Gamma energy spectrum unfolding optimization method and device based on nonlinear iteration

Abstract

The invention discloses a nonlinear iteration-based gamma energy spectrum unfolding optimization method and device, and the method comprises the steps: obtaining known gamma energy spectrum data, and carrying out the preprocessing of the known gamma energy spectrum data, and obtaining the preprocessed known gamma energy spectrum data; nuclide feature extraction is carried out on the preprocessed known gamma energy spectrum data to obtain nuclide feature peak positions and corresponding intensity information; constructing an initial nuclide component model based on the nuclide characteristic peak position and the corresponding intensity information; optimizing the initial nuclide component model by adopting a nonlinear iterative algorithm to obtain an optimized nuclide component model; performing spectrum unfolding on unknown gamma energy spectrum data according to the optimized nuclide component model to obtain the type and activity information of nuclides, adaptively adjusting nuclide component model parameters by introducing a nonlinear iterative algorithm, and further performing spectrum unfolding on the unknown gamma energy spectrum data through the optimized nuclide component model to obtain the type and activity information of the nuclides. And the spectrum unfolding accuracy and efficiency of the gamma energy spectrum are obviously improved.

Description

Technical Field

[0001] This application relates to the field of spectrum optimization technology, and in particular to a method and apparatus for gamma spectrum optimization based on nonlinear iteration. Background Technology

[0002] Gamma-ray energy dispersive spectroscopy (GDS) is widely used in nuclear energy, environmental monitoring, medical imaging, materials science, and other fields, primarily for the identification and quantitative analysis of different nuclides in samples. Gamma-ray spectral data contains energy information about the nuclides in the sample; by analyzing this energy information, the type and corresponding activity information of the nuclides can be obtained.

[0003] Most existing gamma-ray spectral analysis methods employ traditional algorithms based on peak detection and energy calibration. However, these methods are often limited in accuracy and efficiency due to factors such as noise interference, background radiation, and energy drift. Summary of the Invention

[0004] The main objective of this application is to provide a nonlinear iterative method and apparatus for optimizing γ-ray spectral analysis, aiming to solve the technical problems of poor accuracy and efficiency in existing γ-ray spectral analysis methods.

[0005] To achieve the above objectives, this application proposes a nonlinear iterative method for optimizing the γ-ray spectrum, which includes: Acquire known gamma energy spectrum data and preprocess the known gamma energy spectrum data to obtain preprocessed known gamma energy spectrum data. The preprocessing includes at least wavelet threshold denoising, energy drift correction and Compton continuum subtraction. Nuclide feature extraction is performed on the preprocessed known γ-ray spectrum data to obtain the nuclide feature peak positions and corresponding intensity information. The nuclide feature extraction includes at least peak detection, full-energy peak determination, and feature vector construction. Based on the position of the characteristic peaks of the nuclides and the corresponding intensity information, an initial nuclide composition model is constructed; The initial nuclide composition model is optimized using a nonlinear iterative algorithm to obtain an optimized nuclide composition model. The nonlinear iterative algorithm includes one or more combinations of artificial firefly optimization algorithm, culture algorithm, or simulated annealing algorithm. Based on the optimized nuclide composition model, the unknown gamma-ray spectrum data were analyzed to obtain the types and activity information of the nuclides.

[0006] In one embodiment, the step of acquiring known gamma-ray spectrum data and preprocessing the known gamma-ray spectrum data to obtain preprocessed known gamma-ray spectrum data includes: Known gamma spectrum data were acquired using a high-purity germanium detector or a scintillation detector. Adaptive denoising is performed on the known γ energy spectrum data using the Marat multi-resolution analysis algorithm and the Stein unbiased risk threshold method to obtain the denoised known γ energy spectrum data. The known γ energy spectrum data after noise reduction is calibrated using a dynamic time warping algorithm to obtain calibrated known γ energy spectrum data, wherein the energy axis calibration is used to correct energy drift. Based on the calibrated known γ-ray spectrum data, the Compton edge fitting method is used to subtract the Compton continuous spectrum to obtain the preprocessed known γ-ray spectrum data.

[0007] In one embodiment, the step of adaptively denoising the known γ-ray spectrum data according to the Marat multi-resolution analysis algorithm and the Stein unbiased risk threshold method to obtain denoised known γ-ray spectrum data includes: Obtain the wavelet basis function and the number of decomposition levels, wherein the wavelet basis function is one or more combinations of Daubechies wavelet, Symlets wavelet or Coiflets wavelet, and the number of decomposition levels is 3 to 5. Based on the wavelet basis function and the number of decomposition levels, the known γ-energy spectrum data is decomposed into approximation coefficients and detail coefficients at different scales; The detail coefficients are thresholded using a soft thresholding function to obtain the processed detail coefficients, wherein the threshold is determined according to the Stein unbiased risk estimation principle; The processed detail coefficients and approximation coefficients are reconstructed to obtain the known γ-ray energy spectrum data after noise reduction.

[0008] In one embodiment, the step of using a dynamic time warping algorithm to calibrate the energy axis of the denoised known gamma spectrum data to obtain calibrated known gamma spectrum data includes: Obtain reference gamma spectrum data, wherein the reference gamma spectrum data is gamma spectrum data with known energy resolution and energy scale; Calculate the dynamic time-normalized distance between the denoised known gamma spectrum data and the reference gamma spectrum data; Based on the dynamic time warping distance, the known γ energy spectrum data after noise reduction is subjected to energy axis translation and / or scaling transformation to obtain calibrated known γ energy spectrum data, wherein the energy axis translation and / or scaling transformation is used to correct energy drift.

[0009] In one embodiment, the preprocessed known gamma spectrum data is obtained by subtracting the Compton continuum from the calibrated known gamma spectrum data using the Compton edge fitting method, including: The Compton edge energy position is determined based on the Compton scattering formula and the energy range of the calibrated known gamma spectrum data. A nonlinear fitting algorithm was used to fit the energy spectrum data near the Compton edge location to obtain the Compton edge fitting curve; The preprocessed known gamma spectrum data is obtained by subtracting the Compton continuum from the calibrated known gamma spectrum data based on the Compton edge fitting curve.

[0010] In one embodiment, the step of extracting nuclide features from the preprocessed known γ-ray spectrum data to obtain the nuclide feature peak positions and corresponding intensity information includes: Peak detection was performed on the preprocessed known γ-ray spectrum data using an improved golden section search method and Parzen window density estimation method to identify potential full-energy peak positions. Based on the preset full-energy peak energy range and the potential full-energy peak positions, the full-energy peaks in the preprocessed known γ-ray spectrum data are determined; Determine the Zernike moments of each of the full-energy peaks, and construct eigenvectors based on the Zernike moments; The position of the nuclide characteristic peak and the corresponding intensity information are determined based on the feature vector.

