Pulse signal data processing method, system and terminal

Abstract

The invention provides a pulse signal data processing method and system, and a terminal, and the method comprises the steps: carrying out the unsupervised clustering of the pulse height value of each pulse signal through a Gaussian mixture model, and calculating a target contraction covariance matrix of a noise signal data set through employing a Leidoit-Wolf contraction covariance estimation method, so as to achieve the calculation of the target contraction covariance matrix based on a minimum Mahalanobis distance criterion. The optimal pulse height fitting factor of each pulse signal is calculated, and the pulse height value of each pulse signal is corrected according to the optimal pulse height fitting factor, so that the technical problem that the existing pulse signal noise suppression technology is difficult to meet various noise suppression requirements of multi-energy radiation and dynamic radiation scenes is solved; the method improves the correction precision of the pulse height value and the signal-to-noise ratio of the pulse signal, can improve the discrimination capability of the superconducting phase-change edge detector on the pulse signal and the energy analysis accuracy of the to-be-detected sample when being applied to the superconducting phase-change edge detector to detect the to-be-detected sample, and guarantees the high-energy resolution of the superconducting phase-change edge detector.

Description

Technical Field

[0001] The present application relates to the field of signal processing technology, and in particular to a pulse signal data processing method, system and terminal. Background Art

[0002] A superconducting transition-edge sensor (TES) is a highly sensitive thermal detector based on the dramatic change in resistance of superconducting materials near their critical temperature. It is widely used in fields such as particle physics, X-ray spectroscopy, and nuclear radiation detection. The TES used for X-ray spectroscopy can be used as an energy-dispersive spectroscopy device. Based on the elemental analysis principle of X-ray fluorescence, TES performs qualitative and quantitative analysis by measuring the energy of characteristic X-rays released by the sample after excitation. TESs have high energy resolution, reaching eV or even sub-eV levels, which is two orders of magnitude higher than energy-dispersive measurement devices based on semiconductor detectors. Furthermore, compared to wavelength-dispersive spectroscopy devices based on gratings or crystals, TESs can integrate multiple pixel units, resulting in higher collection efficiency and a wider energy coverage range. Therefore, TESs play an important role in X-ray spectroscopy research.

[0003] The core operating principle of a superconducting phase-change edge detector is to measure the interaction between characteristic X-rays emitted by a sample after excitation and the detector. The energy of the characteristic X-rays is absorbed by the superconducting material, causing a local temperature change in the superconducting material, which in turn causes a change in its resistance. This in turn changes the current signal flowing through the detector, generating multiple pulse signals. The height of each pulse signal, i.e., the amplitude of the current signal change, is proportional to the photon energy of the incident characteristic X-ray. Therefore, by measuring the pulse signal height, the energy spectrum data of the characteristic X-rays can be obtained. However, in practical applications, the pulse signals obtained by superconducting phase-change edge detectors are susceptible to various types of noise interference. Specifically, on the one hand, noise interference can cause an unstable pulse signal baseline, resulting in cumulative errors in the extracted pulse signal height, causing peak shifts or broadening of the characteristic X-ray energy spectrum data. On the other hand, random noise superimposed on the pulse signal can cause the formation of spurious peaks, reducing the signal-to-noise ratio of the pulse signal. Therefore, the presence of noise will cause uncertainty in the measured pulse signal height, which in turn seriously affects the accuracy of the characteristic X-ray energy spectrum data and the energy resolution of the superconducting phase-change edge detector. In other words, the energy resolution of a superconducting phase-change edge detector directly depends on the noise suppression capability of the pulse signal.

[0004] Currently, existing technologies primarily rely on preprocessing pulse signals to improve the analysis quality of pulse signals, suppress noise in pulse signals, and enhance the energy resolution of superconducting phase-change edge detectors. Pulse signal preprocessing methods primarily include the following: ① Baseline correction: This technique uses a sliding average filter (such as a Savitzky-Golay filter) to suppress low-frequency baseline drift. However, this technique is inadequate for suppressing high-frequency noise, and the setting of the filter parameters relies on manual experience. ② Pulse alignment and averaging: This technique time-aligns and superimposes repeated pulse signals. However, this technique is only applicable to stable radiation sources and has poor adaptability to dynamic radiation scenarios. ③ Wavelet denoising: This technique uses multi-scale decomposition to filter out high-frequency noise, decomposing the signal or image into subbands of different scales to specifically address high-frequency noise components. However, this technique requires a pre-set scale threshold and has limited effectiveness in suppressing non-stationary noise. ④ Statistical analysis and clustering: This technique uses a Gaussian mixture model (GMM) to distinguish between signal and noise categories. However, the traditional GMM covariance matrix estimate is susceptible to interference from small samples or high-dimensional noise, resulting in insufficient classification accuracy. 5. Optimal filtering technology: Based on the premise that the pulse signal shape does not change with the signal size, a fixed pulse signal template is constructed for fitting. However, this technology does not adequately consider the changes in pulse signal shape due to nonlinear effects in the case of multi-energy radiation. Therefore, existing pulse signal noise suppression technology has many drawbacks and still faces challenges in complex noise environments. It is difficult to meet the diverse noise suppression requirements of multi-energy radiation and dynamic radiation scenarios. Summary of the Invention

[0005] In view of the shortcomings of the existing technology described above, the purpose of this application is to provide a pulse signal data processing method, system and terminal to solve the technical problem that the existing pulse signal noise suppression technology is difficult to meet the various noise suppression requirements of multi-energy radiation and dynamic radiation scenarios.

[0006] To achieve the above-mentioned purpose and other related purposes, the first aspect of the present application provides a pulse signal data processing method, which includes: collecting multiple measurement signals output by a superconducting phase change edge detector when detecting a sample to be tested, and screening the noise signal and pulse signal in each measurement signal, extracting the pulse height value of each pulse signal to construct a noise signal data set, a pulse signal data set and a pulse height data set; using a Gaussian mixture model to perform cluster analysis on the pulse height data set, dividing the pulse height data set into multiple pulse height data subsets, and according to each pulse height data subset, dividing the pulse signal data set into multiple pulse signal data subsets; averaging each pulse signal in each pulse signal data subset to obtain an average pulse template for each pulse signal data subset; calculating the target shrinkage covariance matrix of the noise signal data set, and according to the target shrinkage covariance matrix, calculating the optimal pulse height fitting factor with the minimum Mahalanobis distance between each pulse signal and its corresponding average pulse template, so as to correct each pulse height value, obtain the corresponding corrected pulse height value, and construct a corrected pulse height data set.

