzy rule base contains multiple fuzzy inference rules, and each rule contains antecedent conditions and consequent conclusions; The output fuzzy set is converted into a precise overlay frame number adjustment amount; The fuzzy rule base includes at least one of the following rules: Active gain rule under positive prediction: If the current frame stability score is medium, the prediction stability score is rising, and the fluctuation state is stable, then the number of frames to be adjusted is increased slightly. Risk avoidance rules under negative prediction: If the current frame stability score is high, the prediction stability score is decreasing, and the short-term trend strength is medium, then the superimposed frame adjustment amount is slightly reduced. Conservative strategy rule under high volatility: If the volatility is severe, the amount of frame adjustment is greatly reduced. Strengthening rules under trend confirmation: If the current frame stability score is low, the short-term trend strength is strong, the short-term trend slope is negative, and the predicted stability score is decreasing, then the amount of frame adjustment is significantly reduced.

[0012] In an optional implementation, the N frames of spectral signals selected based on the stability score are weighted and fused to obtain a fused spectral signal, including: Based on the stability score of each frame of spectral signal, the N frames of spectral signals with the highest stability score are selected from the multiple frames of spectral signals continuously acquired within the sliding window as the frames to be fused; the length of the sliding window is adaptively adjusted according to the degree of fluctuation of the discharge state. Based on the stability score of each frame of spectral signal, the fusion weight corresponding to that frame of signal is calculated, wherein the fusion weight is positively correlated with the stability score; Based on the fusion weights, the N frames to be fused are weighted and averaged to obtain the fused spectral signal.

[0013] In one optional implementation, a denoising threshold is adaptively determined based on the local features of the signals at each scale, and the decomposition coefficients are thresholded using the denoising threshold, including... Based on the local energy distribution of signals at each scale, calculate the energy proportion factor corresponding to that scale; The basic threshold model is modified based on the energy proportion factor to obtain the adaptive denoising threshold for this scale. The detail coefficients at this scale are thresholded using the adaptive denoising threshold.

[0014] In an optional implementation, the basic threshold model is modified based on the energy proportion factor to obtain an adaptive denoising threshold for that scale, including: The formula for calculating the basic threshold is:

[0015] in, The noise standard deviation is estimated by the median absolute deviation of the highest frequency subband detail coefficients obtained from wavelet decomposition; N is the number of fused frames. By combining the basic threshold with a function of the energy proportion factor, an adaptive threshold is obtained. ,in For energy percentage factor The monotonically decreasing function allows for the use of smaller thresholds at scales with higher signal energy proportions to preserve details, and larger thresholds at scales with lower signal energy proportions to suppress noise.

[0016] In a second aspect, the present invention provides an adaptive denoising system for glow discharge spectral signals, comprising: The signal acquisition module is used to acquire continuous multi-frame spectral signals generated during the glow discharge process; The stability assessment module is used to assess the stability of each frame of spectral signal and calculate a stability score that characterizes the quality of the signal in that frame. The signal overlay module is used to adjust the number N of spectral frames used for overlay according to the stability score, and to perform weighted fusion of the N spectral signals selected based on the stability score to obtain a fused spectral signal, wherein N is an integer greater than 1, and the number of frames N is positively correlated with the stability score. The wavelet decomposition module is used to perform wavelet decomposition on the fused spectral signal to obtain decomposition coefficients at multiple scales. The wavelet denoising module is used to adaptively determine the denoising threshold based on the local features of the signal at each scale, use the denoising threshold to threshold the decomposition coefficients, and perform wavelet reconstruction on the thresholded decomposition coefficients to obtain a denoised spectral signal with a high signal-to-noise ratio.

[0017] Thirdly, a device is provided, comprising: Memory used to store the adaptive denoising program for glow discharge spectral signals; The processor is configured to implement the adaptive denoising method for the glow spectrum signal as provided in the first aspect when executing the adaptive denoising procedure for the glow spectrum signal.

[0018] Fourthly, a computer-readable medium is provided, on which an adaptive denoising program for a glow discharge spectrum signal is stored, wherein when the program is executed by a processor, it implements the steps of the adaptive denoising method for a glow discharge spectrum signal as provided in the first aspect.

[0019] The adaptive denoising method, system, device, and medium for glow discharge spectral signals provided by this invention have the following significant advantages: First, dynamic adaptive frame fusion overcomes the limitations of fixed overlay. This invention introduces a stability scoring mechanism to evaluate the discharge state in real time and dynamically adjust the number of overlay frames. When the discharge is stable, the number of frames is increased to improve the signal-to-noise ratio, and when the discharge fluctuates, the number of frames is reduced to avoid low-quality frame contamination.

[0020] Second, the hybrid denoising strategy balances fidelity preservation and noise reduction. This invention combines Savitzky-Golay smoothing with energy-ratio-based adaptive wavelet thresholding, suppressing high-frequency noise while preserving weak spectral details through scale-adaptive thresholding.

[0021] Third, it boasts strong real-time performance and broad applicability. This invention exhibits high computational efficiency, can be embedded in hardware platforms such as FPGAs to achieve millisecond-level response, and is suitable for various discharge modes including continuous, pulsed, and micro-plasma discharges, demonstrating significant engineering practical value. Attached Figure Description To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a schematic flowchart of a method according to an embodiment of the present invention.

[0023] Figure 2 This is a dynamic relationship diagram between the stability score and the number of superimposed frames of a method according to an embodiment of the present invention.