[0011] In one embodiment, optimizing the initial nuclide composition model using a nonlinear iterative algorithm to obtain an optimized nuclide composition model includes: An initial firefly population is set up, wherein each firefly in the initial firefly population represents an initial nuclide composition model; The fitness value of each firefly is calculated according to an objective function, wherein the objective function is used to evaluate the degree of matching between the nuclide composition model and known gamma spectral data; The firefly position and brightness are updated based on the fitness value to obtain the updated firefly population, where the firefly position is a nuclide composition model parameter and the firefly brightness is a fitness value. The process iteratively executes the steps of calculating the fitness value of each firefly according to the objective function and updating the position and brightness of the fireflies according to the fitness value to obtain the updated firefly population, until a preset iteration termination condition is met to obtain the optimal firefly position. The iteration termination condition includes reaching the maximum number of iterations or the fitness value converging. Based on the determined optimized nuclide composition model parameters and the parameter update of the initial nuclide composition model, the optimized nuclide composition model is obtained.

[0012] In one embodiment, the step of optimizing the initial nuclide composition model using a nonlinear iterative algorithm to obtain an optimized nuclide composition model further includes: An initial cultural population is set up, wherein each individual in the initial cultural population represents an initial nuclide composition model and a corresponding belief space; The fitness value of each individual is evaluated based on the knowledge base and rule base in the belief space. The knowledge base contains prior knowledge of the nuclide composition model, and the rule base is used to guide the optimization process of the nuclide composition model. The individuals in the cultural population are updated according to the fitness values ​​to obtain the updated cultural population. The update process includes updating the belief space and evolving the population space. The steps of evaluating the fitness value of each individual based on the knowledge base and rule base in the belief space and updating the individuals in the cultural population based on the fitness value are executed iteratively until a preset iteration termination condition is met to obtain the optimal individual. The iteration termination condition includes reaching the maximum number of iterations or the fitness value converges. Based on the belief space and population space information of the optimal individual, the optimized nuclide composition model parameters are determined, and the parameters of the initial nuclide composition model are updated to obtain the optimized nuclide composition model.

[0013] In one embodiment, the step of optimizing the initial nuclide composition model using a nonlinear iterative algorithm to obtain an optimized nuclide composition model further includes: An initial temperature is set, and an initial solution is generated based on the initial temperature, wherein the initial solution represents an initial nuclide composition model; The fitness value of the initial solution is calculated according to the objective function, wherein the objective function is used to evaluate the degree of matching between the nuclide composition model and the known γ-ray spectral data; The temperature is reduced according to a preset cooling strategy, and a new solution is accepted according to the Metropolis criterion to obtain an updated solution. The new solution is obtained by randomly perturbing the current solution. The Metropolis criterion is used to accept solutions that worsen the objective function value with a preset probability. The steps of calculating the fitness value of the initial solution according to the objective function, reducing the temperature according to the preset cooling strategy, and accepting new solutions according to the Metropolis criterion are executed iteratively until the preset iteration termination condition is met to obtain the optimal solution. The iteration termination condition includes reaching the minimum temperature or the fitness value converges. Based on the optimal solution, the optimized nuclide composition model parameters are determined, and the parameters of the initial nuclide composition model are updated to obtain the optimized nuclide composition model.

[0014] Furthermore, to achieve the above objectives, this application also proposes a nonlinear iterative γ-ray spectrum optimization device, which includes: The preprocessing module is used to acquire known γ energy spectrum data and preprocess the known γ energy spectrum data to obtain preprocessed known γ energy spectrum data. The preprocessing includes at least wavelet threshold denoising, energy drift correction and Compton continuity subtraction. The extraction module is used to extract nuclide features from the preprocessed known γ-ray spectrum data to obtain the nuclide feature peak positions and corresponding intensity information. The nuclide feature extraction includes at least peak detection, full-energy peak determination, and feature vector construction. The construction module is used to construct an initial nuclide composition model based on the position of the characteristic peaks of the nuclide and the corresponding intensity information; The optimization module is used to optimize the initial nuclide composition model using a nonlinear iterative algorithm to obtain an optimized nuclide composition model. The nonlinear iterative algorithm includes one or more combinations of artificial firefly optimization algorithm, culture algorithm, or simulated annealing algorithm. The spectrum interpretation module is used to interpret unknown gamma-ray spectrum data based on the optimized nuclide composition model to obtain information on the type and activity of nuclides.

[0015] This application proposes one or more technical solutions, including: acquiring known gamma-ray spectrum data; preprocessing the known gamma-ray spectrum data to obtain preprocessed known gamma-ray spectrum data, wherein the preprocessing includes at least wavelet thresholding denoising, energy drift correction, and Compton continuum subtraction; extracting nuclide features from the preprocessed known gamma-ray spectrum data to obtain the nuclide feature peak positions and corresponding intensity information, wherein the nuclide feature extraction includes at least peak detection, full-energy peak determination, and feature vector construction; constructing an initial nuclide composition model based on the nuclide feature peak positions and corresponding intensity information; optimizing the initial nuclide composition model using a nonlinear iterative algorithm to obtain an optimized nuclide composition model, wherein the nonlinear iterative algorithm includes one or more combinations of artificial firefly optimization algorithm, culture algorithm, or simulated annealing algorithm; and interpreting the unknown gamma-ray spectrum data according to the optimized nuclide composition model to obtain nuclide type and activity information. By introducing a nonlinear iterative algorithm to adaptively adjust the parameters of the nuclide composition model, and then using the optimized nuclide composition model to interpret the unknown γ-ray spectrum data, the accuracy and efficiency of γ-ray spectrum interpretation are significantly improved. Attached Figure Description

[0016] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating an embodiment of the nonlinear iterative gamma-ray spectrum optimization method of this application. Figure 2 This is a flowchart illustrating Embodiment 2 of the γ-ray energy spectrum decomposition optimization method based on nonlinear iteration in this application; Figure 3 This is a schematic diagram of the module structure of the γ-ray spectrum decomposition and optimization device based on nonlinear iteration in an embodiment of this application.

[0019] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0020] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0021] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0022] The main solution of this application embodiment is as follows: First, acquire known gamma-ray spectrum data and preprocess the known gamma-ray spectrum data to obtain preprocessed known gamma-ray spectrum data. The preprocessing includes at least wavelet thresholding denoising, energy drift correction, and Compton continuum subtraction. Second, extract nuclide features from the preprocessed known gamma-ray spectrum data to obtain the nuclide feature peak positions and corresponding intensity information. The nuclide feature extraction includes at least peak detection, full-energy peak determination, and feature vector construction. Third, construct an initial nuclide composition model based on the nuclide feature peak positions and corresponding intensity information. Fourth, optimize the initial nuclide composition model using a nonlinear iterative algorithm to obtain an optimized nuclide composition model. The nonlinear iterative algorithm includes one or more combinations of artificial firefly optimization algorithms, culture algorithms, or simulated annealing algorithms. Fifth, interpret the unknown gamma-ray spectrum data according to the optimized nuclide composition model to obtain nuclide type and activity information.

[0023] Most existing gamma-ray spectral analysis methods employ traditional algorithms based on peak detection and energy calibration. However, these methods are often limited in accuracy and efficiency due to factors such as noise interference, background radiation, and energy drift.

[0024] This application provides a solution that adaptively adjusts the parameters of the nuclide composition model by introducing a nonlinear iterative algorithm, and then uses the optimized nuclide composition model to interpret unknown γ-ray spectral data, which significantly improves the accuracy and efficiency of γ-ray spectral interpretation.