[0007] In some embodiments of the first aspect of the present application, a Gaussian mixture model is used to perform cluster analysis on the pulse height dataset, and the method of dividing the pulse height dataset into multiple pulse height data subsets includes: predicting the number of energy peaks of the sample to be tested based on the elemental composition of the sample to be tested to construct a Gaussian mixture model of the pulse height dataset; wherein the number of each Gaussian distribution in the Gaussian mixture model is the same as the number of energy peaks of the sample to be tested; using the expectation maximization algorithm to step-by-step calculate the posterior probability that each pulse height value belongs to each Gaussian distribution, and iteratively optimize the weight, mean and covariance matrix of each Gaussian distribution until each Gaussian distribution reaches convergence, and obtain corresponding multiple converged Gaussian distributions; calculating the posterior probability that each pulse height value belongs to each convergent Gaussian distribution, and assigning each pulse height value to the convergent Gaussian distribution with the largest posterior probability, thereby dividing the pulse height dataset into multiple pulse height data subsets.

[0008] In some embodiments of the first aspect of the present application, a calculation method for obtaining an average pulse template for each pulse signal data subset includes: Among them, x temp,j is the average pulse template of the jth pulse signal data subset, x ij is the i-th pulse signal in the j-th pulse signal data subset, C j is the number of pulse signals in the j-th pulse signal data subset.

[0009] In some embodiments of the first aspect of the present application, the method for calculating the target shrinkage covariance matrix of the noise signal data set includes: constructing a sample matrix of the noise signal data set and generating a corresponding sample covariance matrix; constructing a target matrix of the noise signal data set, and shrinking the sample covariance matrix toward the target matrix, and calculating the optimal shrinkage coefficient of the sample covariance matrix by minimizing the mean square error between the shrinkage covariance matrix and the true covariance matrix; calculating the target shrinkage covariance matrix of the noise signal data set based on the optimal shrinkage coefficient; wherein the calculation formula of the optimal shrinkage coefficient is: λ opt ≈(∑ i,j Var(S ij )) / ∑ i,j (S ij -T ij ) 2 ; The calculation formula of the target shrinkage covariance matrix is: Where S is the sample covariance matrix of the noise signal dataset, S ij is the i-th row and j-th column element of the sample covariance matrix, Var(S ij ) is the variance of the elements in the i-th row and j-th column of the sample covariance matrix, Tr(S) is the trace of the sample covariance matrix, I is the identity matrix, d is the dimension of the sample covariance matrix; T is the target matrix of the noise signal data set, T ij is the element in the i-th row and j-th column of the target matrix; opt is the optimal shrinkage coefficient, Target shrinkage covariance matrix for the noisy signal dataset.

[0010] In some embodiments of the first aspect of the present application, a method for calculating the optimal pulse height fitting factor with the minimum Mahalanobis distance between each pulse signal and its corresponding average pulse template according to the target shrinkage covariance matrix includes: Among them, α opt,ij is the optimal pulse height fitting factor of the i-th pulse signal in the j-th pulse signal data subset, x ij is the i-th pulse signal in the j-th pulse signal data subset, x temp,j is the average pulse template of the j-th pulse signal data subset, Target shrinkage covariance matrix for the noisy signal dataset.

[0011] In some embodiments of the first aspect of the present application, each pulse height value is corrected according to each optimal pulse height fitting factor, and a method for obtaining a corresponding corrected pulse height value includes: ij =α opt,ij pj ; Among them, p ij is the corrected pulse height value of the i-th pulse signal in the j-th pulse signal data subset, α opt,ij is the optimal pulse height fitting factor of the i-th pulse signal in the j-th pulse signal data subset, p j is the pulse height value of the average pulse template of the j-th pulse signal data subset.

[0012] In some embodiments of the first aspect of the present application, the method for screening out noise signals and pulse signals in each measurement signal includes: setting a signal amplitude threshold, and treating the signal segment in each measurement signal whose signal amplitude is less than or equal to the signal amplitude threshold as a noise signal, and the signal segment whose signal amplitude is greater than the signal amplitude threshold as a pulse signal, thereby splitting each measurement signal into a noise signal and a pulse signal, and screening out the noise signal and pulse signal in each measurement signal.

[0013] In some embodiments of the first aspect of the present application, the pulse signal data processing method also includes: drawing a pulse height distribution histogram of the corrected pulse height data set to perform a visual analysis of the corrected pulse signal height data set; obtaining energy spectrum data of the sample to be tested based on the positive proportional linear relationship between the pulse height and the incident particle energy of the superconducting phase change edge detector, and drawing an energy spectrum diagram of the sample to be tested.

[0014] To achieve the above-mentioned purpose and other related purposes, the second aspect of the present application provides a pulse signal data processing system, which includes: a data acquisition module for collecting multiple measurement signals output by a superconducting phase change edge detector when detecting a sample to be tested, and screening the noise signal and pulse signal in each measurement signal, extracting the pulse height value of each pulse signal to construct a noise signal data set, a pulse signal data set and a pulse height data set; a data classification module, connected to the data acquisition module, for performing cluster analysis on the pulse height data set using a Gaussian mixture model, dividing the pulse height data set into multiple pulse height data subsets, and classifying the pulse height data set according to the pulse height data subsets. A set of pulse signal datasets is divided into multiple pulse signal data subsets; a signal averaging processing module is connected to the data classification module, and is used to average each pulse signal in each pulse signal data subset to obtain an average pulse template for each pulse signal data subset; a pulse height correction module is connected to the signal averaging processing module, and is used to calculate the target shrinkage covariance matrix of the noise signal dataset, and calculate the optimal pulse height fitting factor with the minimum Mahalanobis distance between each pulse signal and its corresponding average pulse template based on the target shrinkage covariance matrix, so as to correct each pulse height value, obtain the corresponding corrected pulse height value, and construct a corrected pulse height dataset.

[0015] To achieve the above-mentioned purpose and other related purposes, the third aspect of the present application provides a pulse signal data processing terminal, which includes: a processor and a memory; the memory is used to store computer programs and data; the processor is used to execute the computer program stored in the memory, so that the pulse signal data processing terminal executes the pulse signal data processing method described in any one of the above embodiments.