[0024] Figure 3 This is a flowchart of a hybrid denoising algorithm according to an embodiment of the present invention.

[0025] Figure 4 This is a test data graph of a signal without the adaptive denoising method for glow discharge spectral signals provided in this invention.

[0026] Figure 5 This is a test data graph using the adaptive denoising method for glow discharge spectral signals provided by this invention.

[0027] Figure 6 This is a system deployment architecture diagram of a method according to an embodiment of the present invention.

[0028] Figure 7 This is a schematic diagram of the structure of a device provided in an embodiment of the present invention. Detailed Implementation

[0029] To enable those skilled in the art to better understand the technical solutions of this invention, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this invention.

[0030] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0031] The method provided in this embodiment of the invention is executed by a computer device, and correspondingly, the adaptive denoising system for glow discharge spectral signals runs in the computer device.

[0032] Figure 1 This is a schematic flowchart illustrating a method according to an embodiment of the present invention. Wherein, Figure 1 The executing entity can be an adaptive denoising system for glow discharge spectral signals. Depending on the specific requirements, the order of the steps in this flowchart can be changed, and some steps can be omitted.

[0033] like Figure 1 As shown, the method includes: S1. Acquire continuous multi-frame spectral signals generated during the glow discharge process; S2. Perform stability assessment on each frame of spectral signal and calculate the stability score that characterizes the quality of the signal in that frame; S3. Based on the stability score, adjust the number N of spectral frames used for superposition, and perform weighted fusion on the N spectral signals selected based on the stability score to obtain a fused spectral signal, wherein N is an integer greater than 1, and the number of frames N is positively correlated with the stability score; S4. Perform wavelet decomposition on the fused spectral signal to obtain decomposition coefficients at multiple scales; S5. Adaptively determine the denoising threshold based on the local features of the signals at each scale, use the denoising threshold to threshold the decomposition coefficients, and perform wavelet reconstruction on the thresholded decomposition coefficients to obtain a denoised spectral signal with a high signal-to-noise ratio.

[0034] In one embodiment of the present invention, based on step S1, the following will provide a possible embodiment and describe its specific implementation in a non-limiting manner.

[0035] A glow discharge spectrometer is used as the signal source. This spectrometer includes a glow discharge excitation source and a spectral detector. The glow discharge excitation source can be a DC glow discharge lamp, a pulsed glow discharge lamp, or a micro-plasma discharge device (such as a liquid cathode glow discharge module), used to excite the sample to generate characteristic emission spectra. The spectral detector is an array detector, such as a charge-coupled device (CCD) or a complementary metal-oxide-semiconductor (CMOS) image sensor, capable of simultaneously acquiring spectral information within a certain wavelength range.

[0036] In practice, the sample to be tested is placed in the glow discharge apparatus, and the discharge parameters are set (e.g., for metal samples, the discharge voltage is set to 700-1000 V, and the discharge current is set to 10-50 mA; for liquid samples, a liquid cathode glow discharge mode is used, and the solution flow rate is controlled at 1-5 mL / min). After the discharge is started, the spectrometer continuously acquires spectral data at a fixed frame rate. For example, the frame rate is set to 100 Hz, that is, one frame of spectrum is acquired every 10 milliseconds, and each frame of spectrum contains the intensity values ​​of several wavelength channels (e.g., 2048 pixels). During the acquisition process, a synchronous trigger signal ensures the time correspondence between each frame of spectrum and the discharge state.

[0037] To eliminate inherent dark current noise and background light interference from the detector, dark spectrum acquisition can be performed before or between acquisitions: the discharge source is turned off or the incident light is blocked, and multiple frames of dark spectra are acquired at the same integration time as the sample measurement. The average value is then used as the dark background. In subsequent processing, this dark background is subtracted from each frame of raw spectral data to achieve dark current correction. Furthermore, to eliminate baseline variations caused by fluctuations in light source intensity, baseline normalization can be performed on the corrected spectrum, for example, by selecting the background intensity of non-spectral line regions as a reference for normalization.

[0038] The acquisition system runs continuously, generating a series of multiple frames of spectral signal sequences. These raw spectral data will serve as input for subsequent stability assessment and denoising processing.

[0039] In one embodiment of the present invention, based on step S2, the following will provide a possible embodiment and describe its specific implementation in a non-limiting manner.

[0040] S201: Extract multi-dimensional stability feature indicators and construct feature vectors. For each frame of spectral signal acquired, this embodiment extracts the following six dimensions of stability feature indicators to form the feature vector F=[f1,f2,f3,f4,f5,f6] of that frame of signal: (1) Peak shift f1: Characterizes the stability of spectral line positions. Select several characteristic spectral lines in the spectrum (for example, for metal samples, Cu I 324.75 nm, Zn I 213.86 nm, etc.), calculate the absolute value of the shift of the peak position of each characteristic spectral line in the current frame relative to the corresponding spectral line position in the reference frame (such as the first frame or average frame), and take the maximum or average value as the peak shift. The unit is pixels or nanometers. The smaller the value, the more stable the spectral line position.

[0041] (2) Light intensity fluctuation rate f2: characterizes the stability of signal intensity. Calculate the peak intensity I of the characteristic spectral lines in the current frame. peak Peak intensity I of the spectral line corresponding to the reference frame ref Relative deviation: The average value of multiple spectral lines is taken as the light intensity fluctuation rate. The smaller this value, the more stable the signal strength.

