Ultra-low background measurement device and method for discriminating beta and gamma
Through the combined method of liquid flash detector and ultra-high-speed acquisition module combined with convolutional neural network, the problem of difficult γ-ray interference in liquid flash detection device is solved, and the rapid and accurate identification of β and γ and ultra-low background measurement is achieved.
Patent Information
- Application Number
- CN202510646889.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-20
- Publication Date
- 2025-08-15
AI Technical Summary
The existing liquid flash detection device is difficult to effectively eliminate γ-ray interference, which makes it difficult to achieve ultra-low background measurement in the presence of small differences in β and γ pulse waveforms.
Using a combination of liquid flash detector, ultra-high-speed acquisition module and convolutional neural network, fluorescent signals are obtained through liquid scintillators and high-speed photomultiplier tubes, signal conversion and feature extraction are used for data processing units, and β and γ are identified by convolutional neural networks to achieve ultra-low background measurement.
It realizes fast and accurate pulse identification, eliminates γ-ray interference, and achieves the effect of ultra-low background measurement.
Smart Images
Figure CN120491143A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of pulse discrimination and ultra-low background measurement, and in particular to an ultra-low background measurement device and method for discriminating β and γ. Background Art
[0002] Liquid scintillation detectors are radiation detection devices based on liquid scintillators. Their core advantages lie in high detection efficiency and sensitivity, low detection background, a wide variety of particle types, and a wide energy range. These advantages make them irreplaceable in fields such as environmental monitoring, nuclear medicine, and basic physics. Liquid scintillation detectors remain the preferred solution, particularly in scenarios requiring low background and high sensitivity. Alpha radiation has weak penetration and a slow light decay time, while beta radiation has slightly stronger penetration and a faster light decay time. Therefore, for alpha or beta measurement, current commercially available devices primarily achieve low-background measurements through shielding designs, while also enabling alpha and beta discrimination through medium- and high-speed sampling. However, gamma rays have strong penetration, and conventional shielding designs cannot completely eliminate them. The gamma ray pulse waveform is not much different from that of beta particles, making it difficult to eliminate gamma ray interference with conventional shielding and acquisition equipment. In the analysis of ultra-low beta activity samples (such as environmental samples and biological tracers), even low background interference can lead to false positive counts or underestimate the true activity. Therefore, it is necessary to design devices that can achieve ultra-low background, or even near-zero background, measurements.
[0003] Current liquid scintillator detection devices, designed to measure α or β, employ shielding to filter out most γ rays, achieving low-background measurements. However, a small portion still penetrates the shield and appears as background interference. These devices primarily employ medium- to high-speed analog-to-digital converters (ADCs) and one or more algorithms for feature recognition. This approach can meet the requirements for particles with significant differences in α and β pulse waveforms. However, for applications where the differences in β and γ pulse waveforms are smaller, features are difficult to express using algorithms, and bandwidth is extremely high, a medium-speed ADC combined with multiple algorithms for feature recognition struggles. Summary of the Invention
[0004] The purpose of this application is to provide an ultra-low background measurement device and method for distinguishing β and γ, which can quickly and accurately perform pulse discrimination, thereby eliminating the background and achieving ultra-low background measurement.
[0005] To achieve the above objectives, this application provides the following solutions:
[0006] In a first aspect, the present application provides an ultra-low background measurement device for discriminating β and γ, comprising: a liquid scintillation detector and an acquisition module connected in sequence;
[0007] The liquid scintillation detector includes a liquid scintillator and a high-speed photomultiplier tube; the acquisition module includes a data processing unit, a main control unit and a neural network processing unit connected in sequence;
[0008] The liquid scintillator is irradiated by a radiation source to output fluorescence; the fluorescence includes: beta rays and gamma rays;
[0009] The high-speed photomultiplier tube is used to perform photoelectric conversion on fluorescence and output a high-speed current pulse signal;
[0010] The data processing unit is used for:
[0011] Converting the high-speed current pulse signal into a high-speed voltage pulse signal;
[0012] Performing direct bias and gain adjustment on the high-speed voltage pulse signal, and performing analog-to-digital conversion to obtain a digital signal;
[0013] The main control unit is used to preprocess the digital signal, and perform feature processing and pulse position adjustment on the preprocessed digital signal to obtain pulse feature data;
[0014] The convolutional neural network built into the neural network processing unit is used to determine the discrimination result regarding β and γ based on the pulse characteristic data and the digital signal; the convolutional neural network is obtained by training and optimizing the initial convolutional neural network based on the data set; the data set includes the pulse characteristic data with known discrimination results and the corresponding digital signal;
[0015] The main control unit is further configured to perform gamma background rejection according to the identification result to achieve ultra-low background measurement.
