Neutron source searching robot
By using plastic scintillator detectors and electronic systems in neutron source search robots, combined with Gaussian process regression method and neutron/photon identification technology, the inability to autonomously search and identify neutron sources in the existing technology is solved. The problem has achieved autonomous positioning and radiation distribution map reconstruction in unknown radiation environments, and improved the safety and efficiency of nuclear accident handling.
Patent Information
- Application Number
- CN202411253002.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-06
- Publication Date
- 2024-11-01
AI Technical Summary
Existing radiation source source search robots cannot independently search and identify neutron sources, and cannot independently move and estimate the location of the radiation source in an unknown radiation environment, reconstruct the radiation hot spot distribution map, and judge the type of radiation source.
A neutron source search robot was designed, using a plastic scintillator detector with n/γ identification function and an electronic system, combined with the upper computer system to generate an environmental map model, plan the walking path, and reconstruct the radiant heat spots through Gaussian process regression method. Distribution map and use neutron/photon identification technology to determine the type of radiation source.
It realizes autonomous positioning of radio sources in an unknown radiation environment, distinguishing neutron sources from gamma sources, and reconstructing the radiation distribution map, improving the location and evaluation efficiency of neutron source loss in nuclear emergency accidents, and enhancing safety and operation efficiency.
Smart Images

Figure CN118876084A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of neutron source tracing, and in particular to a neutron source tracing robot. Background Art
[0002] The statements herein merely provide background information related to the present application and do not necessarily constitute prior art.
[0003] A neutron source is a device that can release neutrons. It includes isotope neutron sources, accelerator neutron sources, and reactor neutron sources. With the development and popularization of the nuclear industry and nuclear technology, isotope neutron sources are widely used in industry, agriculture, medicine, scientific research, and other fields. With the popularization of isotope neutron sources, neutron source loss accidents occur frequently. For radioactive source loss accidents, manual disposal will cause certain health risks. Summary of the invention
[0004] A brief overview of the present application is provided below in order to provide a basic understanding of certain aspects of the present application. It should be understood that this overview is not an exhaustive overview of the present application. It is not intended to identify the key or important parts of the present application, nor is it intended to limit the scope of the present application. Its purpose is merely to present certain concepts in a simplified form as a prelude to a more detailed description discussed later.
[0005] The present application provides a neutron source search robot, comprising: a robot platform, the robot platform comprising a movable chassis, the movable chassis being used to move in a source search area; a plastic scintillator detector, fixed to the robot platform, used to receive neutrons and photon rays when the movable chassis moves, and convert the received neutrons and photon rays into electric pulse signals; an electronics system, fixed to the robot platform, used to receive the electric pulse signals converted by the plastic scintillator detector, and process the electric pulse signals to obtain processed signals; a host computer system, used to generate an environmental map model of the source search area, and plan the walking path of the movable chassis according to the environmental map model and the real-time positioning of the robot platform, and control the movable chassis to move along the walking path, the host computer system is also used to receive the processed signal, and estimate the position of the radiation source according to the processed signal, and reconstruct the radiation hotspot distribution map of the source search area, and is used to determine whether the radiation source is a neutron source or a gamma source.
[0006] Compared with the previous radiation source search robots that can only search for γ radiation sources, the neutron source search robot provided by this application uses a plastic scintillator detector with n / γ discrimination function as the core neutron detector. While locating the radiation source, it can distinguish the type of radiation source as a γ source or a neutron source, providing technical support for finding lost neutron sources in nuclear emergency accidents. In addition, while locating the radiation source, it can reconstruct the radiation distribution map in the area to be searched, providing support for evaluating the radiation distribution in the area. BRIEF DESCRIPTION OF THE DRAWINGS
[0007] Other objects and advantages of the present application will be apparent from the following description of the present application with reference to the accompanying drawings, which will help to provide a comprehensive understanding of the present application.