[0025] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or an electronic device capable of performing the above functions, such as a nonlinear iterative gamma-ray spectrum decomposition and optimization device. The following description uses a nonlinear iterative gamma-ray spectrum decomposition and optimization device as an example to illustrate this embodiment and the subsequent embodiments.

[0026] Based on this, embodiments of this application provide a nonlinear iterative method for optimizing γ-energy spectrum resolution, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the γ-ray spectrum decomposition optimization method based on nonlinear iteration in this application.

[0027] In this embodiment, the γ-ray energy spectrum decomposition optimization method based on nonlinear iteration includes steps S10~S50: Step S10: Obtain known γ energy spectrum data and preprocess the known γ energy spectrum data to obtain preprocessed known γ energy spectrum data. The preprocessing includes at least wavelet threshold denoising, energy drift correction, and Compton continuum subtraction.

[0028] It should be noted that known gamma-ray energy spectrum data, obtained through experimental measurements or simulations, contains gamma-ray energy distribution data containing nuclide characteristic information. This data forms the basis for gamma-ray energy spectrum interpretation, and its quality directly affects the accuracy and reliability of the interpretation results. Therefore, before performing spectral interpretation, the known gamma-ray energy spectrum data needs to be preprocessed to improve its quality and reliability.

[0029] Understandably, in this embodiment, preprocessing includes at least wavelet thresholding denoising, energy drift correction, and Compton continuum subtraction. Wavelet thresholding denoising removes noise interference from known gamma-ray spectrum data, improving the signal-to-noise ratio. Energy drift correction eliminates the impact of energy drift on the spectrum data, ensuring accuracy. Compton continuum subtraction removes the continuous background spectrum generated by Compton scattering, highlighting nuclide characteristic peaks and providing a clearer data foundation for subsequent nuclide feature extraction. The quality and reliability of the preprocessed known gamma-ray spectrum data are significantly improved.

[0030] Step S20: Extract nuclide features from the preprocessed known γ-ray spectrum data to obtain the nuclide feature peak positions and corresponding intensity information. The nuclide feature extraction includes at least peak detection, full-energy peak determination, and feature vector construction.

[0031] It should be noted that, in this embodiment, nuclide feature extraction includes at least peak detection, full-energy peak determination, and feature vector construction. Peak detection is used to identify the positions of potential full-energy peaks in the preprocessed known γ-ray spectrum data, which is the basis for extracting nuclide features. Full-energy peak determination further identifies the full-energy peaks in the known γ-ray spectrum data based on the preset full-energy peak energy range and the positions of potential full-energy peaks. These full-energy peaks correspond to different nuclide types. Feature vector construction is based on feature parameters such as the Zernike moment of the full-energy peaks to construct feature vectors that can characterize the nuclide features. Through this series of steps, nuclide features in the known γ-ray spectrum data can be accurately extracted.

[0032] It is understandable that the position of a nuclide characteristic peak refers to the specific location of the full-energy peak in the energy spectrum data, while the corresponding intensity information reflects the intensity or count rate of that full-energy peak. The accuracy of the nuclide characteristic peak position and intensity information directly affects the accuracy and reliability of the final spectral interpretation results. Therefore, when extracting nuclide features, precise and efficient algorithms are required to ensure the accuracy and completeness of the extraction results.

[0033] In one feasible implementation, step S20 may include: performing peak detection on the preprocessed known γ-ray spectrum data using an improved golden section search method and Parzen window density estimation method to identify potential full-energy peak positions; determining the full-energy peaks in the preprocessed known γ-ray spectrum data based on a preset full-energy peak energy range and the potential full-energy peak positions; determining the Zernike moments of each full-energy peak and constructing eigenvectors based on the Zernike moments; and determining the nuclide characteristic peak positions and corresponding intensity information based on the eigenvectors.

[0034] It should be noted that peak detection using the improved golden section search method and Parzen window density estimation method can more accurately identify the positions of potential full-energy peaks in preprocessed known gamma-ray spectrum data. The golden section search method is a search algorithm for finding a specific element in an ordered array, with a fast convergence speed, suitable for peak detection in this implementation. It includes: setting an initial value for the search interval, i.e., the energy range to be analyzed; dividing the interval into two parts using the golden ratio (approximately 0.618); calculating the function values ​​(i.e., the intensity of the spectrum data) of two points in the interval; determining the search direction based on the calculation results, and gradually narrowing the search interval until the preset accuracy is reached. The Parzen window density estimation method is a non-parametric estimation method that can estimate the probability density function of known gamma-ray spectrum data, thereby helping to identify potential peak positions. In gamma-ray spectrum data, the density of each data point can be estimated using the Parzen window, and the peak position of the density can be used to determine potential full-energy peaks. Combining these two methods can improve the accuracy and efficiency of peak detection.

[0035] When determining the full-energy peak, it is necessary to screen and confirm based on the preset full-energy peak energy range and the potential full-energy peak positions. The preset full-energy peak energy range is pre-set based on factors such as the physical properties of the nuclide and experimental conditions, and is used to limit the possible positions of the full-energy peak and reduce the possibility of misjudgment. By comparing the potential full-energy peak positions with the preset full-energy peak energy range, the full-energy peak in the known gamma-ray spectrum data can be determined.

[0036] After determining the full-energy peaks, it is necessary to further extract the feature parameters of each full-energy peak to construct a feature vector. In this embodiment, Zernike moments are used as feature parameters. Zernike moments are image feature descriptors with rotation invariance and scale invariance, which can accurately describe the shape and texture features of the full-energy peaks. By calculating the Zernike moments of each full-energy peak, a feature vector that can characterize the nuclide features can be constructed. The formula for calculating Zernike moments is:

[0037] in, The original γ-ray spectrum data is converted into a function in polar coordinates. This represents the radial distance in polar coordinates. Angle in polar coordinates Let n and m be Zernike polynomials, where n and m are the order and angle number of the Zernike polynomial, respectively. The complex terms representing rotation help capture rotation-invariant characteristics.

[0038] Step S30: Based on the position of the characteristic peaks of the nuclide and the corresponding intensity information, construct an initial nuclide composition model.

[0039] It should be noted that the initial nuclide composition model is a mathematical model describing the types of nuclides and their relative abundance or activity, and it forms the basis for subsequent nonlinear iterative optimization. When constructing the initial nuclide composition model, it is necessary to fully consider the positions of the characteristic peaks of nuclides in the known gamma-ray spectrum data and their corresponding intensity information to ensure the accuracy and reliability of the model. Specifically, the types of nuclides can be determined by comparing the positions of the characteristic peaks with a pre-defined nuclide database; simultaneously, based on the intensity information of the characteristic peaks, the relative abundance or activity of each nuclide can be preliminarily estimated. Based on this, an initial nuclide composition model containing information such as nuclide types, relative abundance, or activity is constructed, providing an initial solution for subsequent nonlinear iterative optimization.

[0040] Step S40: The initial nuclide composition model is optimized using a nonlinear iterative algorithm to obtain an optimized nuclide composition model. The nonlinear iterative algorithm includes one or more combinations of artificial firefly optimization algorithm, culture algorithm, or simulated annealing algorithm.