[0042] (3) Discharge current stability f3: Characterizes the stability of the discharge state. The discharge current i at the corresponding moment of the current frame is calculated using discharge current monitoring data synchronously acquired with the spectrometer. t With the set current i set Relative deviation: The smaller this value, the more stable the discharge state.

[0043] (4) Centroid shift f4: Characterizes the stability of the spectral line morphology. For each characteristic spectral line, calculate its centroid position. The centroid position is compared with that of the reference frame, and the absolute value of the offset is taken as the centroid drift. This index reflects the change in the overall spectral shape better than the peak position.

[0044] (5) Baseline slope change f5: Characterizes the stability of background changes. Select several wavelength points at both ends of the spectrum or in non-spectral regions, fit a baseline straight line, and calculate the absolute value of the difference between the baseline slope of the current frame and the baseline slope of the reference frame. The smaller this value, the more stable the background.

[0045] (6) High-frequency noise energy density f6: Characterizes the noise level. High-speed filtering or wavelet decomposition is performed on the spectrum of the current frame to extract the energy of the high-frequency components. The energy is then divided by the number of wavelength channels to obtain the high-frequency noise energy density. The smaller this value, the less noise contamination the signal is.

[0046] The above six characteristic indicators describe the stability of the spectral signal from different perspectives and together constitute the characteristic vector F.

[0047] S202: Calculate the distance between the eigenvector and the stable reference vector using weighted Mahalanobis distance. (1) Establish a stable reference vector: Select multiple frames of spectral data collected during a period of stable discharge (e.g., 100 frames continuously collected after the instrument has warmed up and stabilized), calculate the average value of each characteristic index, and use it as the stable reference vector. .

[0048] (2) Calculate the covariance matrix: Based on the stable period data used to establish the reference vector, calculate the covariance matrix Σ between each feature index. The covariance matrix reflects the correlation between each feature and their respective variance.

[0049] (3) Setting the weight matrix: Based on the importance of each feature index to stability, set the weight matrix W. For example, the weights are allocated as follows: spectral peak drift weight 0.2, light intensity fluctuation weight 0.2, discharge current stability weight 0.15, spectral peak centroid drift weight 0.15, baseline slope change weight 0.1, and high-frequency noise energy density weight 0.2. The weight matrix is ​​a diagonal matrix, and the diagonal elements are the weights of each feature.

[0050] (4) Calculate the weighted Mahalanobis distance: For the feature vector F of the current frame, its distance from the stable reference vector F is calculated. ref The formula for calculating the weighted Mahalanobis distance between them is: This distance takes into account the degree to which each feature deviates from the reference value, the correlation between features, and the preset weights, and can comprehensively measure the stability of the current frame spectrum.

[0051] S203: Mapping Mahalanobis distance to stability score Since the Mahalanobis distance D ranges from [0, +∞), a larger distance indicates greater instability, which is not easily understood intuitively. This embodiment maps it to a scoring interval of [0, 100], and the mapping function is:

[0052] in, The decay constant can be determined based on historical data statistics, for example, by taking twice the average Mahalanobis distance during stable periods. This mapping function ensures that: when the Mahalanobis distance is 0, the score is 100 points (completely stable); as the distance increases, the score decays exponentially; when the distance is large enough, the score approaches 0.

[0053] In one embodiment of the present invention, based on step S3, the following will provide a possible embodiment and describe its specific implementation in a non-limiting manner.

[0054] S301. Based on the stability score, adjust the number N of spectral frames used for stacking, including: Establish a mapping relationship between stability score and the number of stacked frames, wherein the mapping relationship is monotonically increasing, such that when the stability score increases, the number of stacked frames N increases, and when the stability score decreases, the number of stacked frames N decreases; please refer to Figure 2 The mapping relationship is a piecewise linear mapping function, which divides the stability score interval into multiple sub-intervals. Each sub-interval corresponds to a number of superimposed frames, and the number of superimposed frames N is within a preset range.

[0055] Let the stability score Q range from [0, 100], and the number of overlay frames N range from the discrete set {3, 4, 5, 6, 7, 8, 10}. The piecewise function f(Q) is defined as follows:

[0056] In another implementation, to further improve the foresight and robustness of the frame number N, the frame number N used for superposition is adjusted according to the stability score, including: 1. Multi-scale trend analysis and forecasting Input: The stability score sequence of the most recent L frames .

[0057] Output: Four precise input feature values ​​provided to the upper-level fuzzy controller.

[0058] Step A: Extraction of trend features across multiple time scales Short-term trend (Trend_short): Calculates the most recent M s Frame (e.g., M) s =5) The slope of the linear regression of the score. Positive values ​​indicate rapid recent improvement, and negative values ​​indicate rapid recent deterioration.

[0059] Mid-term trend (Trend_mid): Calculates the most recent M m Frame (e.g., M) m =15) The slope of the linear regression of the score. Reflects a more persistent direction of change, filtering out short-term fluctuations.

[0060] Volatility: Calculated by the most recent M s The standard deviation of the frame score. This measures the degree of score jitter; high volatility indicates an unstable discharge process and strong random noise.

[0061] Score Prediction (Score_pred): The score sequence is predicted using Holt's Linear Trend Method (quadratic exponential smoothing). This method captures the level and trend of the data, as shown in the following formula: Horizontal component:

[0062] Trend components: k-frame prediction:

[0063] Take the predicted value when k=3 (predicting the next 3 frames). As Score_pred, α and β are smoothing parameters (0~1).