[0016] Optionally, the data processing unit specifically includes: a high-speed preamplifier, a signal conditioning module and an ultra-high-speed ADC;
[0017] The high-speed preamplifier is connected to the liquid scintillation detector and the signal conditioning module respectively; the signal conditioning module is also connected to the ultra-high-speed ADC;
[0018] The high-speed preamplifier is used to receive the high-speed current pulse signal and convert the high-speed current pulse signal into a high-speed voltage pulse signal;
[0019] The signal conditioning module is used to:
[0020] performing direct bias adjustment on the high-speed voltage pulse signal to obtain a differential signal;
[0021] Performing gain adjustment on the differential signal to obtain a gain signal;
[0022] The ultra-high-speed ADC is used to perform analog-to-digital conversion on the gain signal to obtain a digital signal.
[0023] Optionally, the main control unit adopts FPGA+ARM architecture.
[0024] Optionally, the main control unit includes: a data acquisition module, a data processing module and a bus control module;
[0025] The data processing module is connected to the data acquisition module and the neural network processing unit respectively via the bus control module; the data acquisition module is also connected to the data processing unit; the bus control module uses a serial bus for data transmission;
[0026] The data acquisition module is used to acquire the digital signal;
[0027] The data processing module is used for:
[0028] Performing data stream serial-to-parallel conversion on the digital signal, and performing data caching, and caching the data into a data pool;
[0029] Read the data in the data pool and perform baseline subtraction, digital filtering and amplitude normalization to obtain the pre-processed digital signal;
[0030] The pre-processed digital signal is subjected to feature processing and pulse position adjustment to obtain pulse feature data; the feature processing includes: feature calculation of amplitude, rise time, fall time, decay time, rising edge slope, falling edge slope, and ratio of trailing edge to whole peak area;
[0031] transmitting the pulse characteristic data and the digital signal to the neural network processing unit;
[0032] Gamma background is eliminated according to the discrimination result determined by the neural network processing unit to achieve ultra-low background measurement.
[0033] Optionally, the acquisition module further includes: a digitally controlled high-voltage module;
[0034] The digitally controlled high-voltage module is connected to the liquid scintillation detector;
[0035] The digitally controlled high-voltage module is used to provide voltage to the liquid scintillation detector.
[0036] Optionally, the liquid scintillation detector further comprises: a voltage divider tube holder;
[0037] The voltage divider tube seat is connected to the digital control high voltage module and the high-speed photomultiplier tube respectively;
[0038] The voltage divider tube seat is used to provide a bias voltage for the high-speed photomultiplier tube according to the voltage.
[0039] Optionally, the acquisition module further includes: a temperature, voltage and current monitoring module;
[0040] The temperature, voltage and current monitoring modules are respectively connected to the main control unit, the neural network processing unit and the ultra-high-speed ADC.
[0041] Optionally, the acquisition module further includes: an air-cooling heat dissipation module;
[0042] The air cooling module is connected to the main control unit;
[0043] The air-cooling heat dissipation module is used for heat dissipation to reduce the device temperature of the main control unit.
[0044] In a second aspect, the present application provides an ultra-low background measurement method for distinguishing β and γ, comprising:
[0045] Obtaining a high-speed current pulse signal and converting it into a high-speed voltage pulse signal; the high-speed current pulse signal is obtained by photoelectric conversion of the fluorescence output by the liquid scintillator through the radiation source based on a high-speed photomultiplier tube;
[0046] Performing direct bias and gain adjustment on the high-speed voltage pulse signal, and performing analog-to-digital conversion to obtain a digital signal;
[0047] Preprocessing the digital signal, and performing feature processing and pulse position adjustment on the preprocessed digital signal to obtain pulse feature data;
[0048] A convolutional neural network is used to determine the discrimination result of β and γ based on the pulse characteristic data and the digital signal; the convolutional neural network is obtained by training and optimizing an initial convolutional neural network based on a data set; the data set includes the pulse characteristic data with known discrimination results and the corresponding digital signal;
[0049] The gamma background is eliminated according to the screening results to achieve ultra-low background measurement.