[0041] It should be noted that nonlinear iterative algorithms are optimization algorithms that approximate the optimal solution by continuously updating the solution. They are suitable for handling complex optimization problems with a high degree of nonlinearity. In this embodiment, a nonlinear iterative algorithm is used to optimize the initial nuclide composition model. The aim is to continuously adjust the model parameters so that the model can better fit the nuclide characteristics in the known γ-ray energy spectrum data, thereby improving the accuracy and efficiency of spectrum interpretation.

[0042] Specifically, the artificial firefly optimization algorithm simulates the behavior of fireflies in nature, attracting mates by emitting light. It treats each firefly as a potential solution and searches for the optimal solution by continuously adjusting the positions of the fireflies (i.e., solutions). The culture algorithm combines a population-based evolutionary strategy with a knowledge-based belief space, solving optimization problems by simulating the cultural evolution process of human societies. The simulated annealing algorithm borrows from the physical process of metal annealing, accepting inferior solutions with a certain probability to escape local optima and potentially find the global optimum.

[0043] In this embodiment, one or more nonlinear iterative algorithms can be selected and combined according to actual needs and problem characteristics to fully utilize their respective advantages and improve optimization results. Through optimization by nonlinear iterative algorithms, an optimized nuclide composition model can be obtained. This model more accurately describes the types of nuclides and their relative abundance or activity, providing a more reliable basis for subsequent spectral interpretation.

[0044] In one feasible implementation, step S40 may include: setting an initial firefly population, wherein each firefly in the initial firefly population represents an initial nuclide composition model; calculating the fitness value of each firefly according to an objective function, wherein the objective function is used to evaluate the degree of matching between the nuclide composition model and known gamma-ray spectrum data; updating the position and brightness of the fireflies according to the fitness value to obtain an updated firefly population, wherein the position of the fireflies is a parameter of the nuclide composition model, and the brightness of the fireflies is a fitness value; iteratively executing the steps of calculating the fitness value of each firefly according to the objective function and updating the position and brightness of the fireflies according to the fitness value to obtain an updated firefly population, until a preset iteration termination condition is met to obtain the optimal firefly position, wherein the iteration termination condition includes reaching the maximum number of iterations or fitness value convergence; and updating the parameters of the optimized nuclide composition model according to the determined parameters and the initial nuclide composition model to obtain an optimized nuclide composition model.

[0045] It should be noted that, in this embodiment, the artificial firefly optimization algorithm is used to optimize the initial nuclide composition model.

[0046] Understandably, an initial firefly population is used to simulate the possible solution space, with each firefly representing a unique initial nuclide composition model, its position corresponding to different combinations of model parameters. To evaluate the performance of these models, an objective function is defined that quantifies the degree of fit between the nuclide composition model and known gamma-ray spectral data. The value of the objective function, i.e., the fitness value, reflects the model's fit to the data; a higher fitness value indicates a better fit between the model and the data.

[0047] During the iteration process, the fireflies' positions and brightness are continuously updated based on their fitness values. The positional adjustments simulate their search for better solutions in the search space, while changes in brightness reflect improvements in their fitness values. Through continuous iteration, the firefly population gradually converges to the vicinity of the optimal solution, i.e., the positions of the fireflies with the highest fitness values.

[0048] The iterative process continues until a preset termination condition is met. These conditions may include reaching the maximum number of iterations or the fitness value stabilizing, indicating that further iterations may not bring significant improvement. When the iteration terminates, the optimal firefly position is selected as the parameters for the optimized nuclide composition model, and the initial nuclide composition model is updated to obtain the final optimized nuclide composition model. This optimization process not only improves the accuracy of the nuclide composition model but also enhances its ability to interpret unknown gamma-ray spectral data. By introducing a nonlinear iterative algorithm, we can adaptively adjust the model parameters to better fit complex gamma-ray spectral data, thereby improving the accuracy and efficiency of spectral interpretation.

[0049] In one feasible implementation, step S40 may further include: setting an initial cultural population, wherein each individual in the initial cultural population represents an initial nuclide component model and a corresponding belief space; evaluating the fitness value of each individual based on a knowledge base and a rule base in the belief space, wherein the knowledge base contains prior knowledge of the nuclide component model, and the rule base is used to guide the optimization process of the nuclide component model; updating the individuals in the cultural population based on the fitness value to obtain an updated cultural population, wherein the update process includes updating the belief space and evolving the population space; iteratively executing the steps of evaluating the fitness value of each individual based on the knowledge base and the rule base in the belief space and updating the individuals in the cultural population based on the fitness value until a preset iteration termination condition is met to obtain an optimal individual, wherein the iteration termination condition includes reaching the maximum number of iterations or fitness value convergence; determining the optimized nuclide component model parameters based on the belief space and population space information of the optimal individual, and updating the parameters of the initial nuclide component model to obtain an optimized nuclide component model.

[0050] It should be noted that in this embodiment, a cultural algorithm is used to optimize the initial nuclide composition model. The cultural algorithm divides the optimization process into two spaces: a population space and a belief space. The population space contains all possible solutions, i.e., the initial nuclide composition model, while the belief space stores prior knowledge and optimization rules about these solutions. Through the co-evolution of these two spaces, the cultural algorithm can find high-quality solutions to complex optimization problems.

[0051] Specifically, an initial cultural population is established, where each individual represents a unique initial nuclide composition model and its corresponding belief space. The belief space includes a knowledge base and a rule base. The knowledge base stores prior knowledge about the nuclide composition model, such as the physical properties and energy distribution of nuclides, while the rule base provides rules and methods to guide the optimization process of the nuclide composition model.

[0052] During the iteration process, the fitness value of each individual is evaluated based on the knowledge base and rule base in the belief space. The fitness value reflects the degree of matching between the nuclide composition model and the known gamma-ray spectrum data, and is an important indicator for evaluating the model's performance. Then, the individuals in the cultural population are updated based on the fitness values, including updating the belief space and evolving the population space. Updating the belief space involves adjusting the knowledge base and rule base to better guide subsequent optimization processes, while evolving the population space generates new individuals through operations such as selection, crossover, and mutation to increase population diversity.

[0053] The iterative process continues until a preset termination condition is met, such as reaching the maximum number of iterations or the fitness value converging. When the iteration terminates, the optimal individual is selected, i.e., the individual with the highest fitness value. Its belief space and population space information contain the optimized nuclide composition model parameters. These parameters are then applied to the initial nuclide composition model for parameter updates, resulting in the final optimized nuclide composition model.

[0054] By introducing a culture-based algorithm, prior knowledge and optimization rules can be fully utilized to guide the optimization process of nuclide composition models, thereby improving the accuracy and efficiency of spectral interpretation. Simultaneously, the co-evolutionary mechanism of the population space and belief space in the culture-based algorithm enhances its adaptability to complex gamma-ray spectral data.