[0064] Step B: Feature Normalization The four feature values ​​(Trend_short, Trend_mid, Volatility, Score_pred) are normalized to the interval [0, 1] or [-1, 1] (for trend values) to facilitate subsequent fuzzification. Normalization parameters (such as maximum and minimum values) can be obtained from the initial training phase or from historical data statistics.

[0065] 2. Enhanced four-input fuzzy logic controller Input variables (4) and their fuzzy set definitions: Current stability score (S_curr): i.e., s t Fuzzy set: {Very Low (VL), Low (L), Medium (M), High (H), Very High (VH)}. Membership function uses a triangle or trapezoid.

[0066] Prediction stability score (S_pred): derived from the lower-level module's Score_pred. Fuzzy set: {decreasing (D), stable (F), increasing (R)}. Membership function must reflect the change in predicted value relative to the current value.

[0067] Short-term trend strength (T_strength): The absolute value of Trend_short is fuzzified to measure whether the change is significant. Fuzzy set: {weak (W), medium (M), strong (S)}.

[0068] Volatility (V_state): Fuzzyizes the volatility. Fuzzy set: {Stability (C), Mild Volatility (LF), Severe Volatility (HF)}.

[0069] Output variables: Overlay frame adjustment amount (ΔN): Fuzzy set: {significantly decrease (LR), decrease (R), slightly decrease (SR), maintain (H), slightly increase (SI), increase (I), significantly increase (LI)}. The output range can be set to, for example, [-3, 3].

[0070] The rule base needs to comprehensively consider the current state, future predictions, and dynamic changes. Below are examples of core rules: Active gain rule under positive prediction: The current quality is moderate, but it is predicted to improve and the environment is stable. A slight increase in the frame rate can be made to obtain better signals earlier.

[0071] Risk aversion rules under negative forecasting: Although the current quality is high, it is predicted to decline significantly and the downward trend is clear. The number of frames should be reduced slightly to avoid the inclusion of low-quality frames in the average, which would degrade the results.

[0072] Conservative strategy rules under high volatility: During periods of intense fluctuation, signal-to-noise ratio prediction is unreliable. Therefore, it is necessary to significantly reduce the number of superimposed frames to respond quickly to changes and reduce the risk of distortion introduced by superimposing frames with significant differences.

[0073] Reinforcement rules under trend confirmation: The current quality is low, and short-term trends and forecasts strongly point to further deterioration. The most aggressive measures should be taken, such as significantly reducing the number of frames, or even pausing the overlay and waiting for stability.

[0074] The rule base size will expand from a single digit in the original scheme (e.g., 3 × 3 = 9 rules) to the theoretical maximum of 135 rules (5 (current), 3 (predicted), 3 (trend), 3 (fluctuation)). In practice, it can be pruned to 20-40 core rules using expert knowledge, or automated rule generation and parameter optimization can be performed using historical data through genetic algorithms or ANFIS (Adaptive Neural Fuzzy Inference System).

[0075] Reasoning and defuzzification: The Mamdani inference method is employed, and the centroid method is used for deblurring, converting the output blur quantity ΔN into a precise value. The final number of superimposed frames is: , where N min =3, N max =10, where clamp is the amplitude limiting function.

[0076] S302. The N frames of spectral signals selected based on the stability score are weighted and fused to obtain a fused spectral signal, including: S302-1: Screening frames to be fused based on stability scores In the continuously acquired spectral signal stream, a sliding window is set to temporarily store the most recently acquired multiple frames of spectral data and their corresponding stability scores. The length L of the sliding window can be adaptively adjusted according to the degree of fluctuation in the discharge state: when the discharge state is stable and the score fluctuation is small, the window length can be appropriately increased to provide more candidate frames; when the discharge state fluctuates drastically, the window length is decreased to ensure that the selected frames have good timeliness.

[0077] For example, the adjustment rule for the sliding window length LL is defined as follows:

[0078] in, The coefficient of variation (standard deviation / mean) of the scores within the current window. The greater the fluctuation, the smaller the window, ensuring that the window always contains a sufficient number of high-quality candidate frames.

[0079] At the current moment, the sliding window contains L frames of spectral signals and their corresponding stability scores {Q1, Q2, ..., Q}. L The L frames are sorted from highest to lowest stability score, and the top N frames are selected as the frames to be fused (N is the number of stacked frames dynamically determined in step S301). This screening mechanism ensures that the fusion is always performed using the highest quality spectral frames from the most recent period, avoiding contamination of the fusion results by low-quality frames.

[0080] It should be noted that when the number of frames with a score higher than the preset quality threshold (e.g., Q>60) within the sliding window is less than N, one of the following two strategies can be used to supplement them: Strategy A: Continue selecting frames with relatively high scores from the remaining frames until a total of N frames are reached; Strategy B: Select several frames that are closest to the current time in time to ensure the timeliness of the signal.

[0081] This embodiment prioritizes strategy A to ensure the fusion quality to the greatest extent possible.

[0082] S302-2: Calculate the fusion weights for each frame. For the selected N frames to be fused, the corresponding fusion weights are calculated based on their stability scores. The weight calculation follows the principle of "the higher the score, the greater the weight," ensuring that high-quality frames dominate the fusion process.

[0083] An exemplary formula for calculating weights is as follows: in, For example, take the adjustment coefficient (e.g., take...) =0.1), used to control the sensitivity of the weight to score differences. The larger the value, the greater the weight of high-scoring frames.