[0050] Optionally, preprocessing the digital signal, and performing feature processing and pulse position adjustment on the preprocessed digital signal to obtain pulse feature data specifically includes:
[0051] Performing data stream serial-to-parallel conversion on the digital signal, and performing data caching, and caching the data into a data pool;
[0052] Read the data in the data pool and perform baseline subtraction, digital filtering and amplitude normalization to obtain the pre-processed digital signal;
[0053] The pre-processed digital signal is subjected to feature processing and pulse position adjustment to obtain pulse feature data; the feature processing includes: feature calculation of amplitude, rise time, fall time, decay time, rising edge slope, falling edge slope, and ratio of trailing edge to whole peak area.
[0054] According to the specific embodiments provided in this application, this application discloses the following technical effects:
[0055] The present application provides an ultra-low background measurement device and method for discriminating β and γ. For the occasion of measuring β and eliminating the interference of γ rays, it is difficult to eliminate the interference of γ pulses due to the small difference in the waveforms of β and γ pulses and the high bandwidth. Therefore, the present application adopts a liquid scintillation detector, an acquisition module and an artificial intelligence combination, that is, based on the high-speed photomultiplier tube in the liquid scintillation detector, to effectively transmit particle information; then the high-speed current pulse signal is processed based on the data processing unit to effectively ensure that the pulse waveform is not distorted. Finally, the convolutional neural network built into the neural network processing unit is used to quickly discriminate β and γ, determine the discrimination results of β and γ, thereby eliminating the background and achieving ultra-low background measurement. Therefore, the present application can quickly and accurately perform pulse discrimination, thereby eliminating the background and achieving ultra-low background measurement. BRIEF DESCRIPTION OF THE DRAWINGS
[0056] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0057] Figure 1 The structure diagram of the ultra-low background measurement device for distinguishing β and γ;
[0058] Figure 2 Schematic diagram of the overall structure of the ultra-low background measurement device for identifying β and γ;
[0059] Figure 3 This is the circuit structure diagram of the high-speed preamplifier;
[0060] Figure 4 This is a schematic diagram of the circuit structure corresponding to the signal conditioning module;
[0061] Figure 5 This is a structural diagram of the main control unit;
[0062] Figure 6 Schematic diagram of the NPU model deployment process;
[0063] Figure 7 This is a schematic diagram of temperature monitoring by the temperature, voltage and current monitoring module;
[0064] Figure 8 This is a schematic diagram of the temperature, voltage and current monitoring module performing voltage and current monitoring;
[0065] Figure 9 Schematic diagram of the data processing process for convolutional neural networks;
[0066] Figure 10 Flowchart of the ultra-low background measurement method for identifying β and γ. DETAILED DESCRIPTION
[0067] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.
[0068] This application provides a low-noise, high-bandwidth, ultra-high-speed sampling rate module, and uses artificial intelligence algorithms to learn features at different positions and scales, perform pulse discrimination, thereby eliminating background and achieving ultra-low background measurement.
[0069] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.
[0070] In an exemplary embodiment, Figure 1 As shown, an ultra-low background measurement device for distinguishing β and γ is provided, comprising: a liquid scintillation detector and an acquisition module connected in sequence.
[0071] Among them, the liquid scintillation detector includes a liquid scintillator and a high-speed photomultiplier tube; the acquisition module includes a data processing unit, a main control unit and a neural network processing unit connected in sequence.
[0072] The liquid scintillator is irradiated by a radioactive source and outputs fluorescence; the fluorescence includes: beta rays and gamma rays.
[0073] High-speed photomultiplier tubes are used to convert fluorescence into photoelectricity and output high-speed current pulse signals.
[0074] The data processing unit is used to convert the high-speed current pulse signal into a high-speed voltage pulse signal; perform linear deviation and gain adjustment on the high-speed voltage pulse signal, and perform analog-to-digital conversion to obtain a digital signal.
[0075] In one embodiment, the data processing unit specifically includes: a high-speed preamplifier, a signal conditioning module, and an ultra-high-speed ADC.
[0076] The high-speed preamplifier is connected to the liquid scintillation detector and the signal conditioning module respectively; the signal conditioning module is also connected to the ultra-high-speed ADC.
[0077] The high-speed preamplifier is used to receive a high-speed current pulse signal and convert the high-speed current pulse signal into a high-speed voltage pulse signal.