[0055] In one feasible implementation, step S40 may further include: setting an initial temperature and generating an initial solution based on the initial temperature, wherein the initial solution represents an initial nuclide composition model; calculating the fitness value of the initial solution according to an objective function, wherein the objective function is used to evaluate the degree of matching between the nuclide composition model and known gamma-ray spectrum data; reducing the temperature according to a preset cooling strategy and accepting a new solution according to the Metropolis criterion to obtain an updated solution, wherein the new solution is obtained by randomly perturbing the current solution, and the Metropolis criterion is used to accept solutions that worsen the objective function value with a preset probability; iteratively executing the steps of calculating the fitness value of the initial solution according to the objective function, reducing the temperature according to the preset cooling strategy, and accepting a new solution according to the Metropolis criterion until a preset iteration termination condition is met to obtain an optimal solution, wherein the iteration termination condition includes reaching the minimum temperature or the fitness value converging; determining the optimized nuclide composition model parameters according to the optimal solution, and updating the parameters of the initial nuclide composition model to obtain an optimized nuclide composition model.

[0056] It should be noted that, in this embodiment, a simulated annealing algorithm is used to optimize the initial nuclide composition model. The simulated annealing algorithm draws on the physical process of metal annealing, solving complex optimization problems by simulating the optimization of the microstructure of a metal during slow cooling after heating. In the simulated annealing algorithm, the initial nuclide composition model is considered as the initial solution, an initial temperature is set, and a series of initial solutions are generated based on this temperature. Then, the fitness values ​​of these initial solutions are calculated according to the objective function, which quantifies the degree of matching between the nuclide composition model and the known gamma-ray spectrum data; a higher fitness value indicates a better match between the model and the data.

[0057] During the iteration process, the temperature is gradually reduced according to a preset cooling strategy to simulate the cooling process of metal. In each iteration, the current solution is randomly perturbed to generate a new solution, and the Metropolis criterion is used to decide whether to accept the new solution. The Metropolis criterion allows for the acceptance of new solutions that worsen the objective function value with a certain probability, which helps the algorithm escape local optima and potentially find the global optimum. Through continuous iteration, the quality of the solution gradually improves until the preset iteration termination condition is met, such as reaching the minimum temperature or the fitness value converging.

[0058] When the iteration terminates, the optimal solution is selected as the parameters for the optimized nuclide composition model. These parameters describe key information such as the type, relative abundance, or activity of the nuclide. These parameters are then applied to the initial nuclide composition model for parameter updates, resulting in the final optimized nuclide composition model. By introducing a simulated annealing algorithm, high-quality solutions can be found in complex gamma-ray spectral data, improving the accuracy and efficiency of spectral analysis. This optimization process not only relies on the data fitting effect but also fully utilizes the global search capability of the simulated annealing algorithm, enhancing its ability to analyze unknown gamma-ray spectral data.

[0059] Step S50: Decode the unknown γ-ray spectral data according to the optimized nuclide composition model to obtain the nuclide type and activity information.

[0060] It should be noted that the optimized nuclide composition model was used as the basis for spectral analysis. This model has been thoroughly optimized using a nonlinear iterative algorithm, enabling it to more accurately describe the types of nuclides and their relative abundance or activity. Therefore, during spectral analysis, this model can be used to fit and analyze unknown gamma-ray spectral data, thereby extracting information on the types and activities of nuclides.

[0061] Specifically, the spectral decomposition process involves comparing and matching the unknown gamma-ray energy spectrum data with the optimized nuclide composition model. By continuously adjusting the model parameters, the energy spectrum output by the model is made as consistent as possible with the unknown gamma-ray energy spectrum data. Various mathematical methods and algorithms can be used to assist in this process, such as least squares and maximum likelihood estimation. Finally, when the degree of matching between the model and the data reaches a preset standard, the spectral decomposition is considered complete, and the nuclide type and activity information can be extracted from the model.

[0062] This embodiment provides a nonlinear iterative method for optimizing gamma-ray spectrum analysis. Known gamma-ray spectrum data is acquired and preprocessed to obtain preprocessed known gamma-ray spectrum data. The preprocessing includes at least wavelet thresholding denoising, energy drift correction, and Compton continuum subtraction. Nuclide features are extracted from the preprocessed known gamma-ray spectrum data to obtain the positions and corresponding intensity information of nuclide feature peaks. Nuclide feature extraction includes at least peak detection, full-energy peak determination, and feature vector construction. Based on the nuclide feature peak positions and corresponding intensity information, an initial nuclide composition model is constructed. The initial nuclide composition model is optimized using a nonlinear iterative algorithm to obtain an optimized nuclide composition model. The nonlinear iterative algorithm includes one or more combinations of artificial firefly optimization algorithms, culture algorithms, or simulated annealing algorithms. The optimized nuclide composition model is then used to analyze unknown gamma-ray spectrum data to obtain the types and activity information of nuclides. By introducing a nonlinear iterative algorithm to adaptively adjust the parameters of the nuclide composition model, and then using the optimized nuclide composition model to interpret the unknown γ-ray spectrum data, the accuracy and efficiency of γ-ray spectrum interpretation are significantly improved.

[0063] Based on the first embodiment of this application, in the second embodiment of this application, the content that is the same as or similar to that in the first embodiment described above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 2 Step S10 includes steps S101 to S104: Step S101: Acquire known γ-ray spectrum data using a high-purity germanium detector or a scintillation detector.

[0064] It should be noted that, in this embodiment, gamma-ray spectrum data can be acquired using either a high-purity germanium detector or a scintillation detector. Both detectors are characterized by high sensitivity and high resolution, enabling accurate measurement of gamma-ray energy and intensity. High-purity germanium detectors, with their excellent energy resolution and linearity, perform exceptionally well in gamma-ray spectrum measurements and are suitable for detailed analysis of complex gamma-ray spectra. Scintillation detectors, on the other hand, are known for their fast response and high detection efficiency, making them suitable for real-time monitoring and rapid measurement scenarios. In practical applications, the appropriate detector type can be selected based on measurement requirements and experimental conditions.

[0065] In practice, during the acquisition process, a detector is placed near the sample to capture gamma rays emitted by the radioactive decay of the sample. These gamma rays interact with the material within the detector, generating measurable electrical signals. Through appropriate electronics and data processing algorithms, these electrical signals are converted into gamma-ray spectrum data, which contains information about the energy distribution of the gamma rays.

[0066] Step S102: Adaptive denoising is performed on the known γ energy spectrum data according to the Marat multi-resolution analysis algorithm and the Stein unbiased risk threshold method to obtain the denoised known γ energy spectrum data.

[0067] It should be noted that the Marat multi-resolution analysis algorithm is a wavelet analysis technique that decomposes a signal into wavelet coefficients of different frequency components through multi-scale decomposition. These coefficients reflect the local characteristics of the signal at different scales, allowing important signal features to be effectively extracted at different resolutions. In gamma-ray spectrum data processing, the Marat multi-resolution analysis algorithm is used for adaptive denoising of known gamma-ray spectrum data. This algorithm can automatically select appropriate decomposition scales and thresholds based on the characteristics of the gamma-ray spectrum data to process the wavelet coefficients, removing noise components while preserving important signal features. This method yields denoised known gamma-ray spectrum data with a significantly improved signal-to-noise ratio. Combining this with the Stein unbiased risk thresholding method can further optimize the denoising effect, ensuring optimal results when processing complex gamma-ray spectrum data.