[0084] S302-3: Weighted Average Fusion After obtaining the fusion weights for each frame, a weighted average is performed on the N selected frames to be fused to obtain the fused spectral signal. .

[0085] Let the intensity value of the i-th frame to be fused at the k-th wavelength point or pixel point be... The fused spectral signal is then:

[0086] Where K is the total number of wavelength points in the spectral signal. Let be the fusion weight corresponding to the i-th frame.

[0087] Through the aforementioned weighted fusion, spectral frames with high stability scores contribute more to the fusion result, while frames with low scores contribute less, thus preserving the characteristics of high-quality signals to the greatest extent while suppressing noise. Compared to traditional equal-weighted stacking averages, the weighted fusion method of this invention can adaptively reduce the negative impact of unstable frames on the fusion result, further improving the signal quality of the fused spectrum.

[0088] In one embodiment of the present invention, based on step S4, the following will provide a possible embodiment and describe its specific implementation in a non-limiting manner.

[0089] S401: Selection of Wavelet Basis Functions Before performing wavelet decomposition, it is necessary to select appropriate wavelet basis functions. Considering the characteristics of glow discharge spectral signals—containing sharp spectral peaks and continuous background signals—this embodiment preferentially adopts the following two types of wavelet basis functions: Coiflet wavelet series (such as coif2, coif3, coif4): This wavelet series has advantages such as tight support, orthogonality, and approximate symmetry. Moreover, it has better vanishing moments than Daubechies wavelet when the support length is the same, and can more effectively characterize the detailed features in spectral signals.

[0090] Bioorthogonal wavelet series (such as bior2.4, bior3.5, bior4.4): This wavelet series has linear phase characteristics, which can avoid phase distortion during signal reconstruction, and is especially important for maintaining the accuracy of spectral peak positions. At the same time, bioorthogonal wavelets allow the use of different filters during decomposition and reconstruction, and can achieve a balance between smoothness and tight support.

[0091] For example, this embodiment selects the coif4 wavelet as the basis function, which has a 10th-order vanishing moment, and can effectively extract detailed features in the spectral signal while suppressing noise.

[0092] S402: Determining the number of decomposition levels The choice of wavelet decomposition level directly affects the denoising effect and computational efficiency. Too few decomposition levels result in incomplete separation of noise and signal; too many decomposition levels may lead to over-smoothing of the signal and increased computational load.

[0093] In this embodiment, the number of decomposition layers JJ is adaptively determined based on the length and sampling frequency of the fused spectral signal. Let the signal length be LL (number of pixels), and the maximum number of decomposition layers be... To balance denoising effectiveness and computational efficiency, the number of decomposition layers is set as follows:

[0094] That is, it can be decomposed into a maximum of 6 layers to ensure that there are still enough data points in the highest frequency subband for noise statistics.

[0095] S403: Wavelet decomposition process The Mallat pyramid algorithm is used to fuse the spectral signal S. fusion (λ) Perform multi-scale wavelet decomposition. Let the signal length be K and the number of decomposition levels be J. The decomposition process is as follows: Initialization: Let the approximation coefficient a0 of the 0th layer be the original fused spectral signal S. fusion .

[0096] Iterative decomposition: For j=1,2,…,J, perform the following operations: j 1st layer approximation coefficient a j 1. Passed through low-pass decomposition filters respectively and high-pass decomposition filter ; The filtering results are downsampled (one sample is taken at every other point) to obtain the approximate coefficients a of the j-th layer. j and detail coefficient d j :

[0097] in, and These are the decomposition filter coefficients corresponding to the wavelet basis functions.

[0099] After J-level decomposition, the original signal is represented as: That is, an approximation coefficient of the Jth layer. (Reflecting the low-frequency profile of the signal) and J-layer detail coefficients {d1,d2,…,d J (Reflects detailed information about the signal from high frequency to intermediate frequency). Here, d1 corresponds to the highest frequency sub-band, mainly containing noise components; d... J The corresponding lowest frequency detail subband contains information about the signal's edge and peak shape.

[0100] S404: Storage of decomposition coefficients After completing the wavelet decomposition, the obtained approximate coefficients and the detail coefficients d1, d2, ..., d for each layer J These coefficients are stored in memory and used as input for subsequent adaptive thresholding denoising. They will then be thresholded in step S304 to suppress noise components.

[0101] It should be noted that in practical engineering implementations, especially when deployed on FPGAs or embedded platforms, a pipelined approach can be used to perform wavelet decomposition in parallel to improve processing speed. For example, while acquiring the fused spectral signal of frame t, the wavelet decomposition of frame t can be performed simultaneously. A single frame of fused signal is decomposed into wavelet fragments to enable a streamlined acquisition and processing process, meeting real-time requirements.

[0102] In one embodiment of the present invention, based on step S5, the following will provide a possible embodiment and its specific implementation will be described in a non-limiting manner, such as... Figure 3 As shown.

[0103] S501. Adaptively determine the denoising threshold based on the local characteristics of signals at each scale.

[0104] Signal energy distribution varies across different scales. Scales dominated by signal have higher energy levels and should use smaller thresholds to preserve details; scales dominated by noise have lower energy levels and should use larger thresholds to suppress noise. Based on this idea, this embodiment employs an adaptive thresholding method based on energy proportion.