[0078] The signal conditioning module is used to perform direct bias adjustment on the high-speed voltage pulse signal to obtain a differential signal; and perform gain adjustment on the differential signal to obtain a gain signal.
[0079] The ultra-high-speed ADC is used to perform analog-to-digital conversion on the gain signal to obtain a digital signal.
[0080] The main control unit is used to pre-process the digital signal, perform feature processing and pulse position adjustment on the pre-processed digital signal to obtain pulse feature data. The main control unit adopts FPGA+ARM architecture.
[0081] In one embodiment, the main control unit includes: a data acquisition module, a data processing module and a bus control module.
[0082] The data processing module is connected to the data acquisition module and the neural network processing unit respectively via the bus control module; the data acquisition module is also connected to the data processing unit; the bus control module uses a serial bus for data transmission.
[0083] The data acquisition module is used to acquire digital signals.
[0084] The data processing module is used to convert the digital signal into serial and parallel data streams, and cache the data in the data pool; the data processing module is used to read the data in the data pool, and perform baseline subtraction, digital filtering and amplitude normalization to obtain the preprocessed digital signal.
[0085] The data processing module is used to perform feature processing and pulse position adjustment on the pre-processed digital signal to obtain pulse feature data; feature processing includes: feature calculation of amplitude, rise time, fall time, decay time, rising edge slope, falling edge slope, and ratio of trailing edge to whole peak area.
[0086] The data processing module is used to transmit the pulse characteristic data and digital signals to the neural network processing unit.
[0087] The data processing module is also used to perform gamma background rejection based on the discrimination results determined by the neural network processing unit to achieve ultra-low background measurement.
[0088] The convolutional neural network built into the neural network processing unit is used to determine the discrimination results of β and γ based on the pulse characteristic data and the digital signal; the convolutional neural network is obtained by training and optimizing the initial convolutional neural network based on the data set; the data set includes the pulse characteristic data of the known discrimination results and the corresponding digital signal.
[0089] The main control unit is also used to remove the gamma background based on the screening results to achieve ultra-low background measurement.
[0090] As an optional implementation, the acquisition module further includes: a digitally controlled high-voltage module.
[0091] The digitally controlled high-voltage module is connected to the liquid scintillation detector; the digitally controlled high-voltage module is used to provide voltage to the liquid scintillation detector.
[0092] The liquid scintillation detector further includes: a voltage divider tube seat; the voltage divider tube seat is connected to the digital control high voltage module and the high-speed photomultiplier tube respectively; the voltage divider tube seat is used to provide a bias voltage for the high-speed photomultiplier tube according to the voltage.
[0093] In one embodiment, the acquisition module further includes: a temperature, voltage and current monitoring module.
[0094] The temperature, voltage and current monitoring modules are respectively connected to the main control unit, the neural network processing unit and the ultra-high-speed ADC.
[0095] The acquisition module also includes: an air-cooling and heat dissipation module; the air-cooling and heat dissipation module is connected to the main control unit; the air-cooling and heat dissipation module is used to dissipate heat to reduce the device temperature of the main control unit.
[0096] This application uses a liquid scintillation detector, an ultra-high-speed acquisition module (i.e., an acquisition module) and artificial intelligence to distinguish β and γ and achieve ultra-low background measurement.
[0097] like Figure 2 As shown, a liquid scintillation detector primarily consists of a liquid scintillator, a shielded lead chamber, a high-speed PMT (photomultiplier tube), and a voltage divider socket. The liquid scintillator is irradiated by a radioactive source, generating fluorescence. This fluorescence is converted to photoelectricity by the high-speed PMT, which then outputs high-speed current pulses. The liquid scintillation detector uses the high voltage provided by the digitally controlled high-voltage module in the acquisition module to bias each stage of the high-speed PMT via the voltage divider socket, which then outputs high-speed current pulses to the acquisition module.
[0098] The acquisition module mainly includes a high-speed, low-noise current preamplifier (i.e., a high-speed preamplifier), a signal conditioning module, a CNC high-voltage module, a main control unit (FPGA+ARM), an NPU (neural network processing) unit, a high-speed cache and storage, an air-cooled heat dissipation module, and a temperature, voltage, and current monitoring module.
[0099] High-speed preamplifier: The high-speed current pulse output by the liquid scintillation detector is converted into a high-speed voltage pulse signal by the high-speed preamplifier, which is convenient for the back-end ADC to collect. The high-speed preamplifier adopts I / V conversion circuit design, and its operational amplifier is a low bias current, high bandwidth transimpedance amplifier. The circuit diagram of the high-speed preamplifier is shown in Figure 3 .
