Nuclide identification method, system and electronic device

Abstract

A nuclide identification method, a system and an electronic device are provided, which relates to the field of nuclide identification. According to statistical characteristics, a background probability density function and a Compton probability density function in a ROI are determined. An energy Bayesian factor and a time Bayesian factor are determined based on energy and time interval information in a sequence of the nuclear detection events obtained by measurement. By combining the two factors, Compton plateau can be effectively identified and distinguished.

Description

CROSS-REFERENCE TO RELATED APPLICATION

In accordance with 37 C.F.R. 1.76, a claim of priority is included in an Application Data Sheet filed concurrently herewith. Accordingly, the present invention claims priority to Chinese Patent Application No. 202410107288.6, entitled “NUCLIDE IDENTIFICATION METHOD, SYSTEM AND ELECTRONIC DEVICE”, filed Jan. 25, 2024. The contents of the above referenced application is incorporated herein by reference in its entirety.

FIELD OF THE INVENTION

The present disclosure relates to the field of nuclide identification, in particular to a nuclide identification method, a system and an electronic device.

BACKGROUND OF THE INVENTION

At present, the common traditional nuclide identification methods are based on γ-ray spectrum decomposition analysis technology and characteristic full energy peak matching technology. These two technologies mainly include background subtraction, filtering smoothing and peak searching, and both statistically analyze the characteristic γ-ray emitted by radioactive materials by assuming the Gaussian distribution of the peak shape of the full energy peak and by matching the peak position, thus realizing the qualitative and quantitative determination of radioactive materials. However, these methods need to collect enough photons to reduce the statistical fluctuation of characteristic peaks, so that there are certain requirements for detection time and characteristic γ-ray emission intensity. However, in special security inspection scenes such as ports, airports and borders, there is a large number of people and goods, and the detection target can be measured only for a short time (in seconds). At the same time, the detection target is in motion relative to the detector during the security inspection, which will further affect the number of photons collected by the detector. In addition, radioactive materials are usually placed in well-shielded tanks or containers in these scenes, which causes the distortion of γ-spectrum and reduction in counting rate, which in turn affects the rapid and accurate identification of nuclides by additional influence from environmental background and Compton scattering, and further increases the difficulty of processing γ radiation signals. To sum up, in the special security inspection scenes such as ports, airports and borders, the peak searching-nuclide matching identification method based on a Gaussian model becomes difficult and inaccurate in the process of analyzing γ-spectrum, and the prominent contradiction is reflected in the contradiction between identification speed and accuracy. Although the use of fixed large-volume detectors can alleviate this contradiction to some extent, its high price and large size are the main factors limiting its widespread application.

In addition, at present, the new nuclide identification methods based on full-spectrum analysis mainly include a sequential Bayesian method, a fuzzy mathematical method, a neural network method and a deconvolution method. These methods not only use characteristic full energy peaks as useful information, but also incorporate information such as a branching ratio, a half-life period and an γ spectrum shape into the analysis information to reduce the uncertainty of nuclide identification and improve the detection efficiency under the condition of a low counting rate. The sequential Bayesian nuclide identification method has the advantages of a high identification lower limit and a fast identification speed. The fuzzy mathematical nuclide identification method can still perform accurate identification in a complex environment. The neural network nuclide identification method can simulate any function, and can use the full-spectrum information of radionuclides for analysis, and can quickly and accurately identify nuclides without complicated operations such as smoothing, peak searching and peak fitting.

In view of the characteristics that measurement is non-static, detection time is short and the detected signal is weak in the detection of illegal transportation of radioactive materials in special scenes such as airports, ports and nuclear radiation emergency, new requirements are put forward in measurement technology.

• • 1. The measurable time is short, usually in the range of sub-seconds to more than ten seconds.
  • 2. The obtained information is rich, particularly it is necessary to determine not only the energy of γ rays, but also the information such as the types, intensities and dose of corresponding radionuclides.
  • 3. The reliability of data is high. At a high level of confidence, the requirement of identification efficiency and accuracy is high, and the false alarm rate and the missed alarm rate should be as low as possible.

Traditionally, the characteristic full energy peak matching technology of γ-spectrum is generally used to analyze indicators such as radionuclides and dose. This analysis method usually needs to collect enough photons for identification, that is, it needs to be based on a large number of γ ray measurement cases and long-term measurement to reduce statistical fluctuation and ensure the necessary measurement representativeness and accuracy. In the radiation detection process recommended by the International Atomic Energy Agency (IAEA), there are mainly three categories based on this method: a fixed entrance radiation monitor system, a handheld/portable γ spectrometer and a laboratory high-resolution γ spectrometer. The newly developed fixed large-volume sodium iodide (NaI(Tl)) detector and high-purity germanium (HPGe) detector can accurately identify the types of radionuclides, but the system is very expensive, poor in portability and unable to be deployed quickly. The portable γ-ray spectrometer mainly includes a scintillator γ-ray spectrometer and a semiconductor γ-ray spectrometer. Such detector usually has low detection efficiency and needs long-term measurement to collect enough photons to reduce the statistical fluctuation of counting. The laboratory high-resolution γ spectrometer is mainly a HPGe γ spectrometer cooled by liquid nitrogen. The system is poor in universality, unable to be deployed quickly and is expensive. To sum up, it is difficult to realize the rapid measurement and identification of low-level nuclides in a short time by using the traditional method, that is, a nuclide identification method based on retrieval from the nuclide database and matching of characteristic full energy peaks in the γ-spectrum. On the one hand, such method cannot meet the identification requirements of various nuclides in the complex environment. On the other hand, it is difficult to realize rapid nuclide identification because of the long forming time of characteristic full energy peaks at the peak position in the γ-spectrum. At the same time, the traditional nuclide identification method based on γ-spectrum decomposition analysis and characteristic peak matching is prone to missed alarm events when the net counting rate of full energy peaks is not enough.

In addition, at present, the new nuclide identification methods based on full spectrum analysis mainly include a sequential Bayesian method based on Bayesian theory and sequential probability ratio test (SPRT), a fuzzy mathematical method, a neural network method and a deconvolution method, etc. When adapting to different γ-spectrometers, these methods will have some shortcomings in the universality in certain aspects such as the degree of influence from background and Compton plateau, the false alarm rate, the missed alarm rate and the amount of calculation required. The fuzzy mathematical method, the neural network method and the deconvolution method are based on full spectrum analysis, which require a large number of particles and long measurement time. At the same time, the neural network method and the deconvolution method need a large amount of calculation, so that they are not suitable for real-time online analysis.

The sequential Bayesian nuclide identification method based on the Bayesian theory and SPRT uses three characteristics of the half-life period of the radionuclides, the characteristic γ-ray energy and the branch ratio, selects the appropriate prior function and confidence, updates the decision function by using SPRT, and makes statistical inference on the hypothesis test. However, this method needs to preset some parameters that are related to the samples and should be unknown in the actual measurement process, which will significantly limit the universality of this method. For example, due to preset the time interval parameter term in the test model, this method will have a high false alarm rate at Compton plateau generated by high-energy rays in the low-energy region of interest. In addition, this method uses a Gaussian model for the distribution of background, which does not conform to the actual situation. This leads to some limitations in the application of the existing sequential Bayesian method.

SUMMARY OF THE INVENTION

In order to solve the above problems in the traditional technology, the present disclosure provides a nuclide identification method, a system and an electronic device.

In order to achieve the above objectives, the present disclosure provides the following solutions.