[0016] As described above, the present application provides a pulse signal data processing method, system and terminal, which performs unsupervised clustering of the pulse height values of each pulse signal through a Gaussian mixture model, and adopts the Ledoit-Wolf shrinkage covariance estimation method to calculate the target shrinkage covariance matrix of the noise signal data set, so as to calculate the optimal pulse height fitting factor of each pulse signal based on the minimization of the Mahalanobis distance criterion, and correct the pulse height value of each pulse signal accordingly; therefore, the present application has the following beneficial effects: it solves the technical problem that the existing pulse signal noise suppression technology is difficult to meet the various noise suppression requirements of multi-energy radiation and dynamic radiation scenarios, improves the correction accuracy of the pulse height value and the signal-to-noise ratio of the pulse signal, and when applied to the superconducting phase change edge detector to detect the sample to be tested, it can improve the superconducting phase change edge detector's ability to identify the pulse signal and the energy resolution accuracy of the sample to be tested, thereby ensuring the high energy resolution of the superconducting phase change edge detector. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 Shown is a flow chart of a pulse signal data processing method according to an embodiment of the present application.

[0018] Figure 2 Shown is a histogram of the original pulse height distribution of a TES-X-ray pulse signal in a specific embodiment of the present application.

[0019] Figure 3 Shown is a schematic diagram of the clustering results of the TES-X-ray pulse height dataset in a specific embodiment of the present application.

[0020] Figure 4 Shown is a corrected pulse height distribution histogram of a TES-X-ray pulse signal in a specific embodiment of the present application.

[0021] Figure 5 Shown is a structural diagram of a pulse signal data processing system in one embodiment of the present application.

[0022] Figure 6 Shown is a structural diagram of a pulse signal data processing terminal in one embodiment of the present application. DETAILED DESCRIPTION

[0023] The following describes the embodiments of the present application through specific examples. Those skilled in the art can easily understand the other advantages and effects of the present application from the content disclosed in this specification. The present application can also be implemented or applied through other different specific embodiments. The details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present application. It should be noted that the following embodiments and features in the embodiments can be combined with each other unless they conflict.

[0024] In the embodiments of the present application, terms such as "first" and "second" are used to distinguish between identical or similar items with substantially the same functions and effects. For example, the first pulse signal data subset and the second pulse signal data subset are merely used to distinguish between different pulse signal data subsets and do not limit their order. Those skilled in the art will understand that terms such as "first" and "second" do not limit the quantity or execution order, and that terms such as "first" and "second" do not necessarily mean that they are different.

[0025] Superconducting transition-edge sensors (TES) are widely used in fields such as particle physics, X-ray energy spectrum measurement, and nuclear radiation detection. When using a TES to detect a sample, an external source, such as an X-ray source or infrared light, is first used to illuminate the sample to stimulate its energy. This energy is then absorbed by the conductive material of the TES, causing a local temperature change in the superconducting material, which in turn causes a change in its resistance. This changes the current signal flowing through the detector, forming multiple pulse signals for obtaining energy spectrum data of the sample. However, in practical applications, the output pulse signals are often affected by noise signals. To obtain high-precision energy spectrum data, the output pulse signals must be processed to suppress noise and calibrate the pulse height of each pulse signal. Therefore, the present application provides a pulse signal data processing method, system and terminal, which performs unsupervised clustering of the pulse height values of each pulse signal through a Gaussian mixture model, and adopts the Ledoit-Wolf shrinkage covariance estimation method to balance the sample covariance and the unit matrix in a high-dimensional small sample noise signal data set, avoid the ill-conditioned problem of the covariance matrix, and improve the robustness of the noise multi-dimensional feature space modeling; at the same time, based on the minimization of the Mahalanobis distance criterion, an optimal pulse height fitting factor is constructed to achieve correction of the pulse height and effectively suppress the energy spectrum broadening caused by noise; thereby solving the technical problem that the existing pulse signal noise suppression technology is difficult to meet the various noise suppression requirements of multi-energy radiation and dynamic radiation scenarios, significantly improving the ability to identify pulse signals and the accuracy of energy resolution of the sample to be tested, and can be widely used in particle physics, nuclear radiation detection, and high-resolution X-ray energy spectrum measurement and analysis.

[0026] In order to make the invention objectives, technical solutions and advantages of this application more clearly understood, the following embodiments and the accompanying drawings are used to further explain the technical solutions in the embodiments of this application. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0027] like Figure 1 FIG. 1 is a flow chart showing a pulse signal data processing method according to an embodiment of the present application. The pulse signal data processing method according to the embodiment includes steps S1 to S4.

[0028] Step S1: Collect multiple measurement signals output by the superconducting phase change edge detector when detecting the sample to be tested, filter the noise signal and pulse signal in each measurement signal, and extract the pulse height value of each pulse signal to construct a noise signal data set, a pulse signal data set, and a pulse height data set.

[0029] It should be understood that a pulse signal is an electrical signal with extreme duration and abrupt characteristics, which can manifest as sudden changes in voltage or current. Its core characteristics include rapid rising / falling edges, short duration, and aperiodicity. Therefore, based on the abrupt amplitude characteristics of pulse signals, in one embodiment, the method for screening noise signals and pulse signals from each measurement signal includes: setting a signal amplitude threshold, and treating signal segments in each measurement signal with a signal amplitude less than or equal to the signal amplitude threshold as noise signals, and signal segments with a signal amplitude greater than the signal amplitude threshold as pulse signals, thereby splitting each measurement signal into a noise signal and a pulse signal, and screening out the noise and pulse signals from each measurement signal.

[0030] It should be noted that users can also use dynamic threshold detection algorithms, filter filtering, and machine learning algorithms to filter noise signals and pulse signals in the measurement signal for subsequent pulse signal data processing, which is not specifically limited in this application.

[0031] In this embodiment, each measurement signal can be collected based on a certain frequency, and each measurement signal and each pulse signal and each noise signal finally obtained by screening are the measurement values of the signal amplitude at different times, that is, each measurement signal, each pulse signal and each noise signal is an array of a certain length, and the elements of the array are the measurement values of the signal amplitude at different times.

[0032] In a preferred embodiment, each pulse signal and each noise signal can be defined as an array of length n, with the pulse signal dataset being a collection of multiple pulse signal arrays, and the noise signal dataset being a collection of multiple noise signal arrays. Furthermore, the maximum signal amplitude value in each pulse signal array is extracted as the pulse height value of the pulse signal, thereby constructing a pulse height dataset that aggregates multiple pulse height values.

