A prompt gamma neutron activation analysis device based on a d-d neutron generator

Abstract

This invention discloses a transient gamma neutron activation analysis device based on a D-D neutron generator, belonging to the field of nuclear analysis technology. The device includes a nested gamma-ray shielding outer shell and a neutron shielding inner shell, together forming an analysis chamber. The D-D neutron generator is located within the inner shell. The fast neutrons it generates are heated by a high-density polyethylene moderator of a specific thickness, exciting the sample placed in the analysis chamber to produce transient gamma rays. A high-resolution semiconductor detector detects the rays and converts them into digital energy spectrum data via a data acquisition and processing module. The artificial intelligence analysis module employs a dual-track analysis system, which can choose to perform real-time online analysis of the energy spectrum data using either a deep model from a host computer or a lightweight model embedded in the terminal, retrieving sample composition information. This invention fundamentally solves the safety and portability problems of traditional devices and achieves rapid, high-precision intelligent analysis on-site.

Description

Technical Field

[0001] This invention relates to the field of nuclear analysis technology, and specifically to a transient gamma neutron activation analysis device based on a DD neutron generator. Background Technology

[0002] Prompt Gamma Neutron Activation Analysis (PGNAA) is a technique that uses neutron irradiation on a sample to analyze the composition of the material by detecting the characteristic gamma rays released instantaneously after the sample is excited. It has the advantages of being non-destructive, highly sensitive, and capable of deep penetration, and is widely used in fields such as security inspection, industrial testing, and geological exploration.

[0003] Traditional PGNAA devices typically employ isotopic neutron sources (such as Cf-252) or DT (deuterium-deuterium) neutron generators. DT neutron generators utilize deuterium (D) nuclei to bombard a tritium (T) target, producing fast neutrons with energies up to 14 MeV. However, their targets contain the radioactive nuclide tritium, posing a risk of long-term storage and potential leakage. Furthermore, high-energy neutrons require stringent shielding and detector protection. In contrast, DD (deuterium-deuterium) neutron generators produce fast neutrons of approximately 2.45 MeV through deuterium-deuterium reactions. Their targets do not involve radioactive tritium, offering greater safety, and the neutron energies are also lower. However, existing PGNAA technologies based on DD neutron generators suffer from low neutron yields, resulting in weak characteristic gamma signals. Conventional scintillator detectors struggle to effectively extract these weak signals from high background radiation, and the reliance on complex offline spectral analysis algorithms leads to lengthy analysis times.

[0004] Therefore, existing technologies make it difficult to construct a PGNAA device that combines high security, good portability, and the ability to achieve real-time, high-precision intelligent analysis on-site. Summary of the Invention

[0005] The purpose of this invention is to provide a transient gamma neutron activation analysis device based on a DD neutron generator to overcome the above-mentioned problems in the prior art. Through a specific hardware architecture and intelligent analysis software, it can achieve rapid, high-precision and safe on-site analysis of material composition.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A transient gamma neutron activation analysis device based on a DD neutron generator, comprising: A gamma-ray shielding shell, the shell wall of which is made of gamma-ray shielding material, and an analytical chamber is formed inside the gamma-ray shielding shell to accommodate the sample to be tested; A neutron-shielded inner shell is disposed within the analysis chamber, and its shell wall is made of neutron-shielding material. A DD neutron generator is located inside the neutron shielding shell and is used to generate fast neutrons; A neutron moderator is disposed on the shell of the neutron shielding inner shell and located in the fast neutron emission path of the DD neutron generator. It is used to slow down the fast neutrons into thermal neutrons to excite the sample to be tested placed in the analysis chamber to generate transient gamma rays. A high-resolution semiconductor detector is disposed in the analysis chamber with its detection end facing the sample to be tested, for detecting the transient gamma rays and outputting a corresponding analog electrical signal; The data acquisition and processing module is electrically connected to the high-resolution semiconductor detector and is used to receive and process the analog electrical signal to generate digital energy spectrum data. An artificial intelligence analysis module, which is communicatively connected to the data acquisition and processing module, is used to analyze the digital energy spectrum data to obtain the composition information of the sample to be tested.

[0007] Furthermore, the neutron moderator is made of high-density polyethylene, and its thickness in the fast neutron emission direction is 3cm to 5cm.

[0008] Furthermore, the high-resolution semiconductor detector is a high-purity germanium detector and is coupled with an electromechanical refrigerator.