[0105] (1) Noise standard deviation estimation First, the noise standard deviation σ in the signal is estimated. Since the highest frequency detail coefficients d1 of the wavelet decomposition mainly contain noise components, the median absolute deviation (MAD) of d1 is used for estimation: The division by 0.6745 is to ensure that the estimated value is consistent with the standard deviation of Gaussian noise. This estimation method is insensitive to outliers and can accurately reflect the noise level.

[0106] (2) Calculate the basic threshold The base threshold is calculated using the Donoho-Johnstone universal threshold formula: Where N is the length of the fused spectral signal (i.e., the number of pixels). This threshold is the optimal threshold in the sense of minimax, and can effectively remove Gaussian white noise.

[0107] (3) Calculate the energy proportion factor at each scale For the detail coefficient d of the j-th level j Calculate its energy Ej and energy proportion factor p j :

[0108]

[0109] Where J is the total number of wavelet decomposition layers. Energy proportion factor p j This reflects the proportion of energy that the detail coefficients of layer j account for in the entire detail section. j The larger the value, the more signal energy the layer contains, and the more likely it is to be a signal-dominant scale; p j The smaller the value, the more likely the layer is to contain noise.

[0111] (4) Construct an adaptive threshold function Combining the base threshold with the energy proportion factor, we obtain the adaptive threshold for the j-th layer:

[0112] in, For energy percentage factor The monotonically decreasing function allows for the use of smaller thresholds at scales with high energy percentages to preserve signal details, and larger thresholds at scales with low energy percentages to enhance noise suppression.

[0113] This embodiment provides two exemplary monotonically decreasing functions: Function Form 1 (Inverse Proportional): Wherein, γ is an adjustable attenuation coefficient, typically ranging from 1 to 1. 10 The larger γ is, the stronger the regulatory effect of the energy percentage on the threshold. For example, let's take γ = 5.

[0114] Functional Form Two (Exponential Decreasing Type): Where η is an adjustable exponential decay coefficient, typically ranging from 2 to 8. For example, η = 4.

[0115] Both functions satisfy: when When it approaches 0, When it approaches 1, the threshold is close to the base threshold; when When it approaches 1, If the value is significantly less than 1, the threshold is greatly reduced.

[0116] S502: Thresholding the decomposition coefficients using an adaptive threshold. Obtain the adaptive threshold λ at each scale j Then, the detail coefficients of each layer are thresholded. This embodiment preferentially uses a soft thresholding function because it can obtain a smoother reconstructed signal: Where sgn() is the sign function, These are the detail coefficients after thresholding. The soft thresholding function shrinks the coefficients towards zero, effectively suppressing noise and preventing pseudo-Gibberish oscillations in the reconstructed signal.

[0117] For the approximation coefficient a J Since it mainly reflects the low-frequency profile (baseline) of the spectrum, it is usually not thresholded, or only smoothed by a very small amount to maintain the overall shape of the spectrum.

[0118] S503: Wavelet reconstruction yields the denoised spectral signal. After thresholding, wavelet basis functions and reconstruction filters corresponding to wavelet decomposition are used to perform inverse wavelet transform on the processed coefficients to reconstruct the denoised spectral signal.

[0119] Let the approximation coefficient of the Jth layer be a. J The detail coefficients after thresholding at each layer are: The reconstruction process is as follows: Initialization: Let the reconstruction coefficient c of the Jth layer be... J =a J .

[0120] Iterative reconstruction: for j=J, J 1,…,1, perform the following operations: For the reconstruction coefficient c of the j-th layer j Perform upsampling (insert zeros between adjacent points); The upsampled results are then passed through a low-pass reconstruction filter h and a high-pass reconstruction filter g, respectively. Add the filtering results together to obtain the j-th... Level 1 reconstruction coefficient c j 1:

[0121] After J-layer iterative reconstruction, the final denoised spectral signal S is obtained. denoised (λ).

[0122] Stability tests were conducted using the method of this patent and without the method of this patent on a single device, with 28 tests performed over a week (automatic testing every 6 hours). The relative standard deviation (RSD) of the Ga ion content was as follows: Figure 4 and Figure 5 As shown. Figure 4 The RSD of samples that did not undergo stability testing using the method described in this patent was 6.7%. Figure 5 The RSD of the stability test using this patented method was 4.3%. The RSD of the test results not using this patented method did not meet the requirement of RSD≤5% in the standard (HJ 353-2019 "Technical Specification for Installation of Online Monitoring System for Water Pollution Sources").

[0123] Please refer to Figure 6 The system of this invention can be deployed in a stand-alone spectrometer, an industrial process online monitoring system, or a cloud computing analysis platform: in stand-alone mode, fast frame fusion and denoising are performed through a local FPGA; in embedded mode, online signal correction is achieved through the collaborative work of an industrial control motherboard and a spectral acquisition card; in cloud platform mode, the acquired data can be uploaded to a remote server, and the algorithm of this invention can be executed in parallel by distributed nodes to achieve unified signal optimization and model update across devices.

[0124] In some embodiments, the system may include multiple functional modules composed of computer program segments. The computer programs for each segment in the adaptive denoising system for glow discharge spectral signals may be stored in the memory of a computer device and executed by at least one processor to perform (see details). Figure 1 (Description) The adaptive denoising function of glow discharge spectral signals.

[0125] In this embodiment, the adaptive denoising system for the glow discharge spectrum signal can be divided into multiple functional modules according to the functions it performs. A module, as referred to in this invention, is a series of computer program segments that can be executed by at least one processor and perform a fixed function, stored in memory. In this embodiment, the functions of each module will be detailed in subsequent embodiments.