[0033] Step S2: Perform cluster analysis on the pulse height dataset using a Gaussian mixture model to divide the pulse height dataset into multiple pulse height data subsets, and divide the pulse signal dataset into multiple pulse signal data subsets according to each pulse height data subset.

[0034] In one embodiment, cluster analysis is performed on the pulse height dataset using a Gaussian mixture model to divide the pulse height dataset into a plurality of pulse height data subsets, including the following steps.

[0035] ① According to the elemental composition of the sample to be tested, the number of energy peaks of the sample to be tested is predicted to construct a Gaussian mixture model of the pulse height data set.

[0036] It should be understood that energy peaks (such as photoelectron peaks in XPS and characteristic peaks in gamma spectra) generally correspond to electronic transitions or nuclear decays of specific elements in the sample to be tested. The number of energy peaks in the sample to be tested depends on: the elemental composition of the sample to be tested, the chemical state differences, and the detection instrument. Specifically, each element may produce multiple characteristic peaks. For a sample to be tested containing known elements, the theoretical number of characteristic peaks is calculated as follows:

[0037] N 理论 =∑Number of K / L series peaks of each element × number of chemical state splittings; Formula (1)

[0038] Among them, N 理论 is the theoretical number of energy peaks of the sample to be tested. K series peaks include Kα peak and Kβ peak, L series peaks include Lα peak, Lβ peak and Lβ peak. The chemical state splitting number is the number of peaks that may be split in different valence states or chemical environments of the same element.

[0039] The calculated theoretical peak number of the sample to be tested is corrected for interference peaks, so the calculation formula for the predicted number of energy peaks of the sample to be tested is:

[0040] N 预测 =N 理论 +N 逃逸峰 +N 和峰 ; Formula (2)

[0041] Among them, N 预测 is the predicted number of energy peaks of the sample to be tested; N 理论is the theoretical energy peak number of the sample to be tested; N 逃逸峰 is the number of escape peaks, each main peak may produce one escape peak; N 和峰 is the sum peak number, which is a false peak produced by the superposition of two photons at high counting rates.

[0042] In this embodiment, the data distribution of the pulse height dataset is modeled using a Gaussian mixture model, that is, it is assumed that each pulse height value in the pulse height dataset conforms to multiple Gaussian distributions, thereby performing cluster analysis on the pulse height dataset.

[0043] Specifically, the probability density function of the Gaussian mixture model of the pulse height data set can be expressed as:

[0044]

[0045] Among them, p i is the i-th pulse height value in the pulse height data set; K is the number of Gaussian distributions in the Gaussian mixture model. In this embodiment, the number of each Gaussian distribution is the same as the number of energy peaks of the sample to be tested, that is, K = N 预测 ; N(p i |μ k ,Σ k ) is the kth Gaussian distribution, the mean of which is μ k , the covariance matrix is ∑ k ; π k is the weight of the kth Gaussian distribution, satisfying

[0046] ② The expectation maximization algorithm is used to calculate the posterior probability of each pulse height value belonging to each Gaussian distribution step by step, and the weight, mean and covariance matrix of each Gaussian distribution are iteratively optimized until each Gaussian distribution reaches convergence, and the corresponding multiple converged Gaussian distributions are obtained.

[0047] It should be understood that the expectation-maximization algorithm (EM) is an iterative optimization algorithm used to estimate parameters in a probability model with latent variables. It mainly optimizes the parameters of the probability model (such as the Gaussian mixture model) step by step by alternating between the E step and the M step.

[0048] In this embodiment, the expectation maximization algorithm is used. Step E includes calculating the posterior probability that each pulse height value belongs to each Gaussian distribution. The specific calculation formula is:

[0049]

[0050] Use the expectation maximization algorithm M step: update the parameters of each Gaussian distribution, including the weight, mean and covariance matrix of each Gaussian distribution. The specific calculation formula is:

[0051]

[0052] Among them, p i is the ith pulse height value in the pulse height data set; C is the number of pulse height values in the pulse height data set; N(p i |μ k ,∑ k ) is the kth Gaussian distribution, the mean of which is μ k , the covariance matrix is ∑ k ; N(p i |μ j ,∑ j ) is the jth Gaussian distribution, the mean of which is μ j , the covariance matrix is ∑ j ; π k , π j are the weights of the kth and jth Gaussian distributions respectively; γ ik is the i-th pulse height value p i belongs to the kth Gaussian distribution.

[0053] Repeat steps E and M until the log-likelihood function converges to obtain a converged Gaussian mixture model. The converged Gaussian mixture model includes K corresponding converged Gaussian distributions.

[0054] ③ Calculate the posterior probability that each pulse height value belongs to each convergent Gaussian distribution, and assign each pulse height value to the convergent Gaussian distribution with the largest posterior probability, thereby dividing the pulse height data set into multiple pulse height data subsets.

[0055] Step S3: performing averaging processing on each pulse signal in each pulse signal data subset to obtain an average pulse template of each pulse signal data subset.

[0056] It should be noted that averaging the multiple pulse signals in each pulse signal data subset refers to aligning the pulse signals and then averaging them point by point, so as to retain the common characteristics of the pulse signals and weaken random noise.

[0057] Specifically, in one embodiment, the following formula is used to calculate the average pulse template of each pulse signal data subset.

[0058]

[0059] Among them, x temp,j is the average pulse template of the jth pulse signal data subset, x ij is the i-th pulse signal in the j-th pulse signal data subset, C jis the number of pulse signals in the jth pulse signal data subset. It should be noted that the number of pulse signals in each pulse signal data subset is the same as the number of pulse height values in the corresponding pulse height data subset.

[0060] Step S4: Calculate the target shrinkage covariance matrix of the noise signal data set, and calculate the optimal pulse height fitting factor with the minimum Mahalanobis distance between each pulse signal and its corresponding average pulse template based on the target shrinkage covariance matrix, so as to correct each pulse height value, obtain the corresponding corrected pulse height value, and construct a corrected pulse height data set.

[0061] It should be understood that when the data dimension is high or the sample size is insufficient, the sample covariance matrix estimation error is large, leading to overfitting. Therefore, this application adopts the Ledoit-Wolf shrinkage method to optimize the covariance estimation by balancing the sample covariance with the structured target matrix, thereby solving the problem of unstable sample covariance matrix estimation in high-dimensional data.