[0009] Furthermore, the data acquisition and processing module includes a charge-sensitive preamplifier, an analog-to-digital converter, and a field-programmable gate array connected in sequence. The field-programmable gate array is used to execute a digital pulse processing algorithm to construct the digital energy spectrum data.

[0010] Furthermore, the gamma-ray shielding material of the outer shell is lead, and the neutron shielding material of the inner shell includes polyethylene and boron-containing polyethylene.

[0011] Furthermore, the neutron yield of the DD neutron generator can be dynamically adjusted by regulating its operating voltage and beam current.

[0012] Furthermore, the artificial intelligence parsing module includes a dual-track parsing system, which is configured to selectively execute at least one of the following parsing paths: The first path involves transmitting the digital energy spectrum data to an external host computer, where it is analyzed by a deep neural network model deployed on the host computer. The second approach involves a lightweight artificial intelligence model performing real-time analysis of the digital energy spectrum data within an embedded artificial intelligence processor integrated into the device.

[0013] Furthermore, the lightweight artificial intelligence model is obtained by performing network pruning and low-bit quantization on the deep neural network model.

[0014] Furthermore, the deep neural network model and the lightweight artificial intelligence model are one-dimensional convolutional neural networks, and are trained using a mixed dataset containing simulated transient gamma energy spectrum data and measured energy spectrum data.

[0015] Another object of the present invention is to provide a method for transient gamma neutron activation analysis, the method being used in a transient gamma neutron activation analysis device based on a DD neutron generator, comprising the following steps: The DD neutron generator is activated to produce fast neutrons. The fast neutrons are slowed down into thermal neutrons by the neutron moderator and then excite the sample to be tested to produce transient gamma rays. The transient gamma rays are detected by the high-resolution semiconductor detector and converted into analog electrical signals; The analog electrical signal is processed by the data acquisition and processing module to obtain digital energy spectrum data; The digital energy spectrum data is input into the artificial intelligence analysis module; The artificial intelligence analysis module analyzes the digital energy spectrum data and outputs the composition information of the sample to be tested.

[0016] Compared with the prior art, the present invention has the following beneficial effects: I. High safety and portability. Utilizing a DD neutron generator, neutrons are produced only during operation, completely eliminating the risks of continuous radiation from isotope sources and tritium leakage from DT sources. Furthermore, the required shielding is thinner, significantly reducing the size and weight of the device, making it easier to deploy in the field.

[0017] Second, high detection sensitivity. Employing high-resolution semiconductor detectors such as high-purity germanium, whose excellent energy resolution significantly improves the peak-to-background ratio of weak characteristic signals, effectively overcoming the signal weakness problem caused by the low neutron yield of DD sources.

[0018] Third, the analysis is fast and intelligent. An artificial intelligence model is introduced to replace the traditional complex spectral analysis algorithm. Through a dual-track analysis scheme—a deep model on the host computer and a lightweight model on the embedded end—it achieves a leap from offline analysis to online real-time inversion. The analysis time can reach the millisecond to hundreds of milliseconds level, meeting the on-site read-second-level analysis requirements.

[0019] Fourth, it has high analytical accuracy and strong adaptability. The AI ​​model is trained with massive amounts of simulation and experimental mixed data, and has powerful feature extraction and generalization capabilities. It can effectively handle complex energy spectra with high background, weak features and strong statistical fluctuations, and improve the accuracy of nuclide identification and quantitative analysis. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the structure of the transient gamma neutron activation analysis device based on the DD neutron generator provided in an embodiment of the present invention; In the figure: 1-Gamma ray shielding shell, 2-Analysis chamber, 3-Neutron shielding inner shell, 4-DD neutron generator, 5-Neutron moderator, 6-High-resolution semiconductor detector, 7-Data acquisition and processing module, 8-Artificial intelligence analysis module, 9-Sample to be tested, 10-Power supply and control system. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0022] like Figure 1 As shown in this embodiment, a transient gamma neutron activation analysis device based on a DD neutron generator is provided. Its core lies in an innovative double-shell shielding structure and a complete detection and analysis chain integrating real-time AI analysis functionality. The device mainly includes: a gamma-ray shielding outer shell 1, an analysis chamber 2, a neutron shielding inner shell 3, a DD neutron generator 4, a neutron moderator 5, a high-resolution semiconductor detector 6, a data acquisition and processing module 7, and an AI analysis module 8. The sample to be tested 9 is placed inside the analysis chamber 2. The various electronic functional modules of the device, including the DD neutron generator 4, the high-resolution semiconductor detector 6 and its cooler, the data acquisition and processing module 7, and the AI ​​analysis module 8, are coordinated by an integrated power supply and control system 10 to ensure stable operation of each module.