[0100] Signal conditioning module: It mainly realizes the direct bias adjustment and gain adjustment of high-speed current pulses, and converts them into differential signals that match the ultra-high-speed ADC. The differential amplifier (DIF) subtracts the high-speed voltage pulse signal from the DC voltage controlled by the DAC to realize direct bias adjustment and convert it into a differential signal. The differential signal is passed through the digital differential amplifier (DGA) to achieve gain adjustment before entering the ultra-high-speed ADC. The circuit structure corresponding to the signal conditioning module is shown in Figure 4 .
[0101] Digitally controlled high-voltage module: provides a stable high-voltage power supply for the liquid scintillation detector and is combined with a digital-to-analog converter (ADC, DAC) to achieve feedback regulation of the high-voltage power supply.
[0102] Ultra-high-speed ADC: Use the existing ultra-high-speed ADC chip to achieve analog-to-digital conversion, converting high-speed voltage pulse waveforms into digital signals with a sampling rate of ≥5Gsps.
[0103] like Figure 5 As shown, the main control unit uses an FPGA+ARM architecture. It acquires the high-speed data stream from the ADC and processes the waveform data in real time. The processed waveform data is transmitted to the NPU via PCIE. After calculation, the results are returned and the gamma background is removed in the FPGA. The ARM mainly implements process control and external data transmission. A high-speed serial bus is used for data transmission between the FPGA and the ultra-high-speed ADC. After the ultra-high-speed ADC data enters the FPGA, it is first converted from serial to parallel and then cached. The FPGA reads data from the data pool and performs preprocessing such as baseline subtraction, digital filtering, and amplitude normalization. After preprocessing, the data is calculated for amplitude, rise time, fall time, decay time, rising edge slope, falling edge slope, and the ratio of the trailing edge to the whole peak area. Another channel adjusts the pulse position to align the rising edge midpoints of all pulses and ensure that the number of data points before and after the midpoint of all pulses is consistent. Finally, the pulse feature data, raw waveform data (digital signal), and waveform annotations are combined into a dataset and transmitted to the NPU via the PCIE bus.
[0104] The NPU provides computing power and uses artificial intelligence algorithms to process the high-speed data stream collected by the FPGA. Based on the pulse signature data and digital signals, it determines the β and γ discrimination results and returns them to the main control unit. Regarding the convolutional neural network in the NPU, during model training, a large number of β and γ pulse waveforms are first collected and trained on the convolutional neural network. The β pulse waveform is marked as 1 and the γ pulse waveform is marked as 0. After training, the convolutional neural network records the characteristic parameters.
[0105] This application uses NPU because it has the characteristics of high performance, low power consumption, high computing efficiency, etc., and is more suitable for application in embedded platforms. Figure 6 .
[0106] That is, different NPU platforms have corresponding models. The models need to be optimized before they can be deployed on the resource-limited embedded NPU platform. After model optimization, data conversion, and model compilation, the model can be deployed on the NPU platform. Then, the data set (FPGA's high-speed data stream) is prepared and input into the model for model training. After its performance is tuned and monitored, the model is optimized again and can be used for real-time testing and verification.
[0107] Cache and storage: Mainly provides large-capacity, high-bandwidth data cache for the main control and NPU, and provides data and program storage.
[0108] Temperature, voltage and current monitoring module: monitors the temperature, voltage and current of the acquisition module, provides alarm or alarm shutdown processing, and ensures the safe operation of the acquisition module.
[0109] Temperature monitoring: Since the ultra-high-speed ADC, main control unit, and NPU generate a lot of heat, if the temperature is too high, it will affect the effect and even damage the device, so temperature monitoring is necessary; since the liquid scintillation detector is temperature sensitive, temperature fluctuations will cause differences in the pulse waveform. To eliminate this factor, it is necessary to monitor the temperature. Only when the temperature fluctuates within a certain range can measurement be started. The schematic diagram of temperature monitoring by the temperature, voltage, and current monitoring module is shown in Figure 7 .
[0110] Regarding voltage and current monitoring, such as Figure 8 As shown, the ultra-high-speed ADC, main control, and NPU consume large currents. Current fluctuations can cause voltage fluctuations, affecting circuit stability. Furthermore, abnormal conditions may occur during long-term operation. Therefore, it is necessary to monitor voltage and current to form a protection loop to prevent short circuits from damaging components.