[0062] In one embodiment, the method of calculating the target shrinkage covariance matrix of the noise signal data set using the Ledoit-Wolf shrinkage method includes the following steps.

[0063] ① Construct a sample matrix of the noise signal data set and generate the corresponding sample covariance matrix.

[0064] It should be understood that sample covariance is a measure of the trend of joint changes of multiple random variables in statistics, reflecting the linear correlation between random variables. Based on the above embodiment, the noise signal data set includes multiple noise signals, each of which is an array of length n, whose elements are the signal amplitudes of multiple measurements, and the signal amplitudes of each measurement are used as random variables. Thus, the sample covariance matrix S of the noise signal data set generated is of dimension n×n, where the diagonal elements S jj is the variance of the jth random variable, and the off-diagonal elements S ij is the covariance between the ith random variable and the jth random variable.

[0065] ② Construct a target matrix for the noise signal data set, shrink the sample covariance matrix toward the target matrix, and calculate the optimal shrinkage coefficient of the sample covariance matrix by minimizing the mean square error between the shrunken sample covariance matrix and the true covariance matrix.

[0066] In one embodiment, the target matrix is a diagonal matrix of the same dimension as the sample covariance matrix, that is, an n×n diagonal matrix. When the Ledoit-Wolf method is used to shrink the sample covariance matrix, the calculation formula of the target matrix is:

[0067] T=Tr(S)I / d; Formula (9)

[0068] Wherein, T is the target matrix, S is the sample covariance matrix, Tr(S) is the trace of the sample covariance matrix, i is the identity matrix, and d is the dimension of the sample covariance matrix.

[0069] At this time, when the sample covariance matrix is shrunk to the target matrix, the calculation formula is satisfied:

[0070] Σ LW =(1-λ)S+λT; Formula (10)

[0071] Wherein, λ is the shrinkage coefficient, S is the sample covariance matrix, and T is the target matrix.

[0072] By minimizing the shrinkage covariance matrix Σ LW The Frobenius norm mean square error between the true covariance matrix Σ is optimized for the shrinkage coefficient λ, and the calculation formula is:

[0073]

[0074] Among them, λ opt is the optimal shrinkage coefficient, the numerator is the mean square error between the sample covariance matrix S and the true covariance matrix Σ, the denominator is the mean square error between the sample covariance matrix S and the target matrix T.

[0075] In actual operation, the optimal shrinkage coefficient λ is approximately calculated by statistics opt , the calculation formula is:

[0076] λ opt ≈(∑ i,j Var(S ij )) / ∑ i,j (S ij -T ij ) 2 ; Formula (12)

[0077] Among them, λ opt is the optimal shrinkage coefficient; S is the sample covariance matrix, S ij is the i-th row and j-th column element of the sample covariance matrix, Var(S ij ) is the variance of the elements in the i-th row and j-th column of the sample covariance matrix; T is the target matrix, T ij is the element in the i-th row and j-th column of the target matrix.

[0078] ③ According to the optimal shrinkage coefficient, calculate the target shrinkage covariance matrix of the noise signal data set.

[0079] Specifically, the calculation formula of the target shrinkage covariance matrix is:

[0080]

[0081] in, is the target shrinkage covariance matrix of the noise signal dataset; S is the sample covariance matrix of the noise signal dataset, Tr(S) is the trace of the sample covariance matrix, I is the identity matrix, d is the dimension of the sample covariance matrix; T is the target matrix of the noise signal dataset; λ opt is the optimal shrinkage coefficient.

[0082] In this embodiment, the present application adopts the Ledoit-Wolf shrinkage covariance estimation method, which can balance the sample covariance and the unit matrix in the high-dimensional small sample noise signal data set, avoid the covariance matrix morbidity problem, and thus help further improve the correction accuracy of the pulse height value.

[0083] After calculating and obtaining the target shrinkage covariance matrix of the noise signal data set, the optimal pulse height fitting factor with the minimum Mahalanobis distance between each pulse signal and its corresponding average pulse template can be calculated based on the target shrinkage covariance matrix. It should be understood that Mahalanobis distance is a distance measurement method that takes into account the covariance structure of the data. It is used to measure the degree of difference between each pulse signal and its corresponding average pulse template. It solves the limitations of Euclidean distance in data dimension correlation or scale difference, helps to improve the correction accuracy of pulse height, improves the signal-to-noise ratio of the pulse signal, and enhances the suppression of various noises.

[0084] In one embodiment, the relationship between each pulse signal in each pulse signal data set and the corresponding average pulse template can be expressed as:

[0085] x ij =α ij x temp,j +noise; formula (14)

[0086] Among them, x ij is the i-th pulse signal in the j-th pulse signal data subset, x temp,j is the average pulse template of the jth pulse signal data subset, α ij is the pulse height fitting factor of the i-th pulse signal in the j-th pulse signal data subset, and noise is the corresponding mixed noise.

[0087] Therefore, the calculation formula of the Mahalanobis distance from each pulse signal to the corresponding average pulse template is:

[0088]

[0089] Among them, M(α ij ) is the i-th pulse signal x in the j-th pulse signal data subset ij The average pulse template x of the jth pulse signal data subset temp,j The Mahalanobis distance, Target shrinkage covariance matrix for the noisy signal dataset.

[0090] make Based on the minimization Mahalanobis distance criterion, the closed-form solution of the minimization Mahalanobis distance from each pulse signal to the corresponding average pulse template is calculated to obtain the optimal pulse height fitting factor. The specific calculation formula is:

[0091]

[0092] Among them, α opt,ij is the optimal pulse height fitting factor of the i-th pulse signal in the j-th pulse signal data subset, x ij is the i-th pulse signal in the j-th pulse signal data subset, x temp,j is the average pulse template of the j-th pulse signal data subset, Target shrinkage covariance matrix for the noisy signal dataset.

[0093] In one embodiment, each pulse height value in each of the pulse height data subsets is corrected to obtain a corrected pulse height value corresponding to each pulse signal using the following calculation formula:

[0094] p ij =α opt,ijpj ; Formula (17)

[0095] Among them, p ij is the corrected pulse height value of the i-th pulse signal in the j-th pulse signal data subset, α opt,ij is the optimal pulse height fitting factor of the i-th pulse signal in the j-th pulse signal data subset, p j is the pulse height value of the average pulse template of the j-th pulse signal data subset, that is, the maximum signal amplitude corresponding to the average pulse template.