[0008] Figure 1 is a structural block diagram of a neutron source seeking robot according to some embodiments of the present application;
[0009] Figure 2 is a structural block diagram of an electronic system according to some embodiments of the present application;
[0010] Figure 3 is a structural block diagram of an internal control system according to some embodiments of the present application;
[0011] Figure 4 It is the radiation intensity distribution prediction diagram of RBF covariance function;
[0012] Figure 5 is the radiation intensity distribution prediction diagram of the Matern covariance function when ν = 3;
[0013] Figure 6 is the radiation intensity distribution prediction diagram of the Matern covariance function when ν = 5;
[0014] Figure 7 It is the radiation intensity distribution prediction diagram of the Exponential covariance function;
[0015] Figure 8 It is the radiation intensity distribution prediction diagram of the Rational Quadratic covariance function;
[0016] Fig. 9 It is the relationship curve between lengthscale and MSE value;
[0017] Fig.10 It is the relationship curve between variance and MSE value;
[0018] Fig.11 It is the relationship curve between r and MSE value;
[0019] Fig.12It is a point cloud map of the environment generated by laser scanning;
[0020] Fig.13 It is a point cloud map of the environment generated by laser scanning;
[0021] Fig.14 It is the pulse amplitude spectrum and PSD spectrum measured in the source search mission;
[0022] Fig.15 It is the actual source-seeking path diagram of the neutron source-seeking robot;
[0023] Fig.16 It is the radiation hotspot distribution map reconstructed by the neutron source homing robot;
[0024] Fig.17 This is a top view of the installation of the neutron source locating robot of this embodiment.
[0025] Explanation of the accompanying drawings: 100, host computer system; 200, robot platform; 210, plastic scintillator detector; 220, electronics system; 221, processing module; 222, main amplifier module; 230, internal control system; 231, general control layer; 2311, embedded computer; 23111, Linux operating system; 23112, robot operating system; 232, USB bus; 233, sub-control layer; 2331, intelligent sensor; 2332, walking system; 2333, actuator; 234, network remote control; 2341, Linux operating system; 2342, robot operating system; 2343, network; 2344, human-computer interaction module; 235, cloud management platform. DETAILED DESCRIPTION
[0026] In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be clearly and completely described below in conjunction with the drawings of the embodiments of the present application. Obviously, the described embodiment is one embodiment of the present application, not all embodiments. Based on the described embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0027] Unless otherwise defined, technical or scientific terms used in this application should have the common meanings understood by persons having ordinary skills in the field to which this application belongs.
[0028] Compared with manual operation, radioactive source search robots are safer and more efficient, and are suitable for handling radiation accidents. Most of the existing source search robots are manually remote-controlled robots that search for radioactive sources. In this method, the robots do not have autonomy and move along a manually designated route, collecting radioactive information on the way to find radioactive sources. In such tasks, the autonomy and intelligence of the robots are not reflected, and manual operation may reduce the work efficiency of the robots. In addition, the existing source search robots search for X / γ sources and cannot identify neutron sources.
[0029] In order to solve the problem that existing source-seeking robots cannot autonomously search for sources and identify the types of radioactive sources, this embodiment provides a neutron source-seeking robot that moves in an unknown radiation environment and estimates the location of the radioactive source, reconstructs the distribution map of radiation hot spots in the area, and determines whether the found radioactive source is a neutron source or a gamma source. The source-seeking robot samples different unknown radioactive information in the area through the plastic scintillator detector it carries, and uses real-time positioning and mapping technology to build an environmental map. The source-seeking robot autonomously plans the path, and analyzes the radiation count rate collected along the way in combination with the position information to complete the positioning of the radioactive source. After the radioactive source is located, the type of the radioactive source is determined to be a neutron source or a gamma source through neutron / photon discrimination technology.
[0030] Figure 1 is a structural block diagram of a neutron source locating robot according to some embodiments of the present application, such as Figure 1 As shown, the neutron source locating robot provided in this embodiment includes a robot platform 200 , a plastic scintillator detector 210 , an electronics system 220 and a host computer system 100 .
[0031] The robot platform 200 includes a movable chassis, which is used to move in the source area to be found. The plastic scintillator detector 210 is fixed to the robot platform 200, and is used to receive neutrons and photon rays when the movable chassis moves, and convert the received neutrons and photon rays into electrical pulse signals. The electronics system 220 is fixed to the robot platform 200, and is used to receive the electrical pulse signal converted by the plastic scintillator detector 210, and process the electrical pulse signal to obtain a processed signal. The host computer system 100 is used to generate an environmental map model of the source area to be found, and plan the walking path of the movable chassis according to the environmental map model and the real-time positioning of the robot platform 200, and control the movable chassis to move according to the walking path. The host computer system 100 is also used to receive the processed signal, and estimate the position of the radiation source according to the processed signal, and reconstruct the radiation hot spot distribution map of the source area to be found, and is used to determine whether the radiation source is a neutron source or a gamma source.