[0068] In one feasible implementation, step S102 may include: obtaining wavelet basis functions and the number of decomposition levels, wherein the wavelet basis functions are one or more combinations of Daubechies wavelet, Symlets wavelet, or Coiflets wavelet, and the number of decomposition levels is 3 to 5; decomposing the known γ-ray spectrum data into approximation coefficients and detail coefficients at different scales according to the wavelet basis functions and the number of decomposition levels; applying a soft thresholding function to threshold the detail coefficients to obtain processed detail coefficients, wherein the threshold is determined according to the Stein unbiased risk estimation principle; and reconstructing the processed detail coefficients and approximation coefficients to obtain the denoised known γ-ray spectrum data.

[0069] It should be noted that in this embodiment, one or more combinations of Daubechies wavelet, Symlets wavelet, or Coiflets wavelet are used as wavelet basis functions. These wavelet basis functions possess compact support, orthogonality, and good frequency domain localization properties, making them suitable for fine analysis of gamma-ray spectrum data. The number of decomposition levels is determined based on the complexity and noise level of the gamma-ray spectrum data, typically between 3 and 5 levels. By appropriately selecting the wavelet basis functions and the number of decomposition levels, effective multi-scale decomposition of the gamma-ray spectrum data can be ensured.

[0070] During the decomposition process, the known gamma-ray spectrum data is decomposed into approximation coefficients and detail coefficients at different scales. The approximation coefficients reflect the low-frequency components of the signal, while the detail coefficients reflect the high-frequency components, including noise and local features. The decomposition formula is as follows:

[0071] in, These are the approximation coefficients of the k-th layer, representing the low-frequency component. These are the detail coefficients of the k-th layer, representing the high-frequency components. It approximates the basis functions. These are detailed basis functions.

[0072] By thresholding the detail coefficients, noise components can be effectively removed while preserving important signal features. In this embodiment, a soft thresholding function is used to threshold the detail coefficients. The soft thresholding function is a commonly used thresholding method that shrinks the detail coefficients according to a preset threshold, setting coefficients smaller than the threshold to zero and subtracting the threshold from coefficients larger than the threshold. This method can effectively remove noise while avoiding excessive smoothing of important signal features. The choice of threshold is crucial; thresholds that are too large or too small may lead to poor noise reduction. Therefore, in this embodiment, the threshold is determined according to the Stein unbiased risk estimation principle. Stein unbiased risk estimation is an adaptive threshold selection method that can automatically determine the optimal threshold based on the characteristics of the signal, thereby ensuring optimal noise reduction. Finally, the processed detail coefficients and approximation coefficients are reconstructed to obtain the known γ-ray spectrum data after noise reduction. The reconstruction process is the process of recombinating the coefficients after multi-scale decomposition into the original signal. Through a reasonable reconstruction algorithm, the denoised γ-ray spectrum data can be obtained, and its signal-to-noise ratio is significantly improved.

[0073] Step S103: Use the dynamic time warping algorithm to perform energy axis calibration on the denoised known γ energy spectrum data to obtain calibrated known γ energy spectrum data, wherein the energy axis calibration is used to correct energy drift.

[0074] It should be noted that in gamma-ray spectroscopy measurements, factors such as detector response, ambient temperature, and electromagnetic interference can cause the measured gamma-ray spectral data to drift along the energy axis. This drift can shift the position of the energy peaks in the spectral data, thus affecting the accuracy of subsequent nuclide identification and quantitative analysis. To address this issue, this implementation uses a dynamic time warping algorithm to calibrate the energy axis of the denoised known gamma-ray spectral data.

[0075] Dynamic Time Warping (DTW) is a nonlinear programming method used to measure the similarity between two time series. It uses dynamic programming to stretch or shorten the time series to achieve an optimal shape match. In energy axis calibration of gamma-ray spectrum data, denoised known gamma-ray spectrum data can be matched with a standard gamma-ray spectrum data as two time series. DTW finds an optimal time warping path that ensures the denoised known gamma-ray spectrum data achieves the best match with the standard gamma-ray spectrum data on the energy axis, thus correcting energy drift.

[0076] In one feasible implementation, step S103 may include: acquiring reference γ-ray spectrum data, wherein the reference γ-ray spectrum data is γ-ray spectrum data with known energy resolution and energy scale; calculating the dynamic time warping distance between the denoised known γ-ray spectrum data and the reference γ-ray spectrum data; and performing energy axis translation and / or scaling transformation on the denoised known γ-ray spectrum data based on the dynamic time warping distance to obtain calibrated known γ-ray spectrum data, wherein the energy axis translation and / or scaling transformation is used to correct energy drift.

[0077] It should be noted that in this embodiment, the reference gamma spectrum data is gamma spectrum data with known energy resolution and energy scale, which serves as the calibration standard. The degree of mismatch between the denoised known gamma spectrum data and the reference gamma spectrum data can be quantified by calculating the dynamic time-warped distance. The smaller the dynamic time-warped distance, the higher the similarity between the two sequences, and the smaller the drift on the energy axis. The calculation of the dynamic time-warped distance involves defining a difference metric function for the two sequences, which can use absolute difference (or other metrics): d(i,j)=|r i -d j | Then, construct a two-dimensional cumulative cost matrix D. DTW This is used to record the minimum cumulative cost from a point in one sequence to another. Each element in the 7-matrix represents the cost from R(t) to R(t). i ) to D(t) j The minimum matching cost of ). The recursive formula is as follows: D DTW (i,j)=∣r i -d j |+min(D DTW (i-1,j),D DTW (i,j-1),D DTW (i-1,j-1)) Among them, D DTW (i,j) is the minimum cumulative cost to reach R(i) and D(j), |r i -d j| is a measure of the difference between the current two sequence points.

[0078] The goal of calculating the DTW distance is to minimize the cumulative cost, which gives the distance between the two sequences.

[0079] Based on this distance, energy axis translation and / or scaling transformations can be performed on the denoised known gamma-ray spectrum data to correct energy drift. Energy axis translation adjusts the overall position of the spectrum data on the energy axis, while energy axis scaling adjusts the distribution range of the spectrum data on the energy axis. Through these two transformations, the denoised known gamma-ray spectrum data can achieve optimal matching with the reference gamma-ray spectrum data on the energy axis, thereby improving the accuracy of subsequent nuclide identification and quantitative analysis. Finally, the calibrated known gamma-ray spectrum data is obtained.

[0080] Step S104: Based on the calibrated known γ energy spectrum data, the Compton edge fitting method is used to subtract the Compton continuous spectrum to obtain the preprocessed known γ energy spectrum data.

[0081] It should be noted that in gamma-ray spectroscopy measurements, the Compton continuum is a continuous energy distribution generated by Compton scattering of gamma rays with electrons in matter. The presence of this continuum can interfere with the identification of characteristic peaks of nuclides, thus affecting the accuracy of quantitative nuclide analysis. To eliminate this interference, this implementation method uses the Compton edge fitting method to subtract the Compton continuum from the calibrated known gamma-ray spectral data.