[0096] In one embodiment, the pulse signal data processing method further includes step S5 and step S6 (not shown).

[0097] Step S5: drawing a pulse height distribution histogram of the corrected pulse height data set to perform a visual analysis on the corrected pulse signal height data set.

[0098] Step S6: Based on the proportional linear relationship between the pulse height and the incident particle energy of the superconducting phase change edge detector, the energy spectrum data of the sample to be tested is obtained, and an energy spectrum diagram of the sample to be tested is drawn.

[0099] In order to better illustrate the specific steps and beneficial effects of the pulse signal data processing method, this application is further described in detail in conjunction with the following specific embodiments.

[0100] Example 1: TES-X-ray pulse signal data processing method.

[0101] In this embodiment, the TES-X-ray pulse signal data processing method includes the following steps.

[0102] ① Use an external X-ray source to irradiate the sample to be tested, stimulate the X-ray energy of the sample to be tested, and collect multiple X-ray measurement signals output by the superconducting phase change edge detector when detecting the sample to be tested, including 11590 pulse sequences with a length of 10400 and 11590 zero-input sequences with a length of 10400. The 11590 pulse sequences with a length of 10400 are constructed as a pulse signal dataset, and the 11590 zero-input sequences with a length of 10400 are constructed as a noise signal dataset. The original pulse height p of each pulse sequence is extracted. i , stored in an array P value In the example, the array length is 11590 to construct the pulse height dataset. i The distribution histogram of Figure 2 shown.

[0103] ②Use Gaussian mixture model to calculate the length of array P of 11590 value Specifically, according to the elemental composition of the sample to be tested, the number of energy peaks of the sample to be tested is predicted to be 2, thereby setting the number of Gaussian distributions of the Gaussian mixture model to 2; the expectation maximization algorithm is used to optimize the two Gaussian distributions to obtain two convergent Gaussian distributions, thereby converting the pulse height value p i Assign to the convergent Gaussian distribution with the largest posterior probability, such as Figure 3 As shown, the pulse height data set is divided into two pulse height data subsets: a first pulse height data subset and a second pulse height data subset.

[0104] ③ According to each pulse height data subset, the pulse signal data set is divided into two pulse signal data subsets: a first pulse signal data subset and a second pulse signal data subset.

[0105] ④ Calculate the average pulse template x of the two pulse signal data subsets respectively temp,1 and x temp,2 . Among them, the average pulse template x temp,1 and x temp,2 The calculation formulas are:

[0106]

[0107] Among them, x temp,1 is the average pulse template of the first pulse signal data subset, x temp,2 is the average pulse template of the second pulse signal data subset; x i1 is the i-th pulse signal in the first pulse signal data subset, x i2 is the i-th pulse signal in the second pulse signal data subset; C1 is the number of pulse signals in the first pulse signal data subset, and C2 is the number of pulse signals in the second pulse signal data subset.

[0108] ⑤ The Ledoit-Wolf shrinkage estimation method is used to calculate the target shrinkage covariance matrix of the noise signal data set. The calculation formula is:

[0109]

[0110] λ opt ≈(∑ i,j Var(S ij )) / ∑ i,j (S ij -T ij ) 2 ; Formula (21)

[0111] Where S is the sample covariance matrix of the noise signal dataset, S ij is the i-th row and j-th column element of the sample covariance matrix, Var(S ij ) is the variance of the element in the i-th row and j-th column of the sample covariance matrix, Tr(S) is the trace of the sample covariance matrix, I is the identity matrix, d is the dimension of the sample covariance matrix (10400×10400); T is the target matrix of the noise signal dataset, T ij is the element in the i-th row and j-th column of the target matrix; opt is the optimal shrinkage coefficient, Target shrinkage covariance matrix for the noisy signal dataset.

[0112] ⑥ Shrink the covariance matrix Σ according to the target LW , calculate each pulse signal x ij The corresponding average pulse template x temp,1 and x temp,2The optimal pulse height fitting factor α with the smallest Mahalanobis distance opt,i1 and α opt,i2 The specific calculation formulas are:

[0113]

[0114] Among them, α opt,i1 is the optimal pulse height fitting factor of the i-th pulse signal in the first pulse signal data subset, α opt,i2 is the optimal pulse height fitting factor of the i-th pulse signal in the second pulse signal data subset; x i1 is the i-th pulse signal in the first pulse signal data subset, x i2 is the i-th pulse signal in the second pulse signal data subset; x temp,1 is the average pulse template of the first pulse signal data subset, x temp,2 is an average pulse template of a second pulse signal data subset; Target shrinkage covariance matrix for the noisy signal dataset.

[0115] ⑦ According to the optimal pulse height fitting factors, the corrected pulse height value of each pulse signal is calculated. The calculation formula is:

[0116] p new,i1 =α opt,i1 p1; formula (24)

[0117] p new,i2 =α opt,i2 p2; formula (25)

[0118] Among them, p new,i1 is the corrected pulse height value of the i-th pulse signal in the first pulse signal data subset, p new,i2 is the corrected pulse height value of the i-th pulse signal in the second pulse signal data subset;

[0119] α opt,i1 is the optimal pulse height fitting factor of the i-th pulse signal in the first pulse signal data subset, α opt,i2 is the optimal pulse height fitting factor of the i-th pulse signal in the second pulse signal data subset;

[0120] p1 is the pulse height value of the average pulse template of the first pulse signal data subset, that is, x temp,1 The maximum signal amplitude in the middle, p2 is the pulse height value of the average pulse template of the second pulse signal data subset, that is, x temp,2 The maximum signal amplitude.

[0121] ⑧Draw the height value p of each correction pulse new The distribution histogram of Figure 4 shown.

[0122] contrast Figure 2 as well as Figure 4 The distribution histogram of the TES-X-ray pulse signal data processing method described in this application can effectively improve the analysis quality of the original pulse signal height, suppress various types of noise in the pulse signal, including random noise, high-frequency noise, low-frequency noise, etc., and can meet the needs of multi-energy radiation and dynamic radiation scenarios.

[0123] like Figure 5 FIG2 is a block diagram of a pulse signal data processing system 500 according to an embodiment of the present invention. The pulse signal data processing system 500 comprises a data acquisition module 501, a data classification module 502, a signal averaging processing module 503 and a pulse height correction module 504 connected in sequence.