A nuclide identification method is provided, including:

• • acquiring time-energy information of a ray, and determining whether the ray belongs to one of Region of Interests (ROIs) based on energy information in the time-energy information of the ray;
  • describing a sequence of all nuclear detection events in the ROI when the ray belongs to the one of the ROIs;
  • determining an energy Bayesian factor based on energy information in the described sequence of all the nuclear detection events;
  • determining an energy decision function based on the energy Bayesian factor;
  • determining a time interval in the ROI based on time information in the described sequence of all the nuclear detection events;
  • determining a time Bayesian factor based on the time interval;
  • determining a time decision function based on the time Bayesian factor;
  • combining the energy decision function and the time decision function to obtain a joint decision function of the ROI;
  • retrieving a potential nuclide corresponding to the ROI based on a characteristic γ ray corresponding to the ROT, and combining joint decision functions of respective ROIs corresponding to the retrieved nuclide to obtain a nuclide joint decision function; and
  • determining and identifying the retrieved nuclide based on the nuclide joint decision function.

According to the specific embodiment of the present disclosure, the present disclosure provides the following technical effects.

Based on a Bayesian factor and a sequential posterior probability, the present disclosure significantly improves the universality of the method by setting a range of a time interval on a decision function instead of directly giving a predefined time interval parameter. According to statistical characteristics, a background probability density function and a Compton probability density function in a ROI are determined. An energy Bayesian factor and a time Bayesian factor are determined based on energy and time interval information in a sequence of the nuclear detection events obtained by measurement. By combining the two factors, the sequence of the nuclear detection events can effectively identify and distinguish Compton plateau. Under the same identification conditions and the same confidence level, the nuclide identification method according to the present disclosure can effectively identify the existence and types of radionuclides faster than an energy spectrum decomposition analysis-characteristic peak matching method. Compared with the fuzzy mathematical and neural network nuclide identification methods, the nuclide identification method provided by the present disclosure is more universal.

In addition, compared with the existing sequential nuclide identification method based on the Bayesian theory and SPRT, the nuclide identification method provided by the present disclosure can significantly reduce the false alarm rate and the missed alarm rate, and can effectively identify Compton contribution from high-energy rays appearing in the low-energy ROI.

Further, the present disclosure provides a nuclide identification system, wherein the system is used to implement the nuclide identification method described above; wherein the system includes:

• • a ray belonging region determining module, configured to acquire time-energy information of a ray, and determine whether the ray belongs to one of Region of Interests (ROIs) based on energy information in the time-energy information of the ray;
  • a nuclear detection event sequence description module, configured to describe a sequence of all nuclear detection events in the ROI when the ray belongs to the one of the ROIs;
  • an energy Bayesian factor determining module, configured to determine an energy Bayesian factor based on energy information in the described sequence of all the nuclear detection events;
  • an energy decision function determining module, configured to determine an energy decision function based on the energy Bayesian factor;
  • an ROI time interval determining module, configured to determine a time interval in the ROI based on time information in the described sequence of all the nuclear detection events;
  • a time Bayesian factor determining module, configured to determine a time Bayesian factor based on the time interval;
  • a time decision function determining module, configured to determine a time decision function based on the time Bayesian factor;
  • a decision function combining module, configured to combine the energy decision function and the time decision function to obtain a joint decision function of the ROI;
  • a nuclide joint decision function determining module, configured to retrieve a potential nuclide corresponding to the ROI based on a characteristic γ-ray corresponding to the ROI, and combine joint decision functions of respective ROIs corresponding to the retrieved nuclide to obtain a nuclide joint decision function; and
  • a nuclide identification module, configured to determine and identify the retrieved nuclide based on the nuclide joint decision function.

Still further, the present disclosure further provides an electronic device, including:

• • a memory in which a computer program is stored; and
  • a processor connected with the memory, which is configured to call and execute the computer program to implement the nuclide identification method described above.

Since the technical effects achieved by the system and the electronic device according to the present disclosure are the same as those achieved by the nuclide identification method provided above, the technical effects will not be described in detail herein.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to illustrate embodiments of the present disclosure or the technical solutions in the prior art more clearly, the drawings used in the embodiments will be briefly described below. Apparently, the drawings described below are only some embodiments of the present disclosure. For those skilled in the art, other drawings can be obtained according to these drawings without creative efforts.

FIG. 1 is an overall flow chart of a nuclide identification method according to the present disclosure.

FIGS. 2A and 2B together form an implementation flow chart of the nuclide identification method according to the present disclosure.

FIG. 3 is a curve diagram illustrating variations of time and energy decision functions of a first 59 keV γ ray according to the present disclosure.

FIG. 4 is a curve diagram illustrating variations of time and energy decision functions of a first 661 keV γ ray according to the present disclosure.

FIG. 5 is a curve diagram illustrating a variation of a first nuclide joint decision function of ²⁴¹Am according to the present disclosure.

FIG. 6 is a curve diagram illustrating a variation of a first nuclide joint decision function of ¹³⁷Cs according to the present disclosure.

FIG. 7 is a curve diagram illustrating variations of time and energy decision functions of a second 59 keV γ ray according to the present disclosure.

FIG. 8 is a curve diagram illustrating variations of time and energy decision functions of a second 661 keV γ ray according to the present disclosure.

FIG. 9 is a curve diagram illustrating a variation of a second nuclide joint decision function of ²⁴¹Am according to the present disclosure.

FIG. 10 is a curve diagram illustrating a variation of a second nuclide joint decision function of ¹³⁷Cs according to the present disclosure.

DETAILED DESCRIPTION OF THE INVENTION

The technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Apparently, the described embodiments are merely a part of the embodiments of the present disclosure, rather than all of the embodiments. All other embodiments obtained by those skilled in the art based on the embodiment of the present disclosure without creative efforts shall fall within the scope of the present disclosure.

The present disclosure aims to provide a nuclide identification method, a system and an electronic device, which can well distinguish Compton plateau, effectively identify the existence and types of radionuclides faster and have more universality.

In order to make the above objectives, features and advantages of the present disclosure clearer and more comprehensible, the present disclosure will be described in further detail below in conjunction with the drawings and specific implementations.

When γ rays are incident on the sensitive volume of a γ spectrometer measurement system, the measurement system outputs the energy and time data pair (ε, t) of the rays to an upper computer. The upper computer calculates the decision function and completes statistical inference based on the energy and time data pair (ε, t) of the incident rays according to the nuclide rapid identification algorithm, and makes effective inference on the existence and types of radionuclides. Based on this, as shown in FIGS. 1 and 2A-2B, the nuclide identification method according to the present disclosure includes the following steps.

Step 1: Region of Interest (ROI) Identification

When a detector detects a ray, time-energy information ξ⁽⁰⁾=(ε⁽⁰⁾,τ⁽⁰⁾) of the ray is outputted, and it is determined whether the ray belongs to a certain ROI according to the energy information ε⁽⁰⁾. If not, the ray is discarded. If the ray belongs to a certain ROI region, such as ROI_i region, a sequence {right arrow over (ξ)}_i of all nuclear detection events in this ROI_i can be described as follows:{right arrow over (ξ)}_i={ξ_i,j,ξ_i,j−1, . . . }ξ_i,j=(ε_i,j,τ_i,j)=ξ⁽⁰⁾ ε_i,j=ε⁽⁰⁾,τ_i,j=τ⁽⁰⁾

where denotes ξ_i,j the time and energy information of the j-th ray in the ROI_i, where ROI_i represents the i-th region of interest, which is expressed as:ROI_i=(E_i−k·σ_i,E_i+k·σ_i)

where E_i denotes the energy of the characteristic γ ray corresponding to the i-th ROI, k is a coverage factor, and σ_i is a standard deviation of the full energy peak of the characteristic γ ray corresponding to the i-th ROI.

