Decoupling method based on high-field solid-state nuclear magnetic resonance T2 relaxation

Abstract

The invention discloses a decoupling method based on high-field solid-state nuclear magnetic resonance T2 relaxation, and the method comprises the steps: carrying out the preprocessing of a collected T2 relaxation signal, including data conversion, chemical shift sorting, normalization processing and signal screening; converting the time domain signal into a frequency domain model by adopting inverse Laplacian transformation, and extracting pole features; on the basis of frequency domain pole information, automatic estimation and nonlinear fitting of signal parameters are achieved through radial basis function fitting in combination with a numerical optimization algorithm; and storing the processed time domain signal data. The method innovatively solves the technical problem of multi-atomic-nucleus relaxation signal analysis under the high-field condition, has the advantages of high signal-to-noise ratio, good analysis precision, wide application range and the like compared with a traditional low-field nuclear magnetic technology, and is particularly suitable for research and analysis in the fields of energy materials, catalysts and the like. And the processing capability and the decoupling effect of the high-field solid-state nuclear magnetic resonance relaxation signal are remarkably improved through technologies of system integration signal conversion, parameter optimization, regularization processing and the like.

Description

Technical Field

[0001] The present invention relates to the fields of materials, new energy, and high-field solid-state nuclear magnetic resonance, and particularly to a decoupling method based on high-field solid-state nuclear magnetic resonance T2 relaxation. Background Art

[0002] Since the beginning of this century, the relaxation phenomenon in nuclear magnetic resonance has been emphasized in many fields. Low-field nuclear magnetic resonance equipment has been widely used due to its highly competitive price and convenience, and nuclear spin relaxation signals have thus been used to analyze materials, food, biology, chemical engineering, geology, and other research fields.

[0003] There are many different inversion algorithms for relaxation decoupling in low-field nuclear magnetic resonance, including but not limited to singular value decomposition method (Xiao Lizhi, 1998), non-negative least squares method (Xiao Lizhi, 1998), transform inversion algorithm (Wang Weimin, 2001), singular value decomposition method under regularization constraint (Wang Caizhi, 2003), joint iterative inversion method (Yao Xugang, 2003), and so on.

[0004] Due to the equipment limitations of low-field nuclear magnetic resonance and the physical properties of different atomic nuclei, the spin relaxation signals measured in a low-field environment mostly come from 1 H nuclei. The relaxation signals of other atomic nuclei have a low signal-to-noise ratio or are even difficult to measure under low-field conditions. However, atomic nuclei other than 1 H nuclei are widely used in many fields such as energy, catalysis, and materials, and have extremely high research value. Therefore, the relaxation research of high-field solid-state nuclear magnetic resonance is crucial.

[0005] The main challenges faced by the existing technology in the field of high-field solid-state nuclear magnetic resonance relaxation decoupling: The decoupling algorithms of low-field nuclear magnetic resonance cannot be directly applied to a high-field environment, and the existing methods are difficult to effectively handle the low signal-to-noise ratio and complex coupling problems of non- 1 H nuclear relaxation signals under high-field conditions, resulting in insufficient analysis accuracy and stability.

[0006] It should be noted that the information disclosed in the above background art section is only used for understanding the background of the present application, and thus may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0007] The main purpose of the present invention is to overcome the defects existing in the above background art, and to provide a decoupling method based on high-field solid-state nuclear magnetic resonance T2 relaxation.

[0008] To achieve the above purpose, the present invention adopts the following technical solutions:

[0009] A decoupling method based on high-field solid-state nuclear magnetic resonance T2 relaxation, comprising the following steps:

[0010] S1. Data preprocessing: extract the T2 relaxation signal of high-field solid-state NMR, convert the discretized data into a time domain sequence and sort it according to the chemical shift to form a two-dimensional matrix, normalize the signal values ​​in the matrix, screen out the signals that meet the signal intensity gradient requirements, and recombine them to form a data set to be processed;

[0011] S2. Inverse Laplace transform: performing inverse Laplace transform on the coupled relaxation signal in the data set to be processed, converting the time domain signal into a frequency domain signal and determining the poles in the frequency domain;

[0012] S3. Radial basis function fitting: Determine the central value range of the Gaussian radial basis function based on the frequency domain pole information, optimize the parameters of the radial basis function using a numerical optimization algorithm, and approximate the exponential decay component of the signal through radial basis function fitting to achieve nonlinear fitting and smooth decoupling of the signal;

[0013] S4. Data saving: Save the fitted data in the form of time domain signals for subsequent analysis.