[0124] Specifically, the data acquisition module 501 is used to collect multiple measurement signals output by the superconducting phase change edge detector when detecting the sample to be tested, and to filter the noise signals and pulse signals in each measurement signal, and extract the initial pulse height value of each pulse signal to construct a noise signal data set, a pulse signal data set and a pulse height data set.

[0125] The data classification module 502 is used to perform cluster analysis on the pulse height data set using a Gaussian mixture model, divide the pulse height data set into multiple pulse height data subsets, and divide the pulse signal data set into multiple pulse signal data subsets according to each pulse height data subset.

[0126] The signal averaging processing module 503 is used to perform averaging processing on each pulse signal in each pulse signal data subset to obtain an average pulse template for each pulse signal data subset.

[0127] The pulse height correction module 504 is used to calculate the target shrinkage covariance matrix of the noise signal data set, and based on the target shrinkage covariance matrix, calculate the optimal pulse height fitting factor with the minimum Mahalanobis distance between each pulse signal and its corresponding average pulse template, so as to correct each pulse height value, obtain the corresponding corrected pulse height value, and construct a corrected pulse height data set.

[0128] It should be understood that the execution process of each module to implement a specific function has been described in detail in the above method embodiment, and for the sake of brevity, it will not be repeated here.

[0129] It should also be understood that the division of modules in the embodiments of the present application is illustrative and is merely a logical functional division. In actual implementation, other division methods may be used. Furthermore, the functional modules in the various embodiments of the present application may be integrated into a single processor, or may exist physically separately, or two or more modules may be integrated into a single module. The aforementioned integrated modules may be implemented in the form of hardware or software functional modules.

[0130] The pulse signal data processing method provided in the embodiment of the present application can be implemented on the terminal side or the server side. As for the hardware structure of the pulse signal data processing terminal 600, please refer to Figure 6 , is an optional hardware structure diagram of the pulse signal data processing terminal 600 provided in an embodiment of the present application. The pulse signal data processing terminal 600 can be a mobile phone, a computer device, a tablet device, a personal digital processing device, a factory background processing device, etc. The pulse signal data processing terminal 600 includes: at least one processor 601, a memory 602, at least one network interface 604 and a user interface 606. In addition, the various components in the pulse signal data processing terminal 600 are coupled together through a bus system 605. It can be understood that the bus system 605 is used to realize the connection and communication between these components. In addition to the data bus, the bus system 605 also includes a power bus, a control bus and a status signal bus. However, for the sake of clarity, Figure 6 The various buses are all labeled as bus systems in FIG. The user interface 606 may include a display, keyboard, mouse, trackball, click gun, keys, buttons, touch pad or touch screen, etc.

[0131] It can be understood that the memory 602 can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. This application is not specifically limited. The memory 602 in the embodiment of the present application is used to store various categories of data to support the operation of the pulse signal data processing terminal 600. Examples of these data include: any executable program for operating on the pulse signal data processing terminal 600, such as an operating system 6021 and an application 6022; the operating system 6021 includes various system programs, such as a framework layer, a core library layer, a driver layer, etc., for implementing various basic services and processing hardware-based tasks. The application 6022 can include various applications, such as a media player (MediaPlayer), a browser (Browser), etc. The pulse signal data processing method provided by the method embodiment of the present application can be included in the application 6022.

[0132] The pulse signal data processing method disclosed in the above-mentioned method embodiment of the present application can be applied to the processor 601, or implemented by the processor 601. The processor 601 may be an integrated circuit chip with signal processing capabilities. During implementation, the various steps of the above-mentioned method can be completed by the hardware integrated logic circuit in the processor 601 or by instructions in the form of software. The above-mentioned processor 601 can be a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The processor 601 can implement or execute the various methods, steps and logic block diagrams disclosed in the embodiments of the present application. The general-purpose processor 601 can be a microprocessor or any conventional processor, etc.

[0133] In an exemplary embodiment, the pulse signal data processing terminal 600 may be implemented by one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), and complex programmable logic devices (CPLDs) to execute the aforementioned pulse signal data processing method.

[0134] Those skilled in the art will appreciate that all or part of the steps in the above-described method embodiments can be implemented using hardware associated with a computer program. The aforementioned computer program can be stored in a computer-readable storage medium. When executed, the program performs the steps in the above-described method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0135] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

[0136] In summary, the present application provides a pulse signal data processing method, system and terminal, which performs unsupervised clustering of the pulse height values of each pulse signal through a Gaussian mixture model, and adopts the Ledoit-Wolf shrinkage covariance estimation method to calculate the target shrinkage covariance matrix of the noise signal data set, so as to calculate the optimal pulse height fitting factor of each pulse signal based on the minimization of the Mahalanobis distance criterion, and correct the pulse height value of each pulse signal accordingly, thereby solving the technical problem that the existing pulse signal noise suppression technology is difficult to meet the various noise suppression requirements of multi-energy radiation and dynamic radiation scenarios, improving the correction accuracy of the pulse height value and the signal-to-noise ratio of the pulse signal, and when applied to the superconducting phase change edge detector to detect the sample to be tested, it can improve the superconducting phase change edge detector's ability to identify the pulse signal and the accuracy of the energy resolution of the sample to be tested, thereby ensuring the high energy resolution of the superconducting phase change edge detector. Therefore, the present application effectively overcomes the various shortcomings in the existing technology and has a high industrial utilization value.

[0137] The above embodiments are merely illustrative of the principles and effects of this application and are not intended to limit this application. Anyone skilled in the art may modify or alter the above embodiments without departing from the spirit and scope of this application. Therefore, all equivalent modifications or alterations made by one of ordinary skill in the art without departing from the spirit and technical concepts disclosed in this application shall be covered by the claims of this application.

Claims

1. A pulse signal data processing method, characterized in that: include: Collect multiple measurement signals output by a superconducting phase change edge detector when detecting a sample to be tested, filter out noise signals and pulse signals in each measurement signal, and extract the pulse height value of each pulse signal to construct a noise signal dataset, a pulse signal dataset, and a pulse height dataset; Performing cluster analysis on the pulse height dataset using a Gaussian mixture model to divide the pulse height dataset into a plurality of pulse height data subsets, and dividing the pulse signal dataset into a plurality of pulse signal data subsets according to each pulse height data subset; performing averaging processing on each pulse signal in each pulse signal data subset to obtain an average pulse template for each pulse signal data subset; Calculate the target shrinkage covariance matrix of the noise signal data set, and based on the target shrinkage covariance matrix, calculate the optimal pulse height fitting factor with the minimum Mahalanobis distance between each pulse signal and its corresponding average pulse template, so as to correct each pulse height value, obtain the corresponding corrected pulse height value, and construct a corrected pulse height data set.