Step 2: Calculation of the Energy Bayesian Factor

$B_i^{\pi }\left({\epsilon }_{i,j}\right)=\frac{f_{i,0}\left({\epsilon }_{i,j}❘{\sigma }_{i,0}^2\right)}{\int g_{i,1}\left({\sigma }_{i,1}^2\right)f_{i,1}\left({\epsilon }_{i,j}❘{\sigma }_{i,1}^2\right)d\left({\sigma }_{i,1}^2\right)}$

where B_i^π(ε_i,j) is the j-th energy Bayesian factor of the ROI_i calculated according to the energy information of the current ray, f_i,0(ε_i,j|σ_i,0²) and f_i,1(ε_i,j|σ_i,1²) are the energy probability density functions under the energy null hypothesis H₀ (there is no full energy peak) and the energy alternative hypothesis H₁ (there is full energy peak), respectively, σ_i,0² and σ_i,1² are the standard variances of the energy under H₀ and H₁, respectively, and g_i,1(σ_i,1²) is the prior probability under H₁. Moreover:

$\{\begin{array}{c}{H}_0:f_{i,0}\left({\epsilon }_{i,j}❘{\sigma }_{i,0}^2\right)=k_i·\left({\epsilon }_{i,j}-E_i\right)+1/{\Delta }_{E_i},{\sigma }_{i,0}^2={\Delta }_{E_i}^2/12\\ H_1:f_{i,1}\left({\epsilon }_{i,j}❘{\sigma }_{i,1}^2\right)=p\frac{A}{\sqrt{2·\pi ·{\sigma }_{i,1}^2}}·e^{\frac{{\left({\epsilon }_{i,j}-E_i\right)}^2}{2·{\sigma }_{i,1}^2}}+\left(1-p\right)·f_{i,0}\left({\epsilon }_{i,j}❘{\sigma }_{i,0}^2\right),{\Delta }_{E_i}^2/C_k<{\sigma }_{i,1}^2<{\Delta }_{E_i}^2/12\end{array}$ ${\pi }_i\left({\sigma }^2\right)=\{\begin{array}{cc}{\pi }_{i,0},& {\sigma }^2={\Delta }_{E_i}^2/12\\ {\pi }_{i,1}{g}_{i,1}\left({\sigma }^2\right),& {\Delta }_{E_i}^2/C_k<{\sigma }^2<{\Delta }_{E_i}^2/12\end{array}$

where Δ_Ei=2·k·σ_i,1 is an energy width of the ROI_i, π_i(σ²) is an energy prior probability density function of the ROI_i, and π_i,0 and π_i,1 are the energy prior probability under H₀ and H₁, respectively.

Step 3: Calculation of the Energy Decision Function

$p_{i,0}^{\pi }={\left[1+\frac{1-{\pi }_{i,0}}{{\pi }_{i,0}}\frac{1}{B_i^{\pi }\left({\epsilon }_{i,j}\right)}\right]}^{-1};$ $p_{i,1}^{\pi }=1-p_{i,0}^{\pi };$

where p_i,0^π and p_i,1^π are the energy posterior probability of the ROI_i calculated according to energy of the current ray under H₀ and H₁, respectively, which are also the decision functions of γ-ray decision making.

Step 4: Updating of the Energy Prior Probabilityπ_i,0=p_i,0^π;π_i,1=p_i,1^π;Step 5: Calculation of the Time Interval in the ROI_i

The sequence {right arrow over (ξ)}_i of the nuclear detection events in the ROI_i in Step 1 is used to calculate the time interval in the ROI_i, which is:Δt_i,j=τ_i,j−τ_i,j−1

where τ_i,j is the measurement moment of the current ray, and τ_i,j−1 is the measurement moment of the previous ray in the ROI_i.

Step 6: Calculation of the Time Bayesian Factor in the ROI_i

$B_i^{\vartheta }\left(\Delta t_{i,j}\right)=\frac{g_{i,0}\left(\Delta t_{i,j}❘{\tau }_{i,0}\right)}{\int h_{i,1}\left({\tau }_{i,1}\right)·g_{i,1}\left(\Delta t_{i,j}❘{\tau }_{i,1}\right)d{\tau }_{i,1}}$

where B_Δt^ϑ(Δt_i,j) is a time Bayesian factor, g_i,0(Δt_i,j|τ_i,0) is a time interval probability density function under the time null hypothesis M₀ (there are no radionuclides), g_i,1(Δt_i,j|τ_i,1) is a time interval probability density function under the time alternative hypothesis M₁ (there are radionuclides), τ_i,0 is the mathematic expectation of the time interval under the time null hypothesis M₀, τ_i,1 is a mathematic expectation of the time interval under the time alternative hypothesis M₁, Δt_i,j is a time interval, and h_i,1(τ_i,1) is the probability density function under the time alternative hypothesis M₁. Moreover:

$\{\begin{array}{c}{M}_0:g_0\left(\Delta t❘{\tau }_0\right)=\frac{1}{{\tau }_0}·e^{-\frac{\Delta t}{{\tau }_0}},{\tau }_0={\tau }_{\mathrm{bkg}}=\frac{1}{{\stackrel{.}{n}}_{\mathrm{bkg}}}\\ M_1:g_1\left(\Delta t❘{\tau }_1\right)=\frac{1}{{\tau }_1}·e^{-\frac{\Delta t}{{\tau }_1}},{\tau }_1\in \left({\tau }_{\min },{\tau }_{\mathrm{bkg}}\right)\end{array}$ $\vartheta \left(\tau \right)=\{\begin{array}{cc}{\vartheta }_0& ,\tau ={\tau }_{\mathrm{bkg}}\\ {\vartheta }_1·h_1\left(\tau \right)& ,\tau \in \left({\tau }_{\min },{\tau }_{\mathrm{bkg}}\right)\end{array}$

where g₀(Δt_i,j|τ₀) and g₁(Δt_i,j|τ₁) are the time interval probability density function under the time null hypothesis M₀ (there are no radionuclides) and the time alternative hypothesis M₁ (there are radionuclides), respectively, τ₀ and τ₁ the mathematic expectations of the time interval under M₀ and M₁, respectively, h₁(τ₁) is a probability density function under M₁,

$h_1\left(\tau \right)\propto \frac{1}{\tau },{\dot{n}}_{\mathrm{bkg}}$ and τ_bkg are the count rate and the mean of the time interval under the background condition, respectively, which are reciprocal to each other, and τ_min is a parameter determining a lower limit η_min of detection sensitivity under M₁. The lower limit η_min of detection sensitivity is defined as the minimum signal-to-noise ratio η that can make effective determination under active conditions and meet certain detection performance requirements, and the signal-to-noise ratio is the ratio of the full-spectrum net count rate to the background counting rate. ϑ(τ) is the prior probability density function of the full-spectrum time interval. ϑ₀ and ϑ₁ are the prior probabilities under M₀ and M₁, respectively, in which ϑ₀+ϑ₁=1.Step 7: Calculation of the Time Decision Function

$p_{i,0}^{\vartheta }={\left[1+\frac{1-{\vartheta }_{i,0}}{{\vartheta }_{i,0}}\frac{1}{B_i^{\vartheta }\left(\Delta t_{i,j}\right)}\right]}^{-1}$ $p_{i,j}^{\vartheta }=1-p_{i,0}^{\vartheta };$

where p_i,0^ϑ and p_i,1^ϑ are the time posterior probabilities calculated according to the time interval of the current ray under M₀ and M₁, respectively, which are also the decision function of γ-ray decision making.