[0111] Air cooling module: controls air cooling according to the temperature of the acquisition module to reduce the temperature of the device.
[0112] The AI algorithm primarily completes high-speed data stream input, waveform preprocessing, convolutional neural network modeling, and event identification output. High-speed data streams are input via PCIE, waveform preprocessing, and then connected to the convolutional neural network model for identification. Event identification results are output based on the discrimination threshold.
[0113] like Figure 9 As shown in the figure, the convolutional neural network model mainly contains the following components: multi-scale convolution, batch normalization, activation function, random inactivation, deep learning optimizer, depth-separable convolution, knowledge distillation, etc.
[0114] Among them, multi-scale convolution is used to capture pulse features; batch normalization is used to make the data at the same level; activation function is used to regularize the neural network and prevent overfitting.
[0115] Deep learning optimizer: can overcome local optimal solutions and has good consistency between training sets and validation sets.
[0116] Depthwise separable convolution: splits the standard convolution to reduce computational complexity; knowledge distillation: transfers knowledge to a lightweight student model. After the initial convolutional neural network is trained, the resulting convolutional neural network model can be used for real-time pulse waveform identification and determination of β and γ.
[0117] This application is aimed at measuring β and eliminating gamma-ray interference. Since the β and gamma pulse waveforms are relatively small and have very high bandwidths, conventional solutions are difficult to eliminate interference. Therefore, this application proposes the use of a liquid scintillation detector, an ultra-high-speed acquisition module, and an artificial intelligence combination to perform β and gamma discrimination, thereby eliminating background and achieving ultra-low background measurement. For β-ray measurement, liquid scintillation detectors have much higher detection efficiency than other detectors, so liquid scintillation is used. β and gamma pulses have very high bandwidths, so high-speed circuits are used in the signal path. Since artificial intelligence can learn nonlinear characteristics at different positions and scales, these characteristics cannot or are difficult to express using formulas, or require the fusion of multiple features to distinguish them. Currently, there are slight differences in decay time and fall time slope, and other features cannot be seen to the naked eye. However, based on previous artificial intelligence research experience, it is found that it has a good ability to learn pulse waveform characteristics, which is better than conventional algorithms. Features such as amplitude, rise time, fall time, decay time, rising edge slope, falling edge slope, and the ratio of the trailing edge to the entire peak area can be expressed using formulas. These characteristic parameters can be input into the artificial intelligence model to allow it to identify new key information and perform β and γ screening, but the effect may not be very good. Therefore, it is proposed to directly input waveform data into the artificial intelligence model (i.e., convolutional neural network, and open source model can also be used) to analyze a large amount of β and γ waveform data, train a set of characteristic parameters, and use a labeling method to automatically identify β and γ pulses. It is possible to identify other features besides the decay time and the fall time slope.
[0118] Only by combining liquid scintillation detectors, ultra-high-speed acquisition modules and artificial intelligence can β and γ be better identified, thereby eliminating the background and achieving ultra-low background measurement.
[0119] Liquid scintillation detector: It has high detection efficiency and sensitivity. When used with a high-speed PMT, it can more effectively transmit particle information. However, detectors using ordinary PMTs are difficult to fully preserve the characteristics of their current pulse waveforms.
[0120] High-bandwidth, low-noise current preamplifier (i.e., high-speed preamplifier): Compared with ordinary current preamplifiers, it can more effectively ensure that the pulse waveform is not distorted. Ordinary current preamplifiers are limited by their own bandwidth, resulting in waveform distortion and loss of waveform characteristics.
[0121] Signal conditioning module: can effectively reduce the size, save power consumption, and enhance system flexibility, which cannot be achieved by using other forms of processors.
[0122] Ultra-high-speed ADC: It can more accurately express the current pulse waveform characteristics, providing a basis for subsequent waveform identification, and retains more complete information than general medium- and high-speed ADCs.
[0123] Artificial Intelligence: The structure of convolutional neural networks enables them to capture and learn pulse signal characteristics. By sliding the convolution kernel over the input data, local features at different locations are extracted. The neural network can learn features at different locations and scales, thus extracting local waveform features. CNNs typically introduce nonlinear activation functions at each layer to enhance the network's nonlinear expression capabilities, enabling the network to learn more complex waveform features, which cannot be achieved with standard waveform features. Artificial intelligence algorithms effectively discriminate between different pulse waveforms, eliminate interference, and achieve ultra-low background measurements.