[0014] Furthermore, step S1 specifically includes:

[0015] Convert the discretized relaxation signal data into a preset format and sort them according to a preset time domain sequence;

[0016] Constructing a two-dimensional matrix based on the time domain sequence, wherein the rows of the matrix correspond to the time domain sequence and the columns correspond to the chemical shift;

[0017] Normalizing the signal values ​​in the two-dimensional matrix;

[0018] Filter the target signal based on the signal intensity gradient and remove the low-intensity noise component;

[0019] The filtered signals were reordered by chemical shift and reorganized into a data set to be processed.

[0020] Furthermore, step S2 specifically includes:

[0021] The time domain model of the original coupled relaxation signal is converted into a frequency domain model, and the signal components with different time constants are represented by calculating the poles in the frequency domain;

[0022] Based on the Gaussian distribution characteristics of the nuclear spin vector, the response of the frequency domain signal component is analyzed to suppress the cross-interference between signals with different time constants;

[0023] The separation relationship between the amplitude and time constant of the signal components is determined according to the frequency domain poles, providing frequency domain characteristic parameters for subsequent radial basis function fitting.

[0024] Furthermore, step S3 specifically includes:

[0025] Define the central value range of the Gaussian radial basis function based on the frequency domain poles, and use the least squares method to optimize the central parameters of the basis function to minimize the residual;

[0026] Dynamically adjust the fitting range with the basis function width as the adjustable variable, where the basis function width is associated with the approximation degree of the exponential decay signal;

[0027] By combining Gaussian basis functions with different widths, perform smooth separation and nonlinear fitting on the exponential decay components in the coupled relaxation signal.

[0028] Furthermore, the radial basis function fitting in step S3 includes the following optimization mechanisms:

[0029] By dynamically adjusting the basis function width, perform nonlinear fitting on the exponential decay signals with similar time constants to suppress the cross-interference between different signal components;

[0030] Based on the optimization of the basis function width, achieve smooth separation of signal components and avoid relying on fine adjustment of the frequency domain poles;

[0031] Automatically estimate the amplitude, time constant, and width parameters of the basis function through an optimization algorithm to achieve automatic decoupling of the coupled relaxation signal.

[0032] Furthermore, step S3 also includes:

[0033] Perform regularization on the residual after radial basis function fitting. By selecting the L1 regularization or L2 regularization penalty term, constrain the model parameters to suppress overfitting;

[0034] Among them: L1 regularization sparsifies the model by penalizing the sum of the absolute values of the parameters to automatically screen important signal components; L2 regularization achieves smooth convergence of the fitting curve and retains signal integrity by penalizing the sum of the squares of the parameters.

[0035] Furthermore, the L2 regularization specifically includes:

[0036] Introduce a penalty term proportional to the sum of the squares of the model parameters into the loss function, compress the amplitude of large-value parameters to reduce the model degrees of freedom;

[0037] By balancing the weights of the fitting residual and the penalty term, optimize the smoothness and stability of signal decoupling;

[0038] Retain the amplitude correlation of all signal components and avoid directly removing parameters to enhance the physical interpretability of the decoupling results.

[0039] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements the decoupling method based on high-field solid-state nuclear magnetic resonance T2 relaxation described above.

[0040] A computer program product includes a computer program which, when executed by a processor, implements the decoupling method based on high-field solid-state nuclear magnetic resonance T2 relaxation as described above.

[0041] The present invention has the following beneficial effects:

[0042] The present invention proposes a decoupling method based on high-field solid-state nuclear magnetic resonance T2 relaxation. Aiming at the technical problem of decoupling high-field solid-state nuclear magnetic resonance T2 relaxation signals, by innovatively combining data preprocessing, inverse Laplace transform and radial basis function fitting algorithm, and further introducing regularization and normalization techniques, it effectively overcomes the limitation that low-field nuclear magnetic decoupling methods cannot be directly applied to high-field environments, and realizes high signal-to-noise ratio, high precision and high stability analysis of composite exponential decay signals. Compared with the traditional low-field nuclear magnetic decoupling technology which is only applicable to 1 H nuclei and has algorithm limitations, the present invention can effectively process the relaxation signals of multiple nuclei under high-field conditions, solves the problem of immature high-field solid-state nuclear magnetic resonance relaxation decoupling technology, and provides a reliable technical means for nuclear research in the fields of energy, catalysis, materials, etc. Through the systematic integration of data preprocessing, frequency domain conversion, non-linear fitting and optimization algorithms, the present invention significantly improves the automation degree and analysis accuracy of signal decoupling, and fills the technical gap in the processing of high-field solid-state nuclear magnetic resonance relaxation signals.