Step 8: Updating of the Time Prior Probabilityϑ_i,0=p_i,0^ϑ;ϑ_i,1=p_i,1^ϑ;Step 9: γ-Ray Decision Making

The decision functions of time and energy in the ROI_i are combined. To sum up, the decision on the existence of γ rays is made, specifically:

$\{\begin{array}{c}\begin{array}{cc}\mathrm{Action}1:& \mathrm{if}{P}_{i,0}^{\pi }<P_{i,\mathrm{low}\_\mathrm{th}}^{\pi }\&P_{i,0}^{\vartheta }<P_{i,\mathrm{low}\_\mathrm{th}}^{\vartheta },\\ & \mathrm{the}\mathrm{signal}\mathrm{is}\mathrm{judged}\mathrm{as}a\mathrm{chacteristic}\gamma -\mathrm{ray}\mathrm{signal}\end{array}\\ \begin{array}{cc}\mathrm{Action}2:& \mathrm{if}{P}_{i,0}^{\pi }>P_{i,\mathrm{up}\_\mathrm{th}}^{\pi }\&P_{i,0}^{\vartheta }>P_{i,\mathrm{up}\_\mathrm{th}}^{\vartheta },\\ & \mathrm{the}\mathrm{signal}\mathrm{is}\mathrm{judged}\mathrm{is}a\mathrm{background}\mathrm{signal}\end{array}\\ \begin{array}{cc}\mathrm{Action}3:& \mathrm{if}{P}_{i,0}^{\pi }<P_{i,\mathrm{low}\_\mathrm{th}}^{\pi }\&P_{i,0}^{\vartheta }>P_{i,\mathrm{up}\_\mathrm{th}}^{\vartheta },\\ & \mathrm{the}\mathrm{signal}\mathrm{is}\mathrm{judged}\mathrm{as}a\mathrm{natural}\mathrm{background}\mathrm{radioactive}\mathrm{ray}\mathrm{signal}\end{array}\\ \begin{array}{cc}\mathrm{Action}4:& \mathrm{if}{P}_{i,0}^{\pi }>P_{t,\mathrm{up}\_\mathrm{th}}^{\pi }\&P_{t,0}^{\vartheta }<P_{t,\mathrm{low}\_\mathrm{th}}^{\vartheta },\\ & \mathrm{the}\mathrm{signal}\mathrm{is}\mathrm{judged}\mathrm{as}a\mathrm{Compton}\mathrm{signal}\mathrm{of}\mathrm{other}\mathrm{high}‐\mathrm{energy}\mathrm{rays}\mathrm{in}\mathrm{this}{\mathrm{ROI}}_i\mathrm{interval}\end{array}\\ \begin{array}{cc}\mathrm{Action}5:& \mathrm{in}\mathrm{other}\mathrm{cases},\\ & \mathrm{no}\mathrm{judgement}\mathrm{is}\mathrm{made},\mathrm{and}\mathrm{continue}\mathrm{to}\mathrm{detect}\end{array}\end{array}$

where P_{i,up_th}^π and P_{i,low_th}^π are the upper and lower thresholds of the energy decision function, respectively, P_{i,up_th}^ϑ and P_{i,low_th}^ϑ are the upper and lower thresholds of the time decision function, respectively.

$\begin{array}{c}{p}_{i,\mathrm{up}\_\mathrm{th}}^{\pi }=\frac{1-{\alpha }^{\pi }}{1-{\alpha }^{\pi }+{\beta }^{\pi }}\\ p_{i,\mathrm{low}\_\mathrm{th}}^{\pi }=\frac{{\alpha }^{\pi }}{1-{\beta }^{\pi }+{\alpha }^{\pi }}\\ p_{i,\mathrm{up}\_\mathrm{th}}^{\vartheta }=\frac{1-{\alpha }^{\vartheta }}{1-{\alpha }^{\vartheta }+{\beta }^{\vartheta }}\\ p_{i,\mathrm{low}\_\mathrm{th}}^{\vartheta }=\frac{{\alpha }^{\vartheta }}{1-{\beta }^{\vartheta }+{\alpha }^{\vartheta }}\end{array}$

where α^π and β^π are the probabilities of making a first type of mistakes and a second type of mistakes in the energy threshold decision making, respectively, and α^ϑ and β^ϑ are the probabilities of making the first type of mistakes and the second type of mistakes in the time threshold decision making, respectively.

Step 10: Calculation of the Joint Decision Function of Radionuclide I_x

$\begin{array}{c}{P}_X=\sum _m{\lambda }_m·P_{m,0}^{\pi }·P_{m,0}^{\vartheta }\\ {\lambda }_m=\frac{{\eta }_m·{\kappa }_m}{\sum _N{\eta }_N·{\kappa }_N}\end{array}$

where P_X denotes the decision function of nuclide I_X, m denotes the serial number of the characteristic γ-ray of nuclide I_X, λ_m denotes the weight coefficient of the corresponding γ-ray, p_m,0^π and p_m,0^ϑ denote the decision functions of energy and time corresponding to the ray, respectively, η_m and κ_m denote the branching ratio and the intrinsic detection efficiency of γ-ray, respectively, and N is the total number of γ-rays of radionuclide I_X.

Step 11: Radionuclide Decision Making

To sum up, the decision is made as follows:

$\{\begin{array}{c}\mathrm{If},P_X\le P_{X,\mathrm{low}\_\mathrm{th}},\mathrm{it}\mathrm{is}\mathrm{judged}\mathrm{that}\mathrm{there}\mathrm{is}\mathrm{nuclide}{I}_x\\ P_{X,\mathrm{up}\_\mathrm{th}}>P_X>P_{X,\mathrm{low}\_\mathrm{th}},\mathrm{no}\mathrm{judgment}\mathrm{is}\mathrm{made},\mathrm{and}\mathrm{more}\mathrm{information}\mathrm{is}\mathrm{needed}\\ P_{X,\mathrm{up}\_\mathrm{th}}\le P_X,\mathrm{it}\mathrm{is}\mathrm{judged}\mathrm{that}\mathrm{there}\mathrm{is}\mathrm{no}\mathrm{nuclid}e{I}_x\end{array}$

where P_{X,up_th} and P_{X,low_th} are the upper threshold and the lower threshold, respectively. The lower threshold P_{X,low_th} is a first predefined value, and the upper threshold P_{X,up_th} is a second predefined value, where the second predefined value is greater than the first predefined value.

$P_{X,\mathrm{low}\_\mathrm{th}}=\frac{\alpha }{1-\beta +a},P_{X,\mathrm{up}\_\mathrm{th}}=\frac{1-\alpha }{1-\alpha +\beta },$ where α and β are the probabilities of making a first type of mistakes and a second type of mistakes in the nuclide joint decision making, respectively. The specific values of α and β need to be determined based on the practical environment, purpose, objectives, and monitoring requirements (confidence level 1−α). Usually, α means “false alarm”, and α usually is set as 5%, 10%, etc. For scenarios with high monitoring requirements, α can be set as 1%. β means “missed alarm”, which is a more concerned indicator. Generally, β is set as 5%, 10%, etc. For scenarios with high monitoring requirements, β can be set as 1%.

Based on the above description, compared with the prior art, the present disclosure has the following characteristics.