[0082] The Compton edge fitting method is a mathematical model based on the physical process of Compton scattering. It accurately estimates the shape and intensity of the continuous spectrum produced by Compton scattering by fitting the spectrum to the data. When subtracting the continuous spectrum, the location of the Compton edge is first determined based on the energy range and resolution of the calibrated known gamma-ray spectrum data. Then, an appropriate mathematical function (such as an exponential function or a power function) is used to fit the Compton continuous spectrum, obtaining a fitted curve. Finally, the fitted curve is subtracted from the calibrated known gamma-ray spectrum data to obtain the known gamma-ray spectrum data after subtracting the Compton continuous spectrum.

[0083] In one feasible implementation, step S104 may include: determining the Compton edge energy position, wherein the Compton edge energy position is determined based on the Compton scattering formula and the energy range of the calibrated known γ energy spectrum data; fitting the energy spectrum data near the Compton edge position using a nonlinear fitting algorithm to obtain a Compton edge fitting curve; and subtracting the Compton continuous spectrum in the calibrated known γ energy spectrum data from the Compton edge fitting curve to obtain preprocessed known γ energy spectrum data.

[0084] It should be noted that, in this embodiment, the Compton edge energy position is determined based on the Compton scattering formula and the energy range of the known γ-ray energy spectrum data after calibration. The Compton scattering formula describes the energy distribution of scattered photons when γ-rays scatter with electrons in matter. Using this formula, the edge position of the continuous spectrum produced by Compton scattering can be calculated within a given energy range. This position is one of the key parameters for fitting the Compton continuous spectrum.

[0085] A nonlinear fitting algorithm is used to fit the energy spectrum data near the Compton edge to obtain the Compton edge fitting curve. The nonlinear fitting algorithm is a mathematical method that, given data points and a pre-defined mathematical model, finds the optimal model parameters so that the model best describes the distribution of the data points. In this embodiment, an appropriate mathematical function (such as an exponential function or a power function) is used as the fitting model to fit the energy spectrum data near the Compton edge. By continuously adjusting the model parameters, the error between the fitted curve and the data points is minimized, thus obtaining the optimal Compton edge fitting curve.

[0086] Finally, by subtracting the Compton continuum from the calibrated known gamma spectrum data based on the Compton edge fitting curve, the preprocessed known gamma spectrum data can be obtained. The subtraction process involves subtracting the fitted curve from the original spectrum data to obtain the spectrum data after subtracting the continuum. This process effectively eliminates the interference of the Compton continuum on the identification of nuclide characteristic peaks, improving the accuracy of subsequent quantitative analysis of nuclides.

[0087] In this embodiment, a high-purity germanium detector or a scintillation detector is used to acquire known gamma-ray spectrum data. Adaptive denoising is then performed on the known gamma-ray spectrum data using the Marat multi-resolution analysis algorithm and the Stein unbiased risk thresholding method to obtain denoised known gamma-ray spectrum data. The energy axis of the denoised known gamma-ray spectrum data is then calibrated using a dynamic time warping algorithm to obtain calibrated known gamma-ray spectrum data, where energy axis calibration is used to correct energy drift. Based on the calibrated known gamma-ray spectrum data, the Compton edge fitting method is used to subtract the Compton continuum, resulting in preprocessed known gamma-ray spectrum data. Through the above methods, preprocessing operations such as wavelet threshold denoising, energy drift correction, and Compton continuum subtraction significantly improve the quality and reliability of the known gamma-ray spectrum data, thereby effectively improving the accuracy of spectral interpretation.

[0088] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the nonlinear iterative γ-energy spectrum optimization method of this application. Any simple transformations based on this technical concept are within the protection scope of this application.

[0089] This application also provides a γ-ray energy spectrum optimization device based on nonlinear iteration, please refer to... Figure 3 The nonlinear iterative γ-ray spectrum optimization device includes: The preprocessing module 10 is used to acquire known γ energy spectrum data and preprocess the known γ energy spectrum data to obtain preprocessed known γ energy spectrum data. The preprocessing includes at least wavelet threshold denoising, energy drift correction and Compton continuity subtraction.

[0090] The extraction module 20 is used to extract nuclide features from the preprocessed known γ-ray spectrum data to obtain the nuclide feature peak positions and corresponding intensity information. The nuclide feature extraction includes at least peak detection, full-energy peak determination and feature vector construction.

[0091] The construction module 30 is used to construct an initial nuclide composition model based on the position of the characteristic peaks of the nuclide and the corresponding intensity information.

[0092] The optimization module 40 is used to optimize the initial nuclide composition model using a nonlinear iterative algorithm to obtain an optimized nuclide composition model. The nonlinear iterative algorithm includes one or more combinations of artificial firefly optimization algorithm, culture algorithm or simulated annealing algorithm.

[0093] The spectral decomposition module 50 is used to decompose unknown γ-ray spectral data according to the optimized nuclide composition model to obtain the type and activity information of nuclides.

[0094] The nonlinear iterative gamma-ray spectrum optimization device provided in this application employs the nonlinear iterative gamma-ray spectrum optimization method described in the above embodiments, which can solve the technical problems of poor accuracy and efficiency in existing gamma-ray spectrum optimization methods. Compared with the prior art, the beneficial effects of the nonlinear iterative gamma-ray spectrum optimization device provided in this application are the same as those of the nonlinear iterative gamma-ray spectrum optimization method provided in the above embodiments, and other technical features in the nonlinear iterative gamma-ray spectrum optimization device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0095] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A method for optimizing γ-ray energy spectrum resolution based on nonlinear iteration, characterized in that, The method includes: Acquire known gamma energy spectrum data and preprocess the known gamma energy spectrum data to obtain preprocessed known gamma energy spectrum data. The preprocessing includes at least wavelet threshold denoising, energy drift correction and Compton continuum subtraction. Nuclide feature extraction is performed on the preprocessed known γ-ray spectrum data to obtain the nuclide feature peak positions and corresponding intensity information. The nuclide feature extraction includes at least peak detection, full-energy peak determination, and feature vector construction. Based on the position of the characteristic peaks of the nuclides and the corresponding intensity information, an initial nuclide composition model is constructed; The initial nuclide composition model is optimized using a nonlinear iterative algorithm to obtain an optimized nuclide composition model. The nonlinear iterative algorithm includes one or more combinations of artificial firefly optimization algorithm, culture algorithm, or simulated annealing algorithm. Based on the optimized nuclide composition model, the unknown gamma-ray spectrum data were analyzed to obtain the types and activity information of the nuclides.

2. The method as described in claim 1, characterized in that, The process of acquiring known gamma-ray spectrum data and preprocessing the known gamma-ray spectrum data to obtain preprocessed known gamma-ray spectrum data includes: Known gamma spectrum data were acquired using a high-purity germanium detector or a scintillation detector. Adaptive denoising is performed on the known γ energy spectrum data using the Marat multi-resolution analysis algorithm and the Stein unbiased risk threshold method to obtain the denoised known γ energy spectrum data. The known γ energy spectrum data after noise reduction is calibrated using a dynamic time warping algorithm to obtain calibrated known γ energy spectrum data, wherein the energy axis calibration is used to correct energy drift. Based on the calibrated known γ-ray spectrum data, the Compton edge fitting method is used to subtract the Compton continuous spectrum to obtain the preprocessed known γ-ray spectrum data.