[0124] In an exemplary embodiment, Figure 10 As shown, an ultra-low background measurement method for distinguishing β and γ is provided, including:
[0125] Step 100: Obtain a high-speed current pulse signal and convert it into a high-speed voltage pulse signal. The high-speed current pulse signal is obtained by photoelectrically converting the fluorescence output by the liquid scintillator through the radiation source using a high-speed photomultiplier tube.
[0126] Step 200: performing linear deflection and gain adjustment on the high-speed voltage pulse signal, and performing analog-to-digital conversion to obtain a digital signal.
[0127] Step 300: Preprocess the digital signal, and perform feature processing and pulse position adjustment on the preprocessed digital signal to obtain pulse feature data.
[0128] The digital signal is preprocessed, and the preprocessed digital signal is subjected to feature processing and pulse position adjustment to obtain pulse feature data, specifically including:
[0129] Perform data stream serial-to-parallel conversion on digital signals, cache the data, and cache it in a data pool.
[0130] The data in the data pool is read and baseline subtraction, digital filtering and amplitude normalization are performed to obtain the preprocessed digital signal.
[0131] The pre-processed digital signal is subjected to feature processing and pulse position adjustment to obtain pulse feature data; the feature processing includes: feature calculation of amplitude, rise time, fall time, decay time, rising edge slope, falling edge slope, and ratio of trailing edge to whole peak area.
[0132] Step 400: Determine the discrimination results for β and γ using a convolutional neural network based on the pulse feature data and the digital signal. The convolutional neural network is obtained by training and optimizing an initial convolutional neural network based on a dataset; the dataset includes the pulse feature data with known discrimination results and the corresponding digital signal.
[0133] Step 500: Perform gamma background removal based on the screening result to achieve ultra-low background measurement.
[0134] In an exemplary embodiment, a computer device is further provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in the above method embodiments when executing the computer program.
[0135] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.
[0136] In an exemplary embodiment, a computer program product is provided, including a computer program. When the computer program is executed by a processor, the steps in the above method embodiments are implemented.
[0137] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.
[0138] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory may include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).
[0139] The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processors involved in the various embodiments provided herein may include, but are not limited to, general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic units, data processing logic units based on quantum computing, and the like.
[0140] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0141] This document uses specific examples to illustrate the principles and implementation methods of this application. The description of the above examples is only intended to help understand the method and core concept of this application. At the same time, for those skilled in the art, based on the concept of this application, there may be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.
Claims
1. An ultra-low background measurement device for distinguishing β and γ, characterized in that: include: Liquid scintillation detector and acquisition module connected in sequence; The liquid scintillation detector includes a liquid scintillator and a high-speed photomultiplier tube; the acquisition module includes a data processing unit, a main control unit and a neural network processing unit connected in sequence; The liquid scintillator is irradiated by a radiation source to output fluorescence; the fluorescence includes: beta rays and gamma rays; The high-speed photomultiplier tube is used to perform photoelectric conversion on fluorescence and output a high-speed current pulse signal; The data processing unit is used for: Converting the high-speed current pulse signal into a high-speed voltage pulse signal; Performing direct bias and gain adjustment on the high-speed voltage pulse signal, and performing analog-to-digital conversion to obtain a digital signal; The main control unit is used to preprocess the digital signal, and perform feature processing and pulse position adjustment on the preprocessed digital signal to obtain pulse feature data; The convolutional neural network built into the neural network processing unit is used to determine the discrimination result regarding β and γ based on the pulse characteristic data and the digital signal; the convolutional neural network is obtained by training and optimizing the initial convolutional neural network based on the data set; the data set includes the pulse characteristic data with known discrimination results and the corresponding digital signal; The main control unit is further configured to perform gamma background rejection according to the identification result to achieve ultra-low background measurement.
2. The ultra-low background measurement device for distinguishing β and γ according to claim 1, characterized in that: The data processing unit specifically includes: a high-speed preamplifier, a signal conditioning module and an ultra-high-speed ADC; The high-speed preamplifier is connected to the liquid scintillation detector and the signal conditioning module respectively; the signal conditioning module is also connected to the ultra-high-speed ADC; The high-speed preamplifier is used to receive the high-speed current pulse signal and convert the high-speed current pulse signal into a high-speed voltage pulse signal; The signal conditioning module is used to: performing direct bias adjustment on the high-speed voltage pulse signal to obtain a differential signal; Performing gain adjustment on the differential signal to obtain a gain signal; The ultra-high-speed ADC is used to perform analog-to-digital conversion on the gain signal to obtain a digital signal.