[0126] The signal acquisition module is used to acquire continuous multi-frame spectral signals generated during the glow discharge process; The stability assessment module is used to assess the stability of each frame of spectral signal and calculate a stability score that characterizes the quality of the signal in that frame. The signal overlay module is used to adjust the number N of spectral frames used for overlay according to the stability score, and to perform weighted fusion of the N spectral signals selected based on the stability score to obtain a fused spectral signal, wherein N is an integer greater than 1, and the number of frames N is positively correlated with the stability score. The wavelet decomposition module is used to perform wavelet decomposition on the fused spectral signal to obtain decomposition coefficients at multiple scales. The wavelet denoising module is used to adaptively determine the denoising threshold based on the local features of the signal at each scale, use the denoising threshold to threshold the decomposition coefficients, and perform wavelet reconstruction on the thresholded decomposition coefficients to obtain a denoised spectral signal with a high signal-to-noise ratio.

[0127] Figure 7 The adaptive denoising method for glow discharge spectral signals provided in the embodiments of this application can be applied to devices. Those skilled in the art will understand that the device structures involved in the embodiments of this invention do not constitute a limitation on the device. A device may include more or fewer components than illustrated, or combine certain components, or have different component arrangements. In the embodiments of this invention, the device includes, but is not limited to, laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The device may also represent various forms of mobile systems, such as personal digital processing units, cellular phones, smartphones, wearable devices, and other similar computing systems. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the embodiments of this application described and / or claimed herein.

[0128] The device 700 may include a processor 710, a memory 720, and a communication unit 730. These components communicate via one or more buses. Those skilled in the art will understand that the server structure shown in the figures does not constitute a limitation of the present invention. It may be a bus topology or a star topology, and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0129] The memory 720 can be used to store execution instructions of the processor 710. The memory 720 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. When the execution instructions in the memory 720 are executed by the processor 710, the device 700 is able to perform some or all of the steps in the above method embodiments.

[0130] The processor 710 serves as the control center of the storage device, connecting various parts of the electronic device via various interfaces and lines. It executes software programs and / or modules stored in the memory 720, and calls data stored in the memory to perform various functions of the electronic device and / or process data. The processor can be composed of integrated circuits (ICs), such as a single packaged IC or multiple packaged ICs with the same or different functions connected together. For example, the processor 710 may consist only of a central processing unit (CPU). In this embodiment of the invention, the CPU may have a single processing core or include multiple processing cores.

[0131] The communication unit 730 is used to establish a communication channel, enabling the storage device to communicate with other devices. It can receive user data sent by other devices or send user data to other devices.

[0132] The present invention also provides a computer medium, wherein the computer medium may store a program, which, when executed, may include some or all of the steps provided in the embodiments of the present invention. The medium may be a magnetic disk, an optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0133] Those skilled in the art will clearly understand that the techniques in the embodiments of the present invention can be implemented using software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solutions in the embodiments of the present invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product is stored in a medium such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, or any other medium capable of storing program code. It includes several instructions to cause a computer device (which may be a personal computer, a server, or a second device, network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.

[0134] The same or similar parts between the various embodiments in this specification can be referred to mutually. In particular, the device embodiments are basically similar to the method embodiments, so the description is relatively simple, and the relevant parts can be referred to the description in the method embodiments.

[0135] In the embodiments provided by this invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between systems or modules may be electrical, mechanical, or other forms.

[0136] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0137] In addition, the functional modules in the various embodiments of the present invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0138] Although the present invention has been described in detail with reference to the accompanying drawings and preferred embodiments, the present invention is not limited thereto. Various equivalent modifications or substitutions can be made to the embodiments of the present invention by those skilled in the art without departing from the spirit and essence of the invention, and such modifications or substitutions should all be within the scope of the present invention. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should also be covered within the protection scope of the present invention.

Claims

1. An adaptive denoising method for glow discharge spectral signals, characterized in that, include: Acquire continuous multi-frame spectral signals generated during the glow discharge process; For each frame of spectral signal, a stability assessment is performed, and a stability score characterizing the quality of that frame of signal is calculated. Based on the stability score, the number N of spectral frames used for superposition is adjusted, and the N spectral signals selected based on the stability score are weighted and fused to obtain a fused spectral signal, where N is an integer greater than 1, and the number of frames N is positively correlated with the stability score. Wavelet decomposition is performed on the fused spectral signal to obtain decomposition coefficients at multiple scales; The denoising threshold is adaptively determined based on the local features of the signal at each scale. The decomposition coefficients are then thresholded using the denoising threshold, and wavelet reconstruction is performed on the thresholded decomposition coefficients to obtain a denoised spectral signal with a high signal-to-noise ratio.

2. The method according to claim 1, characterized in that, For each frame of spectral signal, a stability assessment is performed, and a stability score characterizing the signal quality of that frame is calculated, including: Extract multiple stability feature indices from each frame of spectral signal, and construct the feature vector of the frame signal based on the multiple stability feature indices; A weighted Mahalanobis distance stability discrimination algorithm is used to calculate the Mahalanobis distance between the feature vector and the preset stable reference vector; The Mahalanobis distance is mapped to the stability score, and the smaller the Mahalanobis distance, the higher the stability score. The stability characteristics include spectral peak drift, which characterizes the stability of spectral line position; light intensity fluctuation rate, which characterizes the fluctuation of signal intensity; discharge current stability, which characterizes the stability of discharge state; spectral peak centroid drift, which characterizes the degree of spectral line shape retention; baseline slope change, which characterizes background changes; and high-frequency noise energy density, which characterizes noise level.