3. The method as described in claim 2, characterized in that, The adaptive denoising of the known γ-ray spectrum data based on the Marat multi-resolution analysis algorithm and the Stein unbiased risk threshold method yields the denoised known γ-ray spectrum data, including: Obtain the wavelet basis function and the number of decomposition levels, wherein the wavelet basis function is one or more combinations of Daubechies wavelet, Symlets wavelet or Coiflets wavelet, and the number of decomposition levels is 3 to 5. Based on the wavelet basis function and the number of decomposition levels, the known γ-energy spectrum data is decomposed into approximation coefficients and detail coefficients at different scales; The detail coefficients are thresholded using a soft thresholding function to obtain the processed detail coefficients, wherein the threshold is determined according to the Stein unbiased risk estimation principle; The processed detail coefficients and approximation coefficients are reconstructed to obtain the known γ-ray energy spectrum data after noise reduction.

4. The method as described in claim 2, characterized in that, The step of using a dynamic time warping algorithm to calibrate the energy axis of the denoised known gamma spectrum data to obtain calibrated known gamma spectrum data includes: Obtain reference gamma spectrum data, wherein the reference gamma spectrum data is gamma spectrum data with known energy resolution and energy scale; Calculate the dynamic time-normalized distance between the denoised known gamma spectrum data and the reference gamma spectrum data; Based on the dynamic time warping distance, the known γ energy spectrum data after noise reduction is subjected to energy axis translation and / or scaling transformation to obtain calibrated known γ energy spectrum data, wherein the energy axis translation and / or scaling transformation is used to correct energy drift.

5. The method as described in claim 2, characterized in that, The known gamma spectrum data, based on the calibrated known gamma spectrum data, is obtained by subtracting the Compton continuum using the Compton edge fitting method, resulting in preprocessed known gamma spectrum data, including: The Compton edge energy position is determined based on the Compton scattering formula and the energy range of the calibrated known gamma spectrum data. A nonlinear fitting algorithm was used to fit the energy spectrum data near the Compton edge location to obtain the Compton edge fitting curve; The preprocessed known gamma spectrum data is obtained by subtracting the Compton continuum from the calibrated known gamma spectrum data based on the Compton edge fitting curve.

6. The method as described in claim 1, characterized in that, The step of extracting nuclide features from the preprocessed known γ-ray spectrum data to obtain the nuclide feature peak positions and corresponding intensity information includes: Peak detection was performed on the preprocessed known γ-ray spectrum data using an improved golden section search method and Parzen window density estimation method to identify potential full-energy peak positions. Based on the preset full-energy peak energy range and the potential full-energy peak positions, the full-energy peaks in the preprocessed known γ-ray spectrum data are determined; Determine the Zernike moments of each of the full-energy peaks, and construct eigenvectors based on the Zernike moments; The position of the nuclide characteristic peak and the corresponding intensity information are determined based on the feature vector.

7. The method as described in claim 1, characterized in that, The optimization of the initial nuclide composition model using a nonlinear iterative algorithm to obtain the optimized nuclide composition model includes: An initial firefly population is set up, wherein each firefly in the initial firefly population represents an initial nuclide composition model; The fitness value of each firefly is calculated according to an objective function, wherein the objective function is used to evaluate the degree of matching between the nuclide composition model and known gamma spectral data; The firefly position and brightness are updated based on the fitness value to obtain the updated firefly population, where the firefly position is a nuclide composition model parameter and the firefly brightness is a fitness value. The process iteratively executes the steps of calculating the fitness value of each firefly according to the objective function and updating the position and brightness of the fireflies according to the fitness value to obtain the updated firefly population, until a preset iteration termination condition is met to obtain the optimal firefly position. The iteration termination condition includes reaching the maximum number of iterations or the fitness value converging. Based on the determined optimized nuclide composition model parameters and the parameter update of the initial nuclide composition model, the optimized nuclide composition model is obtained.

8. The method as described in claim 1, characterized in that, The step of optimizing the initial nuclide composition model using a nonlinear iterative algorithm to obtain the optimized nuclide composition model further includes: An initial cultural population is set up, wherein each individual in the initial cultural population represents an initial nuclide composition model and a corresponding belief space; The fitness value of each individual is evaluated based on the knowledge base and rule base in the belief space. The knowledge base contains prior knowledge of the nuclide composition model, and the rule base is used to guide the optimization process of the nuclide composition model. The individuals in the cultural population are updated according to the fitness values ​​to obtain the updated cultural population. The update process includes updating the belief space and evolving the population space. The steps of evaluating the fitness value of each individual based on the knowledge base and rule base in the belief space and updating the individuals in the cultural population based on the fitness value are executed iteratively until a preset iteration termination condition is met to obtain the optimal individual. The iteration termination condition includes reaching the maximum number of iterations or the fitness value converges. Based on the belief space and population space information of the optimal individual, the optimized nuclide composition model parameters are determined, and the parameters of the initial nuclide composition model are updated to obtain the optimized nuclide composition model.

9. The method as described in claim 1, characterized in that, The step of optimizing the initial nuclide composition model using a nonlinear iterative algorithm to obtain the optimized nuclide composition model further includes: An initial temperature is set, and an initial solution is generated based on the initial temperature, wherein the initial solution represents an initial nuclide composition model; The fitness value of the initial solution is calculated according to the objective function, wherein the objective function is used to evaluate the degree of matching between the nuclide composition model and the known γ-ray spectral data; The temperature is reduced according to a preset cooling strategy, and a new solution is accepted according to the Metropolis criterion to obtain an updated solution. The new solution is obtained by randomly perturbing the current solution. The Metropolis criterion is used to accept solutions that worsen the objective function value with a preset probability. The steps of calculating the fitness value of the initial solution according to the objective function, reducing the temperature according to the preset cooling strategy, and accepting new solutions according to the Metropolis criterion are executed iteratively until the preset iteration termination condition is met to obtain the optimal solution. The iteration termination condition includes reaching the minimum temperature or the fitness value converges. Based on the optimal solution, the optimized nuclide composition model parameters are determined, and the parameters of the initial nuclide composition model are updated to obtain the optimized nuclide composition model.

10. A γ-ray energy spectrum optimization device based on nonlinear iteration, characterized in that, The γ-ray spectrum optimization device based on nonlinear iteration includes: The preprocessing module is used to acquire known γ energy spectrum data and preprocess the known γ energy spectrum data to obtain preprocessed known γ energy spectrum data. The preprocessing includes at least wavelet threshold denoising, energy drift correction and Compton continuity subtraction. The extraction module is used to extract nuclide features from the preprocessed known γ-ray spectrum data to obtain the nuclide feature peak positions and corresponding intensity information. The nuclide feature extraction includes at least peak detection, full-energy peak determination, and feature vector construction. The construction module is used to construct an initial nuclide composition model based on the position of the characteristic peaks of the nuclide and the corresponding intensity information; The optimization module is used to optimize the initial nuclide composition model using a nonlinear iterative algorithm to obtain an optimized nuclide composition model. The nonlinear iterative algorithm includes one or more combinations of artificial firefly optimization algorithm, culture algorithm, or simulated annealing algorithm. The spectrum interpretation module is used to interpret unknown gamma-ray spectrum data based on the optimized nuclide composition model to obtain information on the type and activity of nuclides.

Images