• • 1) The construction of the probability density functions of the null hypothesis and the alternative hypothesis is one of the key points of the present disclosure. In the present disclosure, the hypothesis about whether there are radionuclides is converted into whether there are full energy peaks in the ROI based on the single-energy decomposition theory and a certain transformation.
  • 2) The construction method of the probability density function of energy-related statistics is one of the key points of the present disclosure. The probability density function of the energy in the actual process is actually a linear superposition of Gaussian distribution, uniform distribution and triangular distribution according to their respective counting rate shares and is not limited to Gaussian distribution and uniform distribution in the example of the present disclosure. The specific probability density distribution can be determined according to the actual use scene.
  • 3) The construction of prior probabilities of energy-related statistics and time-related statistics is one of the key points of the present disclosure. In the present disclosure, the prior probability density function can be a non-informative prior probability density function or a conjugate prior probability density function, and is not limited to the noninformative prior probability density function in the present disclosure. At the same time, the initial value of the prior probabilities can be determined according to the actual situation, including but not limited to the parameters (0.5, 0.5) in the example of the present disclosure.
  • 4) The construction of the decision function of energy-related statistics and time-related statistics is one of the key points of the present disclosure. In the present disclosure, the decision function can be constructed using two paths: a) the decision function is obtained according to the prior probability and the sample probability density function; b) in the construction process, the Bayesian factor can first be calculated and obtained according to the sample probability density function, and then the decision function is jointly constructed according to the prior probability and the Bayesian factor. No matter which path it is, the decision function structure constructed according to the above key information is the key point to be protected in the present disclosure. [The Bayesian factor, as an intermediate parameter, may not participate in the construction process in the actual construction process.]
  • 5) The solution of the decision function is one of the key points of the present disclosure, no matter whether it is calculated according to the prior probability and the Bayesian factor or it is directly calculated according to the prior information and the sample information. In the process of calculation based on the prior probability and the Bayesian factor, the Bayesian factor, as an intermediate parameter, may not participate in the calculation process in actual use.
  • 6) The construction method of threshold determination at the end of statistical inference of energy-related statistics and time-related statistics is one of the key points of the present disclosure. In the conventional Bayesian method, the relative size of the posterior probabilities of the null hypothesis and the alternative hypothesis (i.e. the upper and lower thresholds are both 0.5) is generally selected, that is, the probability ratio test (SPRT). When making statistical decisions, in the case of the posterior probabilities of the null hypothesis and the alternative hypothesis are close to each other, a wrong decision may be made with a great probability. In the present disclosure, the upper and lower thresholds are set to different values for the decision functions of the energy and the time interval (for example, the decision thresholds for the ROI are 0.8 and 0.2 in this example, which does not mean that the specific values are the protection points of the present disclosure), thus significantly improving the accuracy of the decision results. At the same time, the introduction of decision thresholds for two types of decisions of the decision function (supporting the null hypothesis and supporting the alternative hypothesis) is the protection point of the present disclosure, and the selection of specific values of the thresholds can be made according to actual use scenes.
  • 7) It is one of the key points of the present disclosure to use the combined discrimination result of energy-related statistics and time-related statistics of the detected γ ray as the discrimination result of this γ ray. In combination with the content in 6), although there are ways that identify nuclides by combining time and energy in the prior art, the probability density function and the decision function used in these ways are different from those used in the present disclosure.
  • 8) It is one of the key points of the present disclosure to construct the decision function of nuclides according to the product of all γ-ray decision functions of radionuclides and the constructed weight coefficient. The same or similar methods are not disclosed in the prior art.
  • 9) In the statistical decision of the decision function, the upper and lower thresholds are set to different values, thereby significantly improving the accuracy of decision results. At the same time, the introduction of decision thresholds for two types of decisions of the decision function (supporting the null hypothesis and supporting the alternative hypothesis) is one of the key points of the present disclosure, and the selection of specific values of the thresholds can be made according to the actual use scenes.
  • 10) In the construction of the weight coefficient of the decision function, the normalized result of the product of the branching ratio of the ray and the intrinsic detection efficiency in the corresponding nuclide is taken as the weight coefficient in the nuclide.

Taking background, ²⁴¹Am, ¹³⁷Cs as examples, the specific implementation process of the nuclide identification method according to the present disclosure will be described hereinafter. The lanthanum bromide (LaBr₃) γ spectrometer is used as an example, and any γ spectrometer can be selected in the actual operation process.

Embodiment 1

In this example, a 1.5-inch LaBr₃(Ce) detection system is used, and the energy scale and the half-width scale are:E=0.07525·ch+0.9285(R²=1)FWHM=2.75077+0.56398·√{square root over (E+1.57004·E²)}

A nuclide library containing 22 nuclides and 46 characteristic gamma rays is constructed. Accordingly, 46 ROIs are delineated in the whole spectrum. The ROI is taken as full width at tenth maximum (FWTM). The coverage factor is k=2.146, the confidence is about 96.8%, and the energy width of the ROI_i is ΔE_i=4.296*σ_i. The background data is acquired, the background counting rate of the whole spectrum of the scale is {dot over (n)}_bkg≈59.1 s⁻¹, and the background counting rate in 46 ROIs is {dot over (n)}_i,bkg.

Due to the unbiasedness of the prior probability, the initial value of the prior probability is set to (0.5, 0.5). The lower limit of detection sensitivity of this method in the ROI_i is set as 30%, in which τ_i,min=50%·τ_i,bkg, α^π=β^π=0.2 and α^ϑ=β^ϑ=0.2 are set. The upper and lower thresholds of making decisions are P_{i,up_th}^π=P_{i,up_th}^ϑ=0.8, P_{i,low_th}^π=P_{i,low_th}^ϑ=0.2, P_{X,up_th}=0.64, and P_{X,low_th}=0.04.

Based on this, the background is identified.

The detector has detected the first ray, and the time and energy information is (0.045657025, 44)/(time/second, energy/channel).

Based on above Step 1 according to the present disclosure, it is concluded that the ray belongs to ROI₁ (33, 51) corresponding to the 32 keV ray and ROI₂ (38, 58) corresponding to the 35 keV ray.

The energy Bayesian factors of ROI₁ and ROI₂ are calculated in Step 2, which are: B₁^π(ϑ_1,1)=0.6417 and B₂^π(ϑ_2,1)=0.8062.

The energy decision functions of ROI₁ and ROI₂ are calculated in Step 3, which are: p_1,0^π=0.3909 and p_2,0^π=0.4464.

The energy prior probabilities of ROI₁ and ROI₂ are updated in Step 4, which are: π_1,0=0.3909 and π_2,0=0.4464.

The time intervals, the time Bayesian factors and the time decision functions of ROI₁ and ROI₂ are calculated in Steps 5 to 8, which are:Δt₁=0.0457,B₁^ϑ(Δt₁)=0.8858,p_1,0^ϑ=0.4697; andΔt₂=0.0457,B₂^ϑ(Δt₂)=1.0280,P_2,0^ϑ=0.5069

A decision is made on the 32 keV characteristic γ ray corresponding to ROI₁ and the 35 keV characteristic γ ray corresponding to ROI₂ in Step 9.

Because p_1,0^π=0.3909 and p_1,0^ϑ=0.4697, ROI₁ makes no decision and waits for the next ray. Because p_2.0^π=0.4464 and p_2.0^ϑ=0.5069, ROI₂ makes no decision and waits for the next ray.

It is assumed that it is determined in Step 9 that there is 32 keV characteristic γ ray corresponding to ROI₁ or ROI₁ is the background signal. Since the 32 keV ray belongs to radionuclide ¹³³Ba, it is necessary to investigate whether there is radionuclide ¹³³Ba at this time.

In Step 10, the time and energy decision functions of ROI₁, ROI₅, ROI₁₇, ROI₁₈, ROI₂₂ and ROI₂₃ corresponding to characteristic γ rays of 32 keV, 81 keV, 276 keV, 302 keV, 356 keV and 383 keV are used to calculate the joint decision function of the radionuclide ¹³³Ba.

In Step 11, radionuclides are identified according to the joint decision function of ¹³³Ba.

After the arrival of the next ray, Step 1 to Step 8 are repeated according to the time and energy information of the ray to calculate the time and energy decision functions of the corresponding ROI. Thereafter, Step 9 is repeated to identify the characteristic γ rays. Finally, Step 10 is repeated to identify radionuclides.

In this example, a total of 10⁴ radiation particles are measured and identified, with a total time of 166.25 seconds and a full-spectrum counting rate of 60.15 s¹.

In order to explain the process of nuclide identification in detail, take the 3424-th ray detected by the detector as an example (that is, the ray particle when it is determined for the first time that there is no 661 keV characteristic γ ray and the corresponding nuclide ¹³⁷Cs), so as to observe the changes of the γ-ray decision function and the nuclide joint decision function before and after this ray.