[0023] The gamma-ray shielding shell 1 constitutes the main housing of the device, and its shell wall is made of gamma-ray shielding material, such as lead bricks or lead layers. This shell encloses a sealed analysis chamber 2, specifically designed to contain the sample 9 and perform gamma-ray detection, thus achieving environmental shielding and external radiation protection for the detection area. The neutron shielding inner shell 3, as an independent sub-shielding body, is located within the analysis chamber 2. Its shell wall is made of neutron shielding material; in this embodiment, a composite structure of pre-moderating and post-absorbing is adopted, comprising a polyethylene layer and a boron-containing polyethylene layer from the inside out. The polyethylene layer is rich in hydrogen atoms, used to modulate high-energy fast neutrons into thermal neutrons through elastic collisions; the boron-containing polyethylene layer utilizes the large absorption cross-section of the boron-10 isotope to efficiently absorb the moderated neutrons, preventing neutron leakage. The lead layer of the gamma-ray shielding shell 1 is used to shield secondary gamma rays generated by neutron capture reactions such as hydrogen radiation capture reactions, as well as the portion of characteristic gamma rays generated by the sample itself that leaks into the environment. This multi-layered synergistic shielding design, which combines polyethylene moderation, boron-containing polyethylene absorption, and lead shielding for gamma, along with the dual-shell spatial partitioning of the inner shell shielding the neutron source and the outer shell shielding the analysis area, not only provides high radiation protection efficiency but also achieves high efficiency and compactness of the shielding structure through spatial partitioning and material optimization. This significantly reduces the necessary size and weight of the device, forming the core structural design for achieving device portability.

[0024] The DD neutron generator 4 is located inside the neutron shielding inner shell 3. Its working principle involves generating a deuterium ion beam through internal ion source ionization, accelerating the deuterium ions to tens of thousands to hundreds of thousands of electron volts using a high-voltage electric field, and then bombarding a solid target rich in deuterium atoms, thereby triggering a deuterium-deuterium nuclear fusion reaction. The DD generator stably produces monoenergetic fast neutrons with an energy of approximately 2.45 MeV. This electrically controlled source generates neutrons only when powered on, completely resolving the persistent radioactivity issue of traditional isotope neutron sources such as Cf-252, as well as the risk of radioactive tritium target leakage from the DT neutron generator, ensuring the intrinsic safety of the device from the source. Furthermore, the 2.45 MeV fast neutrons produced by the DD generator have significantly lower energies than the 14 MeV fast neutrons produced by the DT generator. This not only reduces the requirements for shielding thickness, contributing to device lightweighting, but more importantly, it significantly reduces the risk of radiation damage to the high-resolution semiconductor detector 6, effectively extending the detector's lifespan. In addition, the DD neutron generator 4 is provided with precisely adjustable voltage and current through a power supply and control system 10. The neutron yield of the generator can be linearly and dynamically adjusted by regulating the operating voltage and ion beam current, thereby adapting to the detection needs of samples with different densities and volumes. More importantly, the shielding thickness required for 2.45 MeV fast neutrons is much lower than that for 14 MeV fast neutrons, which directly reduces the weight and overall volume of the shield, laying the physical foundation for the overall lightweighting and miniaturization of the device.

[0025] The neutron moderator 5 is disposed on the shell of the neutron shielding inner shell 3 and located in the fast neutron emission path of the DD neutron generator 4. Its function is to moderate (thermally convert) the 2.45 MeV fast neutrons into thermal neutrons with an energy of approximately 0.025 eV, because thermal neutrons have the largest cross-section for inducing radiative trapping reactions in the sample nuclei. The neutron moderator 5 is made of high-density polyethylene (HDPE). Its key design parameter is its thickness in the fast neutron emission direction, which, through theoretical calculations and experimental optimization, is strictly limited to the range of 3 cm to 5 cm. This optimization range is determined based on the physical balance between fast neutron moderation efficiency and thermal neutron absorption attenuation. If the thickness is less than 3 cm, the fast neutrons penetrate without sufficient thermalization, resulting in insufficient thermal neutron flux at the sample; if the thickness exceeds 5 cm, although the moderation is more thorough, the hydrogen atoms in the polyethylene will strongly absorb the generated thermal neutrons through radiative trapping, leading to a sharp attenuation of the thermal neutron flux reaching the sample. This optimized thickness ensures that the highest thermal neutron flux can be obtained at position 9 of the sample under test, thereby significantly improving the excitation efficiency of characteristic gamma rays and effectively making up for the relatively low neutron yield of the DD source.