[0043] Other beneficial effects in the embodiments of the present invention will be further described below. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 is the overall flowchart of the decoupling method based on high-field solid-state nuclear magnetic resonance T2 relaxation according to the embodiment of the present invention.

[0045] Figure 2 is the fitting data obtained through the BRUKER data processing software topspin.

[0046] Figure 3 is the fitting data obtained according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0047] The following makes a detailed description of the embodiments of the present invention. It should be emphasized that the following description is merely exemplary and not intended to limit the scope of the present invention and its applications.

[0048] In addition, the terms "first" and "second" are for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more such features. In the description of the embodiments of the present invention, "a plurality of" means two or more unless otherwise specifically defined.

[0049] Due to the mismatch between the systems and software of low-field and high-field nuclear magnetic resonance, and the significant differences in the nature of the relaxation signals measured under low-field and high-field nuclear magnetic resonance conditions, it is impossible to directly use the existing low-field nuclear magnetic resonance relaxation decoupling technology to invert and analyze the high-field nuclear magnetic resonance relaxation signals. In view of the challenges and problems faced by the existing technology, the present invention provides a decoupling method based on high-field solid-state nuclear magnetic resonance T2 relaxation, which combines data preprocessing, inverse Laplace transform, radial basis function fitting algorithm, and regularization technique to achieve high signal-to-noise ratio, high precision, and high stability analysis of composite exponential decay signals.

[0050] Referring to Figure 1 , an embodiment of the present invention provides a decoupling method based on high-field solid-state nuclear magnetic resonance T2 relaxation, including the following steps:

[0051] Step S1. Data preprocessing: Extract the T2 relaxation signal of high-field solid-state nuclear magnetic resonance, convert the discretized data into a time-domain sequence and sort it according to the chemical shift to form a two-dimensional matrix, normalize the signal values in the matrix, screen out the signals that meet the signal intensity gradient requirements, and recombine them to form a dataset to be processed;

[0052] In some embodiments, step S1 specifically includes: converting the discretized relaxation signal data into a preset format and sorting it according to a preset time-domain sequence; constructing a two-dimensional matrix based on the time-domain sequence, where the rows of the matrix correspond to the time-domain sequence and the columns correspond to the chemical shift; normalizing the signal values in the two-dimensional matrix; screening target signals based on the signal intensity gradient and removing low-intensity noise components; re-sorting the screened signals according to the chemical shift and recombining them into a dataset to be processed.

[0053] Step S2. Inverse Laplace transform: Perform an inverse Laplace transform on the coupled relaxation signals in the dataset to be processed, convert the time-domain signals into frequency-domain signals, and determine the poles in the frequency domain;

[0054] In some embodiments, step S2 specifically includes: converting the time-domain model of the original coupled relaxation signal into a frequency-domain model, and characterizing signal components with different time constants by calculating poles in the frequency domain; based on the Gaussian distribution characteristic of the nuclear spin vector, analyzing the response of the frequency-domain signal components to suppress the cross-interference between signals with different time constants; determining the separation relationship between the amplitude and time constant of the signal components according to the frequency-domain poles, and providing frequency-domain characteristic parameters for subsequent radial basis function fitting.

[0055] Step S3. Radial basis function fitting: determining the central value range of the Gaussian radial basis function according to the frequency-domain pole information, optimizing the parameters of the radial basis function by using a numerical optimization algorithm (such as the least squares method), and approximating the exponential decay component of the signal through radial basis function fitting to achieve nonlinear fitting and smooth decoupling of the signal;

[0056] In some embodiments, step S3 specifically includes: defining the central value range of the Gaussian radial basis function based on the frequency-domain poles, and using the least squares method to optimize the central parameter of the basis function to minimize the residual; dynamically adjusting the fitting range with the basis function width as an adjustable variable, where the basis function width is associated with the approximation degree of the exponential decay signal; and performing smooth separation and nonlinear fitting on the exponential decay component in the coupled relaxation signal by combining Gaussian basis functions with different widths.