The detector has detected the 3424-th ray, and the time and energy information is (57.8514355, 881)/(time/second, energy/channel). At this time, the prior probabilities of energy and time are updated to π_31,0=0.999975 and ϑ_31,0=0.6936.

According to Step 1, it is concluded that this ray belongs to ROI₃₁ (857,900) corresponding to the ray with energy of 661 keV.

The energy Bayesian factor of ROI₃₁ is calculated in Step 2, which is B₃₁^π(ϑ_31,28)=0.6251.

The energy decision function of ROI₃₁ is calculated in Step 3, which is p_31,0^π=0.999960.

The energy prior probability of ROI₃₁ is updated in Step 4, which is π_31,0=0.999960.

The time interval, the time Bayesian factor and the time decision function of ROI₃₁ are calculated in Step 5 to Step 8, which are:Δt₃₁=0.0042,B₃₁^ϑ(Δt₃₁)=1.8946, and p_31,0^ϑ=0.8109

A decision is made on the 661 keV characteristic γ ray corresponding to ROI₃₁ in Step 9. Because p_31,0^π=0.999960>P_{31,up_th}^π=0.8 and p_31,0^ϑ=0.8109>P_{31,up_th}^ϑ=0.8, a determination is made that all ROI₃₁ signals come from the background. Because the characteristic energy corresponding to ROI₃₁ signals is 661 keV, the corresponding characteristic γ ray belongs to nuclide ¹³⁷Cs. Thereafter, the joint decision function of nuclide ¹³⁷Cs needs to be calculated to judge whether there is nuclide ¹³⁷Cs for updating.

The joint decision function of nuclide ¹³⁷Cs is calculated in Step 10, which is: P_137Cs=0.8109.

In Step 11, a decision is made on whether there is nuclide ¹³⁷Cs. Because P_137Cs=0.8109>P_{137Cs,up_th}=0.64, a judgment is made that there is no nuclide ¹³⁷Cs. Because the nuclide ¹³⁷Cs has only one 661 keV characteristic γ ray, the decision on that there is no nuclide ¹³⁷Cs can be made when all the ROI₃₁ signals come from the background in Step 8. The joint decision function of the nuclide ¹³⁷Cs is calculated here only to explain the detailed process of nuclide identification.

It is determined that there are no nuclide ²⁴¹Am and its 59 keV characteristic γ ray at the same time in 14.9575 seconds (a total of 881 samples and 17 ROI₄ samples). It is determined that there are no nuclide ¹³⁷Cs and its 661 keV characteristic γ ray at the same time in 57.8611 seconds (a total of 3424 samples and 28 ROI₃₁ samples). The changes of two γ ray decision functions with time are shown in FIGS. 3 and 4. The changes of the nuclide joint decision functions of ²⁴¹Am and ¹³⁷Cs with time are shown in FIGS. 5 and 6. In FIG. 3, the final identification result shows that: there is no 59.54 keV γ ray, the spent time is 14.9757 s, the number of full-spectrum rays is 881, and the number of rays in ROI is 17. In FIG. 4, the final identification result shows that: there is no 661 keV γ ray, the spent time is 57.8511 s, the number of full-spectrum rays is 3424, and the number of rays in ROI is 28. In FIG. 5, the final identification result shows that there is no nuclide Am-241. In FIG. 6, the final identification result shows that there is no nuclide Cs-137.

In addition, under the background condition, there is no incident of radionuclide false alarm (the false alarm is defined as: there is no radionuclide, but it is judged that there is radionuclide). When it is determined that the above two characteristic γ rays do not exist and all the signals in the corresponding ROI come from the background, the time required is too long because the sample collection rate in the ROI corresponding to the characteristic γ ray under the background condition is too low. When making an effective determination, it can be found by observation of the number of samples collected in the ROI that effective identification can be made if a small number of samples are collected in the ROI.

Embodiment 2

Components and parameters used in this embodiment are the same as those in Embodiment 1 described above. There is the identification result of ¹³⁷Cs (9.22*10³Bq) (the distance between the source and the front end of the detector is 35 cm) (the equivalent dose rate is about 5.52 nGy/h).

The detector detects the first ray, and the time and energy information is (0.030928838, 161)/(time/second, energy/channel).

According to Step 1, it is concluded that the ray belongs to ROI₈ (150, 171) corresponding to the 122 keV ray.

The energy Bayesian factor of the ROI₈ is calculated in Step 2, which is: B₁^π(ε_1,1)=0.6258.

The energy decision function of the ROI₈ is calculated in Step 3, which is: p_1,0^π=0.3849.

The energy prior probability of the ROI₈ is updated in Step 4, which is: π_1,0=0.3849.

The time interval, the time Bayesian factor and the time decision function of the ROI₈ are calculated in Steps 5 to 8, respectively, which are:Δt₁=0.0309,B₁^ϑ(Δt₁)=0.7079 and p_1,0^ϑ=0.4145

In Step 9, a decision is made on the 122 keV ray corresponding to ROI₈. Because p_1,0^π=0.3849 and p_1,0^ϑ=0.4145, ROI₈ makes no decision and waits for the next ray.

If it is determined in Step 8 that there is γ ray, assuming that there is the 122 keV ray corresponding to ROTS, it is necessary to investigate whether there are radionuclides ⁵⁷Co and ¹⁵²Eu at this time, specifically as follows.

In Step 10, the time and energy decision functions of ROI₈ and ROI₁₀ corresponding to the 122 keV ray and the 136 keV ray are used to calculate the joint decision function of nuclide ⁵⁷Co. The time and energy decision functions of ROI₈, ROI₁₆, ROI₂₁, ROI₃₂, ROI₃₆, and ROI₄₂ corresponding to 122 keV, 244 keV, 344 keV, 779 keV, 964 keV and 1408 keV are used to calculate the joint decision function of nuclide ¹⁵²Eu.

In Step 11, according to the joint decision function of ⁵⁷Co and ¹⁵²Eu, the determination of the hypothesis test of the existence of two radionuclides is made.

After the arrival of the next ray, Step 1 to Step 8 are repeated according to the time and energy information of the ray to calculate the time and energy decision functions of the corresponding ROI. Thereafter, Step 9 is repeated to identify the γ rays. Finally, Step 10 and Step 11 are repeated to identify radionuclides.

In this example, a total of 10⁴ radiation particles are measured and identified, with a total time cost of 153.70 seconds and a full-spectrum counting rate of 64.14 s⁻¹.

In order to explain the process of nuclide identification in detail, take the 281st ray detected by the detector as an example (that is, the ray particle when it is determined for the first time that there is 661 keV characteristic γ ray and the nuclide ¹³⁷Cs), so as to observe the changes of the γ-ray decision function and the nuclide joint decision function before and after this ray.

The detector has detected the 281st ray, and the time and energy information is (4.260600813, 865)/(time/second, energy/channel). At this time, the prior probabilities of energy and time are updated to π_31,0=0.1818 and ϑ_31,0=0.2220.

According to Step 1, it is concluded that this ray belongs to ROI₃₁ (857,900) corresponding to 661 keV ray.

The energy Bayesian factor of ROI₃₁ is calculated in Step 2, which is: B₃₁^π(ε_31.28)=1.3158.

The energy decision function of ROI₃₁ is calculated in Step 3, which is p_31,0^π=0.1818.

The energy prior probability of ROI₃₁ is updated in Step 4, which is π_31,0=0.1818.

The time interval, the time Bayesian factor and the time decision function of ROI₃₁ are calculated in Step 5 to Step 8, respectively, which are:Δt₃₁=0.0075,B₃₁^ϑ(Δt₃₁)=0.6932, and p_31,0^ϑ=0.1698.