[0026] A high-resolution semiconductor detector 6 is disposed within the analysis chamber 2, with its detection end facing the sample 9 to be tested. In this embodiment, the detector is a high-purity germanium detector. High-purity germanium crystals operate at a cryogenic temperature of approximately 77K, exhibiting excellent energy resolution, with a full width at half maximum (FWHM) of less than 2keV at 1.33MeV. When characteristic transient gamma rays generated by thermal neutron excitation of the sample 9 are incident, energy is deposited within the crystal through interactions such as the photoelectric effect and Compton scattering, exciting electron-hole pairs whose number is strictly proportional to the incident photon energy. These charge carriers drift directionally under the influence of the detector's bias electric field, forming an analog electrical signal. The extremely high energy resolution can clearly separate the weak characteristic full-energy peaks from the strong Compton continuous background, greatly improving the peak-to-background ratio (signal-to-noise ratio). This is key to overcoming the weak signal caused by the low neutron yield of DD sources and the inability of traditional scintillator detectors such as NaI(Tl) to identify weak characteristic peaks due to poor resolution. To achieve portability, the detector is coupled with an electromechanical cryostat, such as a Stirling cryostat, replacing the traditional liquid nitrogen Dewar flask. This allows the detector to operate stably for extended periods without liquid nitrogen replenishment. To ensure heat dissipation, the hot end of the cryostat is connected to the outer casing via an internal heat conduction path, allowing heat to dissipate naturally through the casing surface. This design eliminates the need for openings in the shielding, ensuring both heat dissipation and radiation safety. These improvements completely eliminate the logistical dependence on bulky liquid nitrogen Dewar flasks and periodic liquid nitrogen replenishment, enabling the integration of a high-resolution detector into a mobile device. This is a prerequisite for the entire system to be used in field testing outside of laboratory environments. The power supply and control system 10 provides the power to the detector and cryostat, ensuring stable operation in mobile scenarios.

[0027] The data acquisition and processing module 7 is electrically connected to the high-resolution semiconductor detector 6. In this embodiment, it is specifically implemented as a highly integrated Digital Multichannel Analyzer (Digital MCA) architecture. Its hardware link consists of a charge-sensitive preamplifier, a high-speed analog-to-digital converter (ADC), and a field-programmable gate array (FPGA) connected sequentially and electrically. The preamplifier converts the weak charge pulses output by the detector into voltage pulses. The ADC digitizes the analog pulses at a high frequency, such as a sampling rate of 100MHz. The FPGA incorporates digital pulse processing algorithms, such as digital laddering, digital baseline recovery, and anti-pulse stacking algorithms, to process the digitized pulse signal in real time, accurately extract the amplitude of each pulse, and rapidly construct, for example, an 8192-channel digital energy spectrum histogram, i.e., the digital energy spectrum data. This digitization processing method has strong anti-interference capabilities and high energy spectrum quality, providing a clean and reliable data source for subsequent AI analysis.

[0028] The artificial intelligence analysis module 8 is communicatively connected to the data acquisition and processing module 7 and serves as the intelligent brain of this device. Its core lies in a dual-track analysis system that can flexibly select the analysis path based on the computing power and portability requirements of the application scenario.

[0029] The first path (host computer deep model analysis): suitable for laboratory or fixed security inspection scenarios. The data acquisition and processing module 7 transmits the generated digital energy spectrum data to an external high-performance host computer, such as a workstation, via a high-speed communication interface such as Gigabit Ethernet. The host computer runs a deep neural network model, which has a large number of parameters and can mine extremely weak deep features in the energy spectrum to achieve quantitative and qualitative analysis with extreme accuracy.