[0057] In some embodiments, the radial basis function fitting includes the following optimization mechanisms: nonlinearly fitting exponential decay signals with similar time constants by dynamically adjusting the basis function width to suppress the cross-interference between different signal components; achieving smooth separation of signal components based on the optimization of the basis function width to avoid relying on fine adjustments of the frequency-domain poles; and automatically estimating the amplitude, time constant, and width parameters of the basis function through an optimization algorithm to achieve automatic decoupling of the coupled relaxation signal.

[0058] In some embodiments, step S3 further includes: performing regularization and normalization processing on the residual after radial basis function fitting, and constraining the model parameters by selecting the L1 regularization or L2 regularization penalty term to suppress overfitting; where: L1 regularization sparsifies the model by penalizing the sum of the absolute values of the parameters to automatically screen important signal components; L2 regularization achieves smooth convergence of the fitting curve and retains signal integrity by penalizing the sum of the squares of the parameters.

[0059] In a further preferred embodiment, the L2 regularization specifically includes: introducing a penalty term proportional to the sum of the squares of the model parameters into the loss function, compressing the amplitude of large-value parameters to reduce the model degrees of freedom; optimizing the smoothness and stability of signal decoupling by balancing the weights of the fitting residual and the penalty term; and retaining the amplitude correlation of all signal components to avoid directly removing parameters to enhance the physical interpretability of the decoupling result.

[0060] Step S4. Data saving: Save the fitted data in the form of a time-domain signal for subsequent analysis.

[0061] The decoupling method based on high-field solid-state nuclear magnetic resonance T2 relaxation of the present invention effectively solves the technical problem of analyzing multi-nuclear relaxation signals under high-field conditions. In the method of the present invention, after specific data preprocessing of the collected T2 relaxation signals, through the inverse Laplace transform, the time-domain signals are converted into a frequency-domain model and the pole characteristics are extracted, and a radial basis function fitting algorithm is adopted, combined with a numerical optimization method to automatically estimate the signal parameters, realizing non-linear fitting and smooth decoupling; preferably, the model parameters are further optimized by the L2 regularization technique to further improve the decoupling stability and accuracy. Compared with the traditional low-field nuclear magnetic resonance technology, the present invention significantly improves the signal-to-noise ratio and analysis accuracy of non-hydrogen nuclear relaxation signals in a high-field environment, fills the technical gap in the field of high-field solid-state nuclear magnetic resonance relaxation decoupling, and provides a reliable analysis means for the research of energy materials, catalysts, etc. This method has significant technical advantages such as high automation degree and stable and reliable analysis results.

[0062] The specific embodiments of the present invention, their algorithm examples and experimental verifications are further described below.

[0063] A decoupling method based on high-field solid-state nuclear magnetic resonance T2 relaxation specifically includes the following processing procedures:

[0064] Extract the T2 relaxation signal measured by the CPMG pulse sequence, and the model of this signal is as follows:

[0065]

[0066] where y(t) is the initial data function with time t as the independent variable, T2 is the time constant corresponding to the relaxation primitive, and the relaxation primitive with f(T2) as the intensity value is integrated in the interval from to , and ε(t) is the noise that causes signal perturbation to the initial data function. Since the measurement and extraction of data are discretized, the above model shows the following transformation:

[0067] Since the measurement and extraction of data are discretized, the above model presents the following transformation:

[0068]

[0069] where t i is the discretized time node, different from the above continuous independent variable t, and the initial data function y(t ) is formed by the coupled superposition of a finite number (here m) of discretized relaxation primitives i .

[0070] Convert the obtained discretized data into txt format and perform time-domain sorting based on the time series in vclist with the benchmark coordinates. The time series is required to have the following relationship with the echo time (D20) in the CPMG test parameters:

[0071] t CPMG,n = 2 × D 20 × C vclist,n

[0072] where t CPMG,n is the nth time node corresponding in the CPMG sequence, and C vclist,n corresponds to the nth sequence constant in the vclist time series.

[0073] Stack the relaxation signals layer by layer according to the time domain sequence to form a two-dimensional matrix. It is required that the matrix stacks the time domain sequence by rows and differentiates the chemical shifts by columns. After sorting the data according to the above method, normalize the relaxation signal values in the obtained two-dimensional matrix, perform gradient screening according to the signal intensity according to the algorithm, and require that the signal screening gradient is not greater than 5% of the peak signal. Re-sort and combine the screened column signals in the order of chemical shift to form a new two-dimensional matrix, and save the matrix variable as the dataset to be processed.