A decision is made on the 661 keV characteristic γ ray corresponding to ROI₃₁ in Step 9. Because p_31,0^π=0.1818<P_{31,low_th}^π=0.2 and p_31,0^ϑ=0.1698<P_{31,low_th}^ϑ=0.2, a judgment is made that there is the 661 keV characteristic γ ray. Because the 661 keV characteristic γ ray corresponding to ROI₃₁ signals belongs to nuclide ¹³⁷Cs, thereafter, the joint decision function of nuclide ¹³⁷Cs needs to be calculated to judge whether there is nuclide ¹³⁷Cs for updating.

The joint decision function of nuclide ¹³⁷Cs is calculated in Step 10, which is: P_137Cs=0.0309.

In Step 11, a decision is made on whether there is nuclide ¹³⁷Cs. Because P_137Cs=0.0309<P_{137Cs,low_th}=0.04, a determination is made that there is nuclide ¹³⁷Cs. Because the nuclide ¹³⁷Cs has only one 661 keV characteristic γ ray, the decision on that there is no nuclide ¹³⁷Cs can be made when all the ROI₃₁ signals come from the background in Step 11. The joint decision function of the nuclide ¹³⁷Cs is calculated here only to explain the detailed process of nuclide identification.

It is determined that there are no nuclide ²⁴¹Am and its 59 keV characteristic γ ray at the same time in 72.6919 seconds (a total of 4613 samples and 98 ROI₄ samples). It is determined that there are nuclide ¹³⁷Cs and its 661 keV characteristic γ ray at the same time in 4.2606 seconds (a total of 281 samples and 7 ROI₃₁ samples). The changes of two γ ray decision functions with time are shown in FIGS. 7 and 8. The changes of the nuclide joint decision functions of ²⁴¹Am and ¹³⁷Cs with time are shown in FIGS. 9 and 10. In FIG. 7, the final identification result shows that there is no 59.54 keV γ ray, the spent time is 72.6919 s, the number of full-spectrum rays is 4613, and the number of rays in ROI is 98. In FIG. 8, the final identification result shows that there is 661.7 keV γ ray, the spent time is 4.2606 s, the number of full-spectrum rays is 281, and the number of rays in ROI is 7. In FIG. 9, the final identification result shows that there is no nuclide ²⁴¹Am. In FIG. 10, the final identification result shows that there is ¹³⁷Cs.

Compared with the nuclide identification process under the above background conditions (namely Embodiment 1 and Embodiment 2), the following content can be found.

Under the conditions of background and ¹³⁷Cs radioactive source, the characteristic γ-ray decision function and the nuclide joint decision function of ²⁴¹Am are both close to 1, and the determination that there is no characteristic γ-ray and ²⁴¹Am nuclide is made. At the same time, under the conditions of background and ¹³⁷Cs radioactive source, there is no false alarm event. The Compton count of ROI₄ corresponds to 661 keV ray of ¹³⁷Cs at 59 keV of ²⁴¹Am only influences the speed of its convergence to 1, without influencing its precision (the results of other characteristic gamma rays and corresponding nuclides not corresponding to 661 keV are the same as the determination and identification result of ²⁴¹Am).

Under the background condition, the characteristic gamma-ray decision function of ¹³⁷Cs quickly approaches to 1, making a determination that there is no characteristic γ ray and making a determination that there is no ¹³⁷Cs nuclide. Under the condition that there is ¹³⁷Cs nuclide, both the characteristic γ ray decision function and the nuclide joint decision function quickly approach to zero, a determination that there is characteristic γ ray is made and a determination that there is ¹³⁷Cs nuclide is made. It can be seen that this example has realized the rapid identification of ¹³⁷Cs nuclides (the identification time is 4.2606 seconds, there is a total of 281 samples, and the equivalent dose rate is about 5.52 nGy/h).

The nuclide identification method of the present disclosure significantly is improved in items of the universality, based on a Bayesian principle and a sequential posterior probability, through changing the test model of the nuclide identification method and setting a value range of a time interval on the prior probability, instead of directly assigning a predefined fixed value to the time interval parameter. According to statistical characteristics, a background probability density function and a Compton probability density function in an ROI are determined. An energy Bayesian factor and a time Bayesian factor are determined based on energy and time interval information in a sequence of the nuclear detection events obtained by measurement. By combining the two factors, Compton plateau can be effectively identified and distinguished. Under the same identification conditions and the same confidence level, the method in this disclosure can effectively identify the existence and types of radionuclides faster than an energy spectrum decomposition analysis-characteristic peak matching method. Compared with the fuzzy mathematics and neural network nuclide identification methods, the method in this disclosure is more universal.

In addition, compared with the sequential nuclide identification method based on the Bayesian theory and the sequential probability ratio test published in previous research, aiming at the problem that Compton plateau cannot be distinguished and the universality is poor in previous research, for the sequential Bayesian nuclide identification method based on the Bayesian method and the sequential posterior probability according to the present disclosure, it can be found, according to the above examples, that the false alarm rate and the missed alarm rate are very low, and it shows that the method can effectively identify Compton particles from high-energy rays appearing in the low-energy ROI.

Further, the present disclosure provides a nuclide identification system, which is used to implement the nuclide identification method described above. The system includes: a ray belonging region determining module, a nuclear detection event sequence description module, an energy Bayesian factor determining module, an energy decision function determining module, an ROI time interval determining module, a time Bayesian factor determining module, a time decision function determining module, a decision function combining module, a nuclide joint decision function determining module, and a nuclide identification module. The above modules are modules implemented on computer.

The ray belonging region determining module is configured to acquire time-energy information of a ray, and determine whether the ray belongs to one of ROIs based on energy information in time-energy information of the ray.

The nuclear detection event sequence description module is configured to, when the ray belongs to the one of ROIs, describe a sequence of all the nuclear detection events in the ROI.

The energy Bayesian factor determining module is configured to determine an energy Bayesian factor based on the energy information in the described sequence of all the nuclear detection events.

The energy decision function determining module is configured to determine an energy decision function based on the energy Bayesian factor.

The ROI time interval determining module is configured to determine a time interval in the ROI based on the time information in the described sequence of all the nuclear detection events.

The time Bayesian factor determining module is configured to determine a time Bayesian factor based on the time interval.

The time decision function determining module is configured to determine a time decision function based on the time Bayesian factor.

The decision function combining module is configured to combine the energy decision function and the time decision function to obtain a joint decision function of the ROI.

The nuclide joint decision function determining module is configured to, based on a characteristic γ ray corresponding to the ROI, retrieve a potential nuclide corresponding to the ROI, and combine the joint decision functions of respective ROIs corresponding to the retrieved nuclide to obtain a nuclide joint decision function.

The nuclide identification module is configured to determine and identify the retrieved nuclide based on the nuclide joint decision function.

Still further, the present disclosure further provides an electronic device, which includes a memory and a processor.

A computer program is stored in the memory.

The processor is connected with the memory, and is configured to call and execute the computer program to implement the nuclide identification method described above.

The computer program in the above-mentioned memory can be stored in a computer-readable storage medium when it is implemented in the form of software functional units and sold or used as an independent product. Based on this understanding, according to the present disclosure, the essence of the technical solution or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions to make a computer device (which can be a personal computer, a server or a network device, etc.) execute all or part of the steps of the method described in various embodiments of the present disclosure. The aforementioned storage media include: a USB flash disk, a mobile hard disk, a read-only memory, a random access memory, a magnetic disk or an optical disk and other media that can store program codes.