2. The pulse signal data processing method according to claim 1, characterized in that: The method of performing cluster analysis on the pulse height data set by using a Gaussian mixture model to divide the pulse height data set into a plurality of pulse height data subsets includes: Predicting the number of energy peaks of the sample to be tested based on the elemental composition of the sample to be tested to construct a Gaussian mixture model of the pulse height data set; wherein the number of Gaussian distributions in the Gaussian mixture model is the same as the number of energy peaks of the sample to be tested; The expectation-maximization algorithm is used to calculate the posterior probability of each pulse height value belonging to each Gaussian distribution in steps, and the weight, mean, and covariance matrix of each Gaussian distribution are iteratively optimized until each Gaussian distribution reaches convergence, obtaining corresponding multiple converged Gaussian distributions; The posterior probability that each pulse height value belongs to each convergent Gaussian distribution is calculated, and each pulse height value is assigned to the convergent Gaussian distribution with the largest posterior probability, thereby dividing the pulse height data set into multiple pulse height data subsets.

3. The pulse signal data processing method according to claim 1, characterized in that: The calculation method for obtaining the average pulse template of each pulse signal data subset includes: Among them, x temp,j is the average pulse template of the jth pulse signal data subset, x ij is the i-th pulse signal in the j-th pulse signal data subset, C j is the number of pulse signals in the j-th pulse signal data subset.

4. The pulse signal data processing method according to claim 1, wherein: The method of calculating the target shrinkage covariance matrix of the noise signal data set includes: Constructing a sample matrix of the noise signal data set and generating a corresponding sample covariance matrix; Constructing a target matrix of the noise signal data set, shrinking the sample covariance matrix toward the target matrix, and calculating the optimal shrinkage coefficient of the sample covariance matrix by minimizing the mean square error between the shrunken covariance matrix and the true covariance matrix; Calculating a target shrinkage covariance matrix of the noise signal data set according to the optimal shrinkage coefficient; The calculation formula of the optimal shrinkage coefficient is: λ opt ≈(∑ i,j Var(S ij )) / ∑ i,j (S ij -T ij ) 2 ; The calculation formula of the target shrinkage covariance matrix is: Where S is the sample covariance matrix of the noise signal dataset, S ij is the i-th row and j-th column element of the sample covariance matrix, Var(S ij ) is the variance of the elements in the i-th row and j-th column of the sample covariance matrix, Tr(S) is the trace of the sample covariance matrix, I is the identity matrix, d is the dimension of the sample covariance matrix; T is the target matrix of the noise signal data set, T ij is the element in the i-th row and j-th column of the target matrix; opt is the optimal shrinkage coefficient, Target shrinkage covariance matrix for the noisy signal dataset.

5. The pulse signal data processing method according to claim 1, characterized in that: The method of calculating the optimal pulse height fitting factor with the minimum Mahalanobis distance between each pulse signal and its corresponding average pulse template according to the target shrinkage covariance matrix includes: Among them, α opt,ij is the optimal pulse height fitting factor of the i-th pulse signal in the j-th pulse signal data subset, x ij is the i-th pulse signal in the j-th pulse signal data subset, x temp,j is the average pulse template of the j-th pulse signal data subset, Target shrinkage covariance matrix for the noisy signal dataset.

6. The pulse signal data processing method according to claim 1, characterized in that: According to each optimal pulse height fitting factor, each pulse height value is corrected to obtain the corresponding corrected pulse height value in the following manner: p ij =α opt,ij p j ; Among them, p ij is the corrected pulse height value of the i-th pulse signal in the j-th pulse signal data subset, α opt,ij is the optimal pulse height fitting factor of the i-th pulse signal in the j-th pulse signal data subset, p j is the pulse height value of the average pulse template of the j-th pulse signal data subset.

7. The pulse signal data processing method according to claim 1, characterized in that: Methods for filtering noise signals and pulse signals from each measurement signal include: A signal amplitude threshold is set, and the signal segment in each measurement signal with a signal amplitude less than or equal to the signal amplitude threshold is treated as a noise signal, and the signal segment with a signal amplitude greater than the signal amplitude threshold is treated as a pulse signal, thereby splitting each measurement signal into a noise signal and a pulse signal, and filtering out the noise signal and pulse signal in each measurement signal.

8. The pulse signal data processing method according to claim 1, characterized in that: Also includes: Drawing a pulse height distribution histogram of the corrected pulse height data set to perform visual analysis on the corrected pulse signal height data set; Based on the proportional linear relationship between the pulse height and the incident particle energy of the superconducting phase change edge detector, the energy spectrum data of the sample to be tested is obtained and the energy spectrum diagram of the sample to be tested is drawn.

9. A pulse signal data processing system, characterized in that: include: The data acquisition module is used to collect multiple measurement signals output by the superconducting phase change edge detector when detecting the sample to be tested, filter the noise signal and pulse signal in each measurement signal, and extract the pulse height value of each pulse signal to construct a noise signal data set, a pulse signal data set, and a pulse height data set; a data classification module, connected to the data acquisition module, for performing cluster analysis on the pulse height data set using a Gaussian mixture model, dividing the pulse height data set into a plurality of pulse height data subsets, and dividing the pulse signal data set into a plurality of pulse signal data subsets corresponding to each pulse height data subset; a signal averaging processing module, connected to the data classification module, for performing averaging processing on each pulse signal in each pulse signal data subset to obtain an average pulse template for each pulse signal data subset; A pulse height correction module is connected to the signal averaging processing module and is used to calculate the target shrinkage covariance matrix of the noise signal data set, and based on the target shrinkage covariance matrix, calculate the optimal pulse height fitting factor with the minimum Mahalanobis distance between each pulse signal and its corresponding average pulse template, so as to correct each pulse height value, obtain the corresponding corrected pulse height value, and construct a corrected pulse height data set.

10. A pulse signal data processing terminal, characterized in that: include: processor and memory; The memory is used to store computer programs and data; The processor is configured to execute the computer program stored in the memory, so as to enable the pulse signal data processing terminal to execute the pulse signal data processing method according to any one of claims 1 to 8.

Images