[0074] Import the matrix variable into the decoupling algorithm for processing, and the algorithm will automatically analyze the coupled relaxation signals under different chemical shift variables.

[0075] The algorithm will perform an inverse Laplace transform (ILT) on the original coupled relaxation signals, and the time domain model of the relaxation signals will be transformed into a frequency domain model to facilitate the calculation of the poles in the frequency domain. The frequency domain model of the relaxation signals is as follows:

[0076]

[0077] where Y(s) is the frequency domain signal after the inverse Laplace transform, which is expressed as a combination of n frequency domain basis elements, and P i represents the component intensity of the relaxation signal with τ i as the time constant, and ν i is the frequency constant corresponding to τ i respectively.

[0078] A potential challenge in the inverse Laplace transform is the problem of signal reconstruction and decoupling. Especially when the signal contains multiple time constants, the inverse Laplace transform of complex signals may cause cross-interference between signals due to the coupling of different time constants, making it difficult to directly analyze the amplitude and time constant of each component from the experimental data. Since the phase dispersion state of the nuclear spin vector shows a certain degree of Gaussian distribution driven by thermodynamics, the inverse Laplace transform of the time domain signal will respond to the signal components in the frequency domain.

[0079] Based on the above frequency-domain response, approximate pole values are obtained to define the range of function center values of the radial basis function. Then, the least squares method is used to return the pole value with the minimum residual, and the function center of the Gaussian radial basis function is initialized. Next, with the width of the Gaussian function as a variable, the fitting range is adjusted. The specific model of the radial basis function fitting (RBF) is as follows:

[0080]

[0081] Among them, in the radial basis function φ(t) with the time domain t as the independent variable, σ represents the width of the radial basis function centered on the time constant c.

[0082] The form of RBF has a certain similarity to the exponential decay function. When the width of the basis function is appropriately selected, the Gaussian function can well fit the exponential decay signal. In particular, when the width of the basis function takes a small value (i.e., the Gaussian function is narrow), RBF will approximate the form of exponential decay.

[0083] Through RBF fitting, it is not necessary to explicitly calculate the Laplace transform of each component, but to approximate each exponential decay component of the signal through the combination of Gaussian basis functions. Specifically, RBF fitting optimizes the decoupling process in the following aspects:

[0084] Nonlinear fitting: RBF fitting can handle nonlinear signals. Especially when the time constants of different exponential decay components are relatively close, RBF can smooth the signal by adjusting the function width, avoiding the possible cross-interference in the inverse Laplace transform.

[0085] Smooth decoupling: RBF can smoothly separate the components of the signal by optimizing the width of the basis function. This process is similar to optimizing the decoupling by adjusting the pole positions in the Laplace transform, but RBF can find a smooth fitting curve in the parameter space, thus avoiding fine adjustments of the poles.

[0086] Automatic parameter estimation: When performing RBF fitting, the amplitude, time constant, and width of the basis function can be directly estimated by the least squares method (or other optimization methods), which provides an automated signal decoupling method. While a separate inverse Laplace transform usually requires precise pole extraction, which may not be easily achieved in some complex signals.

[0087] After determining the function center and function width, the residual between the fitting function and the coupled relaxation signal will be further converged with the help of regularization norms. Regularization refers to adding an additional penalty term to the loss function to prevent the model from overfitting, especially when fitting complex signals. Without regularization, the fitting process may over-focus on noise, resulting in poor generalization ability of the model and unstable decoupling results. Regularization improves the robustness and interpretability of the model by introducing constraints on the model parameters, restricting the degrees of freedom of the model.

[0088] There are two methods of regularization norms, namely L1 regularization (Lasso regression) and L2 regularization (Ridge regression). L1 regularization penalizes by adding the sum of the absolute values of the parameters to the loss function. This method tends to sparsify the model, that is, directly eliminate some relatively small parameter values, achieving automatic screening of important signals and ignoring unimportant signal components. L2 regularization constrains the fitting process by adding the sum of the squares of the parameters to the loss function. Most importantly, this method tends to smooth the solution, compressing some large-value parameters to smaller values as much as possible without directly eliminating them, which will significantly enhance the integrity and interpretability of the signal.