Further, based on the above description, (1) the nuclide identification method according to the present disclosure can be used in cooperation with various types of detectors, including but not limited to scintillator detectors, semiconductor detectors and other detectors with energy resolution. (2) Parameters such as the number and types of nuclides in the nuclide standard library, the number and importance of γ rays selected by each nuclide are not limited to the form in the example of the present disclosure. (3) The probability density of the energy and the time interval in the process of nuclide identification is not limited to Gaussian distribution and uniform distribution in the example of the present disclosure, and the specific probability density distribution can be determined according to the actual use scene. (4) The initial values of the prior probabilities of the null hypothesis and the alternative hypothesis in the process of nuclide identification can be determined according to the actual situation, including but not limited to the parameters (0.5, 0.5) in the example of the present disclosure. (5) The prior probability density function can be a non-informative prior probability density function or a conjugate prior probability density function, and is not limited to the non-informative prior probability density function in the example of the present disclosure. (6) The selection of the time interval value range in the prior probability is related to background conditions and detector types, including but not limited to the parameters in the example of the present disclosure. (7) In the process of nuclide identification, the upper and lower thresholds are related to the false alarm rate and the missed alarm rate, and their settings can be set according to the actual use scenes and requirements, including but not limited to the parameters in the example of the present disclosure. (8) For the division of the ROI of the full energy peak region, the size of the k value can be set according to the specific situation, including but not limited to the parameters in the example of the present disclosure. (9) The decision function in the present disclosure is solved according to the prior probability and the Bayesian factor, and the Bayesian factor, as an intermediate parameter, may not participate in the calculation process in the actual use calculation process.

Various embodiments of the present disclosure are described in a progressive way, and each embodiment focuses on the description that is different from the other embodiments, and the same and similar parts between various embodiments can be referred to with each other. Since the system disclosed in the embodiment corresponds to the method disclosed in the embodiment, the system is described simply. Refer to the description of the method for the relevant points.

In the present disclosure, specific examples are applied to illustrate the principles and implementations of the present disclosure, and the explanations of the above embodiments are only used to help understand the method and core ideas of the present disclosure. At the same time, according to the idea of the present disclosure, there will be some changes in the specific implementations and application scope for those skilled in the art. To sum up, the contents of the specification should not be construed as limiting the present disclosure.

Claims

1. A nuclide identification method, comprising: receiving, at a computer, the time-energy information of a ray from a detector, and performing following steps: determining whether the ray belongs to one of Region of Interests (ROIs) based on energy information in the time-energy information of the ray;describing a sequence of all nuclear detection events in the ROI when the ray belongs to the one of the ROIs;determining an energy Bayesian factor based on energy information in the described sequence of all the nuclear detection events;determining an energy decision function based on the energy Bayesian factor;determining a time interval in the ROI based on time information in the described sequence of all the nuclear detection events;determining a time Bayesian factor based on the time interval;determining a time decision function based on the time Bayesian factor;combining the energy decision function and the time decision function to obtain a joint decision function of the ROI;retrieving a potential nuclide corresponding to the ROI based on a characteristic γ ray corresponding to the ROT, and combining joint decision functions of respective ROIs corresponding to the retrieved nuclide to obtain a nuclide joint decision function;determining whether the retrieved nuclide exists based on the nuclide joint decision function; andoutputting a result that the retrieved nuclide exists in respond to a determination that a value of the nuclide joint decision function is less than or equal to a first predefined value; or a result that the retrieved nuclide does not exist in respond to a determination that the value of the nuclide joint decision function is greater than or equal to a second predefined value, wherein the second predefined value is greater than the first predefined value; or a result of failure in respond to a determination that the value of the nuclide joint decision function is greater than the first predefined value and less than the second predefined value.

2. The nuclide identification method according to claim 1, wherein the energy Bayesian factor is:

3. The nuclide identification method according to claim 1, wherein the energy decision function of the ROI is:

4. The nuclide identification method according to claim 3, wherein after the determining an energy decision function based on the energy Bayesian factor, the method further comprises: updating the energy prior probability using the energy decision function.

5. The nuclide identification method according to claim 1, wherein the time Bayesian factor is:

6. The nuclide identification method according to claim 1, wherein the time decision function is:

7. The nuclide identification method according to claim 6, wherein after the determining a time decision function based on the time Bayesian factor, the method further comprises: updating the time prior probability based on the time decision function.

8. The nuclide identification method according to claim 1, wherein prior to the combining the energy decision function and the time decision function to obtain a joint decision function of the ROI, the method further comprises: based on a relationship between a calculation result of the energy decision function and upper and lower thresholds of the energy decision function and a relationship between a calculation result of the time decision function and upper and lower thresholds of the time decision function, obtaining a ray signal identification result in the ROI.

9. A nuclide identification system, which is used to implement the nuclide identification method according to claim 1, comprises: a ray belonging region determining module, configured to receive time-energy information of a ray from the detector, and determine whether the ray belongs to one of Region of Interests (ROIs) based on energy information in the time-energy information of the ray;a nuclear detection event sequence description module, configured to describe a sequence of all nuclear detection events in the ROI when the ray belongs to the one of the ROIs;an energy Bayesian factor determining module, configured to determine an energy Bayesian factor based on energy information in the described sequence of all the nuclear detection events;an energy decision function determining module, configured to determine an energy decision function based on the energy Bayesian factor;an ROI time interval determining module, configured to determine a time interval in the ROI based on time information in the described sequence of all the nuclear detection events;a time Bayesian factor determining module, configured to determine a time Bayesian factor based on the time interval;a time decision function determining module, configured to determine a time decision function based on the time Bayesian factor;a decision function combining module, configured to combine the energy decision function and the time decision function to obtain a joint decision function of the ROI;a nuclide joint decision function determining module, configured to retrieve a potential nuclide corresponding to the ROI based on a characteristic γ ray corresponding to the ROI, and combine joint decision functions of respective ROIs corresponding to the retrieved nuclide to obtain a nuclide joint decision function;a nuclide identification module, configured to determine whether the retrieved nuclide exists based on the nuclide joint decision function; andan outputting module, configured to output a result that the retrieved nuclide exists in respond to a determination that a value of the nuclide joint decision function is less than or equal to a first predefined value; or a result that the retrieved nuclide does not exist in respond to a determination that the value of the nuclide joint decision function is greater than or equal to a second predefined value, wherein the second predefined value is greater than the first predefined value; or a result of failure in respond to a determination that the value of the nuclide joint decision function is greater than the first predefined value and less than the second predefined value.

10. The nuclide identification system according to claim 9, wherein the energy Bayesian factor is:

11. The nuclide identification system according to claim 9, wherein the energy decision function of the ROI is:

12. The nuclide identification system according to claim 11, wherein after the determining an energy decision function based on the energy Bayesian factor, the method further comprises: updating the energy prior probability using the energy decision function.

13. The nuclide identification system according to claim 9, wherein the time Bayesian factor is:

14. The nuclide identification system according to claim 9, wherein the time decision function is:

15. The nuclide identification system according to claim 14, wherein after the determining a time decision function based on the time Bayesian factor, the method further comprises: updating the time prior probability based on the time decision function.

16. The nuclide identification system according to claim 9, wherein prior to the combining the energy decision function and the time decision function to obtain a joint decision function of the ROI, the method further comprises: based on a relationship between a calculation result of the energy decision function and upper and lower thresholds of the energy decision function and a relationship between a calculation result of the time decision function and upper and lower thresholds of the time decision function, obtaining a ray signal identification result in the ROI.

17. An electronic device, comprising: a memory in which a computer program is stored; anda processor, which is connected with the memory, and is configured to call and execute the computer program to implement the nuclide identification method according to claim 1.

18. The electronic device according to claim 17, wherein the energy Bayesian factor is:

19. The electronic device according to claim 17, wherein the energy decision function of the ROI is:

20. The electronic device according to claim 19, wherein after the determining an energy decision function based on the energy Bayesian factor, the method further comprises: updating the energy prior probability using the energy decision function.