3. The ultra-low background measurement device for distinguishing β and γ according to claim 1, characterized in that: The main control unit adopts FPGA+ARM architecture.
4. The ultra-low background measurement device for distinguishing β and γ according to claim 1, characterized in that: The main control unit includes: a data acquisition module, a data processing module and a bus control module; The data processing module is connected to the data acquisition module and the neural network processing unit respectively via the bus control module; the data acquisition module is also connected to the data processing unit; the bus control module uses a serial bus for data transmission; The data acquisition module is used to acquire the digital signal; The data processing module is used for: Performing data stream serial-to-parallel conversion on the digital signal, and performing data caching, and caching the data into a data pool; Read the data in the data pool and perform baseline subtraction, digital filtering and amplitude normalization to obtain the pre-processed digital signal; The pre-processed digital signal is subjected to feature processing and pulse position adjustment to obtain pulse feature data; the feature processing includes: feature calculation of amplitude, rise time, fall time, decay time, rising edge slope, falling edge slope, and ratio of trailing edge to whole peak area; transmitting the pulse characteristic data and the digital signal to the neural network processing unit; Gamma background is eliminated according to the discrimination result determined by the neural network processing unit to achieve ultra-low background measurement.
5. The ultra-low background measurement device for distinguishing β and γ according to claim 1, characterized in that: The acquisition module also includes: a digitally controlled high-voltage module; The digitally controlled high-voltage module is connected to the liquid scintillation detector; The digitally controlled high-voltage module is used to provide voltage to the liquid scintillation detector.
6. The ultra-low background measurement device for distinguishing β and γ according to claim 5, characterized in that: The liquid scintillation detector further comprises: a voltage divider tube holder; The voltage divider tube seat is connected to the digital control high voltage module and the high-speed photomultiplier tube respectively; The voltage divider tube seat is used to provide a bias voltage for the high-speed photomultiplier tube according to the voltage.
7. The ultra-low background measurement device for distinguishing β and γ according to claim 2, characterized in that: The acquisition module also includes: a temperature, voltage and current monitoring module; The temperature, voltage and current monitoring modules are respectively connected to the main control unit, the neural network processing unit and the ultra-high-speed ADC.
8. The ultra-low background measurement device for distinguishing β and γ according to claim 1, characterized in that: The acquisition module also includes: an air cooling module; The air cooling module is connected to the main control unit; The air-cooling heat dissipation module is used for heat dissipation to reduce the device temperature of the main control unit.
9. An ultra-low background measurement method for distinguishing β and γ, characterized in that: include: Obtain high-speed current pulse signals and convert them into high-speed voltage pulse signals; The high-speed current pulse signal is obtained by photoelectric conversion of the fluorescence output by the liquid scintillator through the radiation source based on the high-speed photomultiplier tube; Performing direct bias and gain adjustment on the high-speed voltage pulse signal, and performing analog-to-digital conversion to obtain a digital signal; Preprocessing the digital signal, and performing feature processing and pulse position adjustment on the preprocessed digital signal to obtain pulse feature data; A convolutional neural network is used to determine the discrimination result of β and γ based on the pulse characteristic data and the digital signal; the convolutional neural network is obtained by training and optimizing an initial convolutional neural network based on a data set; the data set includes the pulse characteristic data with known discrimination results and the corresponding digital signal; The gamma background is eliminated according to the screening results to achieve ultra-low background measurement.
10. The ultra-low background measurement method for distinguishing β and γ according to claim 9, characterized in that: Preprocessing the digital signal, and performing feature processing and pulse position adjustment on the preprocessed digital signal to obtain pulse feature data, specifically including: Performing data stream serial-to-parallel conversion on the digital signal, and performing data caching, and caching the data into a data pool; Read the data in the data pool and perform baseline subtraction, digital filtering and amplitude normalization to obtain the pre-processed digital signal; The pre-processed digital signal is subjected to feature processing and pulse position adjustment to obtain pulse feature data; the feature processing includes: feature calculation of amplitude, rise time, fall time, decay time, rising edge slope, falling edge slope, and ratio of trailing edge to whole peak area.