[0089] It is preferred to use the L2 regularization norm to optimize the signal decoupling process:

[0090]

[0091] Using λ as the constraint factor to normalize the relaxation primitive intensity value f j so as to reduce the difference between the data curve y(t i ) and the fitting curve and achieve higher-precision data fitting.

[0092] The optimal decoupling curve is obtained according to the above model.

[0093] Finally, the fitting data is saved in the form of a time-domain signal as a txt format for subsequent analysis.

[0094] Experimental test

[0095] Figure 2 is the fitting data obtained by the BRUKER data processing software topspin (LPSC is the abbreviation of the test sample, and Vclist represents the reference coordinates). It can be seen from Figure 2 that the standard deviation between the function fitted by the TopSpin software under BRUKER and the known data curve is 0.04233, and the matching degree between the two is relatively low, and the fitting function is distorted to a certain extent; Figure 3 is the fitting data obtained according to the present invention. It can be seen from Figure 3 that the χ between the fitting function obtained according to the present invention and the known data curve2 is 5E-5, and the extremely high fitting matching degree ensures the authenticity and reliability of data parsing.

[0096] An embodiment of the present invention also provides a storage medium for storing a computer program, which when executed, at least executes the method described above.

[0097] An embodiment of the present invention also provides a control device, including a processor and a storage medium for storing a computer program; wherein, the processor is used to execute the computer program to at least execute the method described above.

[0098] An embodiment of the present invention also provides a processor, and the processor executes a computer program to at least execute the method described above.

[0099] The storage medium can be implemented by any type of non-volatile storage device, or a combination thereof. Among them, the non-volatile memory can be a read-only memory (ROM, Read Only Memory), a programmable read-only memory (PROM, Programmable Read-Only Memory), an erasable programmable read-only memory (EPROM, Erasable Programmable Read-Only Memory), an electrically erasable programmable read-only memory (EEPROM, Electrically Erasable Programmable Read-Only Memory), a ferromagnetic random access memory (FRAM, Ferromagnetic Random Access Memory), a flash memory (Flash Memory), a magnetic surface memory, an optical disc, or a compact disc read-only memory (CD-ROM, Compact Disc Read-Only Memory); the magnetic surface memory can be a disk memory or a tape memory. The storage medium described in the embodiments of the present invention is intended to include, but is not limited to, these and any other suitable types of memories.

[0100] In several embodiments provided by the present invention, it should be understood that the disclosed system and method can be implemented in other ways. The device embodiments described above are only illustrative. For example, the division of the units is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units or components can be combined, or can be integrated into another system, or some features can be ignored, or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed with each other can be through some interfaces, and the indirect coupling or communication connection of devices or units can be electrical, mechanical, or other forms.

[0101] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units; some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0102] In addition, in each embodiment of the present invention, all the functional units may be integrated into one processing unit, or each unit may be separately used as a unit, or two or more units may be integrated into one unit; the above integrated units may be implemented in the form of hardware, or in the form of a combination of hardware and software functional units.

[0103] Those of ordinary skill in the art can understand that all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions. The foregoing program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps including the above method embodiments; and the foregoing storage medium includes: removable storage devices, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disks or optical disks and other various media that can store program codes.

[0104] Alternatively, if the above integrated units of the present invention are implemented in the form of software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the embodiments of the present invention essentially or the part that contributes to the prior art can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in the various embodiments of the present invention. And the foregoing storage medium includes: removable storage devices, ROM, RAM, magnetic disks or optical disks and other various media that can store program codes.

[0105] The methods disclosed in several method embodiments provided by the present invention can be arbitrarily combined without conflict to obtain new method embodiments.

[0106] The features disclosed in several product embodiments provided by the present invention can be arbitrarily combined without conflict to obtain new product embodiments.

[0107] The features disclosed in several method or device embodiments provided by the present invention can be arbitrarily combined without conflict to obtain new method embodiments or device embodiments.

[0108] The above content is a further detailed description of the present invention in combination with specific preferred embodiments. It cannot be determined that the specific implementation of the present invention is only limited to these descriptions. For those skilled in the technical field to which the present invention pertains, without departing from the concept of the present invention, several equivalent substitutions or obvious variations can be made, and as long as the performance or use is the same, they should all be regarded as belonging to the protection scope of the present invention.