3. The method according to claim 1, characterized in that, Based on the stability score, the number N of spectral frames used for stacking is adjusted, including: Establish a mapping relationship between stability score and number of stacked frames, wherein the mapping relationship is monotonically increasing, such that when the stability score increases, the number of stacked frames N increases, and when the stability score decreases, the number of stacked frames N decreases. The mapping relationship is a piecewise linear mapping function, which divides the stability score interval into multiple sub-intervals. Each sub-interval corresponds to a number of superimposed frames, and the number of superimposed frames N is within a preset range.

4. The method according to claim 1, characterized in that, Based on the stability score, the number N of spectral frames used for stacking is adjusted, including: Based on historical stability score sequences, multi-scale trend features containing stability scores in the near future are extracted. The multi-dimensional prediction features are mapped to corresponding fuzzy sets to obtain fuzzy input variables; The corresponding rules in the fuzzy rule base are activated according to the fuzzy input variables, and fuzzy inference aggregation is performed to obtain the output fuzzy set; wherein, the fuzzy rule base contains multiple fuzzy inference rules, and each rule contains antecedent conditions and consequent conclusions; The output fuzzy set is converted into a precise overlay frame number adjustment amount; The fuzzy rule base includes at least one of the following rules: Active gain rule under positive prediction: If the current frame stability score is medium, the prediction stability score is rising, and the fluctuation state is stable, then the number of frames to be adjusted is increased slightly. Risk avoidance rules under negative prediction: If the current frame stability score is high, the prediction stability score is decreasing, and the short-term trend strength is medium, then the superimposed frame adjustment amount is slightly reduced. Conservative strategy rule under high volatility: If the volatility is severe, the amount of frame adjustment is greatly reduced. Strengthening rules under trend confirmation: If the current frame stability score is low, the short-term trend strength is strong, the short-term trend slope is negative, and the predicted stability score is decreasing, then the amount of frame adjustment is significantly reduced.

5. The method according to claim 1, characterized in that, The N frames of spectral signals selected based on the stability score are weighted and fused to obtain a fused spectral signal, including: Based on the stability score of each frame of spectral signal, the N frames of spectral signals with the highest stability score are selected from the multiple frames of spectral signals continuously acquired within the sliding window as the frames to be fused; the length of the sliding window is adaptively adjusted according to the degree of fluctuation of the discharge state. Based on the stability score of each frame of spectral signal, the fusion weight corresponding to that frame of signal is calculated, wherein the fusion weight is positively correlated with the stability score; Based on the fusion weights, the N frames to be fused are weighted and averaged to obtain the fused spectral signal.

6. The method according to claim 1, characterized in that, The denoising threshold is adaptively determined based on the local features of the signals at each scale, and the decomposition coefficients are thresholded using the denoising threshold, including... Based on the local energy distribution of signals at each scale, calculate the energy proportion factor corresponding to that scale; The basic threshold model is modified based on the energy proportion factor to obtain the adaptive denoising threshold for this scale. The detail coefficients at this scale are thresholded using the adaptive denoising threshold.

7. The method according to claim 1, characterized in that, The basic threshold model is modified based on the energy proportion factor to obtain the adaptive denoising threshold for this scale, including: The formula for calculating the basic threshold is: in, The noise standard deviation is estimated by the median absolute deviation of the highest frequency subband detail coefficients obtained from wavelet decomposition; N is the number of fused frames. By combining the basic threshold with a function of the energy proportion factor, an adaptive threshold is obtained. ,in For energy percentage factor The monotonically decreasing function allows for the use of smaller thresholds at scales with higher signal energy proportions to preserve details, and larger thresholds at scales with lower signal energy proportions to suppress noise.

8. An adaptive denoising system for glow discharge spectral signals, characterized in that, include: The signal acquisition module is used to acquire continuous multi-frame spectral signals generated during the glow discharge process; The stability assessment module is used to assess the stability of each frame of spectral signal and calculate a stability score that characterizes the quality of the signal in that frame. The signal overlay module is used to adjust the number N of spectral frames used for overlay according to the stability score, and to perform weighted fusion of the N spectral signals selected based on the stability score to obtain a fused spectral signal, wherein N is an integer greater than 1, and the number of frames N is positively correlated with the stability score. The wavelet decomposition module is used to perform wavelet decomposition on the fused spectral signal to obtain decomposition coefficients at multiple scales. The wavelet denoising module is used to adaptively determine the denoising threshold based on the local features of the signal at each scale, use the denoising threshold to threshold the decomposition coefficients, and perform wavelet reconstruction on the thresholded decomposition coefficients to obtain a denoised spectral signal with a high signal-to-noise ratio.

9. An adaptive noise reduction device for glow discharge spectral signals, characterized in that, include: Memory used to store the adaptive denoising program for glow discharge spectral signals; A processor, configured to implement the steps of the adaptive denoising method for the glow spectrum signal as described in any one of claims 1-7 when executing the adaptive denoising program for the glow spectrum signal.

10. A computer-readable storage medium storing a computer program, characterized in that, The readable storage medium stores an adaptive denoising program for the glow discharge spectrum signal, which, when executed by a processor, implements the steps of the adaptive denoising method for the glow discharge spectrum signal as described in any one of claims 1-7.

Images