[0030] The second approach (real-time analysis using an embedded lightweight model): This approach integrates a dedicated embedded artificial intelligence processor, such as an NPU, within the device to run a lightweight model, eliminating the need for external high-performance computers or network connections to analyze complex energy spectra. This significantly enhances the device's independent operational capabilities, enabling it to complete the entire process from detection to analysis in portable scenarios such as field operations and mobile inspections where connecting to a host computer is inconvenient. This is a key software guarantee for achieving real-time intelligent analysis on-site. An embedded artificial intelligence processor, such as a neural network processing unit (NPU), is integrated within the device. To accommodate the limited computing power and memory of embedded devices, the trained deep neural network model undergoes operator fusion, network pruning, and extremely low bit quantization, such as INT8 processing, to obtain a lightweight artificial intelligence model. This model is converted into NPU instructions and stored in the device's memory. The digital energy spectrum data constructed by the FPGA is directly transmitted to the NPU via an internal high-speed system bus, such as the AXI bus. The NPU loads the lightweight model and completes forward inference calculations within milliseconds. The results are output through the device's built-in miniature touchscreen or wireless module.

[0031] The deep neural network model and its lightweight version preferably employ a one-dimensional convolutional neural network (1D-CNN) architecture. This network can automatically extract local shape features of the energy spectrum, such as the full-energy peak morphology, and global distribution patterns through its convolutional kernels. To ensure the model's generalization ability and accuracy in complex physical environments, a hybrid data-driven training approach is adopted: firstly, a massive dataset of simulated transient gamma energy spectra covering multiple sample conditions is generated using Monte Carlo particle transport simulation software such as MCNP; then, measured energy spectrum data obtained on real devices using standard radioactive sources is combined to form a training dataset for supervised learning of the network. This dual-track design breaks the dependence of PGNAA technology on offline processing and high-computing peripherals, achieving a leap from data acquisition terminals to on-site intelligent cognitive terminals.

[0032] This embodiment also provides a method for the activation analysis of transient gamma neutrons, used in the aforementioned device for the activation analysis of transient gamma neutrons based on a DD neutron generator, comprising the following steps: The DD neutron generator 4 is activated to generate fast neutrons. The fast neutrons are slowed down into thermal neutrons by the neutron moderator 5 and then excite the sample to be tested 9 to generate transient gamma rays. The transient gamma rays are detected by the high-resolution semiconductor detector 6 and converted into analog electrical signals; The analog electrical signal is processed by the data acquisition and processing module 7 to obtain digital energy spectrum data; The digital energy spectrum data is input into the artificial intelligence analysis module 8; The artificial intelligence analysis module 8 analyzes the digital energy spectrum data and outputs the composition information of the sample to be tested.

[0033] In one specific embodiment, the sample 9 to be tested is first placed in the analysis chamber 2. The DD neutron generator 4 is activated to generate fast neutrons. After passing through the neutron moderator 5 and being efficiently thermalized, the fast neutrons irradiate the sample 9, exciting it to produce characteristic transient gamma rays. A high-resolution semiconductor detector 6 then detects the transient gamma rays and converts them into analog electrical signals. The data acquisition and processing module 7 amplifies, digitizes, and pulses the analog electrical signals to construct digital energy spectrum data in real time. The digital energy spectrum data is input to the artificial intelligence analysis module 8. Finally, depending on the scenario requirements, the first or second path of the dual-track analysis system is selected, and the corresponding AI model analyzes the digital energy spectrum data, ultimately outputting the elemental composition, content, or identification result of a specific substance category of the sample 9.

[0034] In summary, this embodiment employs an intrinsically safe DD neutron source to reduce shielding requirements, a compact double-layer shielding structure, a high-purity germanium detector that does not require liquid nitrogen, and a built-in embedded AI real-time calculation unit. Through comprehensive optimization of the radiation source, shielding body, detection core, and data analysis chain, it achieves a high degree of integration, lightweight design, and independent operation capability while ensuring safety. This fundamentally solves the problems of traditional PGNAA equipment being bulky, reliant on fixed facilities, and unable to produce results quickly on-site, achieving true portability. The entire process can be completed within seconds, enabling safe, portable, and real-time non-destructive on-site analysis.

[0035] Contents not described in detail in this specification are prior art known to those skilled in the art. It is hereby indicated that the above description is intended to help those skilled in the art understand this invention, but does not limit the scope of protection of this invention. Any equivalent substitutions, modifications, improvements, or simplifications of the above descriptions that do not depart from the essential content of this invention fall within the scope of protection of this invention.