Claims

1. A decoupling method based on high-field solid-state nuclear magnetic resonance T2 relaxation, characterized in that: The following steps are involved: S1. Data preprocessing: extract the T2 relaxation signal of high-field solid-state NMR, convert the discretized data into a time domain sequence and sort it according to the chemical shift to form a two-dimensional matrix, normalize the signal values ​​in the matrix, screen out the signals that meet the signal intensity gradient requirements, and recombine them to form a data set to be processed; S2. Inverse Laplace transform: performing inverse Laplace transform on the coupled relaxation signal in the data set to be processed, converting the time domain signal into a frequency domain signal and determining the poles in the frequency domain; S3. Radial basis function fitting: Determine the central value range of the Gaussian radial basis function based on the frequency domain pole information, optimize the parameters of the radial basis function using a numerical optimization algorithm, and approximate the exponential decay component of the signal through radial basis function fitting to achieve nonlinear fitting and smooth decoupling of the signal; S4. Data saving: Save the fitted data in the form of time domain signals for subsequent analysis.

2. The decoupling method according to claim 1, characterized in that: Step S1 specifically includes: Convert the discretized relaxation signal data into a preset format and sort them according to a preset time domain sequence; Constructing a two-dimensional matrix based on the time domain sequence, wherein the rows of the matrix correspond to the time domain sequence and the columns correspond to the chemical shift; Normalizing the signal values ​​in the two-dimensional matrix; Filter the target signal based on the signal intensity gradient and remove the low-intensity noise component; The filtered signals were reordered by chemical shift and reorganized into a data set to be processed.

3. The decoupling method according to claim 1 or 2, characterized in that: Step S2 specifically includes: The time domain model of the original coupled relaxation signal is converted into a frequency domain model, and the signal components with different time constants are represented by calculating the poles in the frequency domain; Based on the Gaussian distribution characteristics of the nuclear spin vector, the response of the frequency domain signal component is analyzed to suppress the cross-interference between signals with different time constants; The separation relationship between the amplitude and time constant of the signal components is determined according to the frequency domain poles, providing frequency domain characteristic parameters for subsequent radial basis function fitting.

4. The decoupling method according to claim 1 or 2, characterized in that: Step S3 specifically includes: Defining the central value range of the Gaussian radial basis function based on the frequency domain poles, and optimizing the central parameters of the basis function using the least squares method to minimize the residual; Dynamically adjusting the fitting range by taking the basis function width as an adjustable variable, wherein the basis function width is associated with the degree of morphological approximation of the exponential decay signal; By combining Gaussian basis functions of different widths, the exponential decay components in the coupled relaxation signal are smoothly separated and nonlinearly fitted.

5. The decoupling method according to claim 1 or 2, characterized in that: The radial basis function fitting in step S3 includes the following optimization mechanisms: By dynamically adjusting the basis function width, nonlinear fitting is performed on exponential decay signals with similar time constants to suppress cross-interference between different signal components. Achieve smooth separation of signal components based on optimization of basis function width, avoiding reliance on subtle adjustments of frequency domain poles; The amplitude, time constant and width parameters of the basis function are automatically estimated through the optimization algorithm to achieve automatic decoupling of the coupled relaxation signal.

6. The decoupling method according to any one of claims 1 to 5, characterized in that: Step S3 also includes: Regularize the residual after radial basis function fitting, and constrain the model parameters by selecting L1 regularization or L2 regularization penalty terms to suppress overfitting; Among them: L1 regularization sparses the model by penalizing the sum of the absolute values ​​of the parameters to automatically screen important signal components; L2 regularization achieves smooth convergence of the fitting curve and retains signal integrity by penalizing the sum of the squares of the parameters.

7. The decoupling method according to claim 6, characterized in that: The L2 regularization specifically includes: A penalty term proportional to the sum of squares of model parameters is introduced into the loss function to compress the amplitude of large-value parameters to reduce the model's degrees of freedom. By balancing the weights of the fitting residual and the penalty term, the smoothness and stability of the signal decoupling are optimized; The amplitude correlation of all signal components is retained to avoid direct parameter elimination to enhance the physical interpretability of the decoupling results.

8. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the decoupling method based on high-field solid-state nuclear magnetic resonance T2 relaxation as described in any one of claims 1 to 7 is implemented.

9. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the decoupling method based on high-field solid-state nuclear magnetic resonance T2 relaxation as described in any one of claims 1 to 7 is implemented.

Images