Claims

1. A prompt gamma neutron activation analysis apparatus based on a D-D neutron generator, characterized by, include: A gamma-ray shielding shell (1) has a shell wall made of gamma-ray shielding material, and an analytical chamber (2) is formed inside the gamma-ray shielding shell to accommodate the sample to be tested (9). A neutron-shielded inner shell (3) is disposed inside the analysis chamber (2), and its shell wall is made of neutron-shielding material; A DD neutron generator (4) is disposed inside the neutron shielding shell (3) and is used to generate fast neutrons; The neutron moderator (5) is disposed on the shell of the neutron shielding inner shell (3) and located in the fast neutron emission path of the DD neutron generator (4). It is used to slow down the fast neutron into thermal neutrons to excite the sample (9) to be tested placed in the analysis chamber (2) to generate transient gamma rays. A high-resolution semiconductor detector (6) is disposed in the analysis chamber (2) with its detection end facing the sample to be tested (9) for detecting the transient gamma rays and outputting a corresponding analog electrical signal; The data acquisition and processing module (7) is electrically connected to the high-resolution semiconductor detector (6) and is used to receive and process the analog electrical signal to generate digital energy spectrum data; The artificial intelligence analysis module (8) is connected in communication with the data acquisition and processing module (7) and is used to analyze the digital energy spectrum data to obtain the composition information of the sample to be tested.

2. The prompt gamma neutron activation analysis apparatus based on a D-D neutron generator according to claim 1, characterized in that, The neutron moderator (5) is made of high-density polyethylene and has a thickness of 3 cm to 5 cm in the fast neutron emission direction.

3. The prompt gamma neutron activation analysis apparatus based on a D-D neutron generator according to claim 1, characterized by, The high-resolution semiconductor detector (6) is a high-purity germanium detector and is coupled with an electromechanical refrigerator.

4. The transient gamma neutron activation analysis device based on a DD neutron generator according to claim 1, characterized in that, The data acquisition and processing module (7) includes a charge-sensitive preamplifier, an analog-to-digital converter, and a field-programmable gate array connected in sequence. The field-programmable gate array is used to execute a digital pulse processing algorithm to construct the digital energy spectrum data.

5. The instantaneous gamma neutron activation analysis device based on a DD neutron generator according to claim 1, characterized in that, The gamma-ray shielding material of the shell wall of the gamma-ray shielding outer shell (1) is lead, and the neutron shielding material of the shell wall of the neutron shielding inner shell (3) includes polyethylene and boron-containing polyethylene.

6. The transient gamma neutron activation analysis device based on a DD neutron generator according to claim 1, characterized in that, The neutron yield of the DD neutron generator (4) can be dynamically adjusted by regulating its operating voltage and beam current.

7. A transient gamma neutron activation analysis device based on a DD neutron generator according to any one of claims 1-6, characterized in that, The artificial intelligence parsing module (8) includes a dual-track parsing system configured to selectively execute at least one of the following parsing paths: The first path involves transmitting the digital energy spectrum data to an external host computer, where it is analyzed by a deep neural network model deployed on the host computer. The second approach involves a lightweight artificial intelligence model performing real-time analysis of the digital energy spectrum data within an embedded artificial intelligence processor integrated into the device.

8. The transient gamma neutron activation analysis device based on a DD neutron generator according to claim 7, characterized in that, The lightweight artificial intelligence model is obtained by performing network pruning and low-bit quantization on the deep neural network model.

9. The transient gamma neutron activation analysis device based on a DD neutron generator according to claim 7, characterized in that, The deep neural network model and the lightweight artificial intelligence model are both one-dimensional convolutional neural networks, and are trained using a mixed dataset containing simulated transient gamma energy spectrum data and measured energy spectrum data.

10. A method for activation analysis of transient gamma neutrons, characterized in that, The method is used in any one of the DD neutron generator-based transient gamma neutron activation analysis devices according to claims 1-9, and includes the following steps: The DD neutron generator (4) is activated to generate fast neutrons. The fast neutrons are slowed down into thermal neutrons by the neutron moderator (5) and then excite the sample to be tested (9) to generate transient gamma rays. The transient gamma rays are detected by the high-resolution semiconductor detector (6) and converted into analog electrical signals; The analog electrical signal is processed by the data acquisition and processing module (7) to obtain digital energy spectrum data; The digital energy spectrum data is input into the artificial intelligence analysis module (8). The digital energy spectrum data is analyzed by the artificial intelligence analysis module (8), and the composition information of the sample to be tested is output.

Images