[0032] Compared with the previous radiation source search robots that can only search for γ radiation sources, the neutron source search robot provided in this embodiment uses a plastic scintillator detector 210 with n / γ discrimination function as the core neutron detector, which can distinguish the type of radiation source as a γ source or a neutron source while locating the radiation source, providing technical support for finding lost neutron sources in nuclear emergency accidents. In addition, while locating the radiation source, it can reconstruct the radiation distribution map in the area to be searched, providing support for evaluating the radiation distribution in the area.
[0033] Specifically, in the present neutron source search robot, the plastic scintillator detector 210 receives neutrons or photon rays, converts them into electrical pulse signals, and outputs the signals to the electronics system 220 via the BNC interface. The electronics system 220 amplifies, conditions, filters, waveform samples, acquires amplitude, and identifies the waveform of the pulse signals. The processed signal data is transmitted to the robot internal control system 230 via the network cable interface. The electronics system 220 and the robot internal control system 230 can exchange information. The robot internal control system 230 provides low-voltage power supply to the electronics system 220 via the J30JZ interface, and the electronics system 220 provides working high voltage to the plastic scintillator detector 210 via the SHV interface. The robot internal control system 230 remotely interacts with the host computer system 100 via a built-in router in the form of a wireless remote LAN.
[0034] The main components of the plastic scintillator detector 210 are scintillator, optical collection system, photomultiplier tube, and voltage divider for supplying power to each electrode of the photomultiplier tube. The detector uses the property of scintillator to emit light under the action of nuclear radiation to detect nuclear radiation. The photoelectric device converts the scintillation light into photoelectrons, which are amplified multiple times to output an electrical pulse. The pulse signal shape generated by neutrons and photons in the scintillator is different to a certain extent, and the type of incident particles can be judged by neutron / photon discrimination technology.
[0035] Figure 2 is a block diagram of an electronic system 220 according to some embodiments of the present application, such as Figure 2As shown, the electronic system 220 realizes the functions of amplifying, collecting, processing and transmitting the detector signal, and is mainly divided into two parts: the processing module 221 and the main amplifier module 222. The main amplifier module 222 can amplify the charge signal, filter and condition the signal, and feed the detector with high voltage. It has the ability to supply high voltage, and can also realize the conditioning, filtering and amplification of the circuit detection signal; the processing module 221 can perform signal acquisition and processing, state machine timing control, data management, and use the field programmable gate array for full digital parallel processing to realize the functions of pulse waveform sampling, pulse amplitude acquisition, pulse waveform identification, etc. The main amplifier module 222 and the processing module 221 can interact with the data signal and the control signal, the electronic system 220 can receive the detection signal of the plastic scintillator detector 210, and can supply high voltage to the plastic scintillator detector 210. The electronic system 220 can interact with the host computer system 100 of the robot platform 200 through the network port, and the power supply device of the robot platform 200 supplies power to the electronic system 220.
[0036] Figure 3 is a structural block diagram of an internal control system 230 according to some embodiments of the present application, such as Figure 3As shown, the architecture of the internal control system 230 is a layered architecture, including: a master control layer 231 and a sub-control layer 233, which can be connected via a USB bus 232. The master control layer 231 has an embedded computer 2311 (x86 architecture CPU) as the core control unit, running a Linux operating system 23111 (CPU) and a robot operating system 23112 (ROS), and can communicate with a cloud management platform 235 and a network remote controller 234, wherein the network remote controller 234 includes a Linux operating system 2341 (CPU), a robot operating system 2342 (ROS), a network 2343, and a human-computer interaction module 2344; the sub-control layer 233 is composed of a series of MCU controllers, for example, including an intelligent sensor 2331, a walking system 2332, and an actuator 2333, and each MCU controller manages an independent functional unit. The sub-control unit can be connected to the master control unit through the standard connection of the robot to achieve overall control and data transmission. The master control software architecture is the robot operating system open source software (ROS). The bottom layer of the system is the Linux operating system, which is responsible for the file system, hardware driver, memory management, task scheduling, etc. The middle layer is the robot operating system ROS, which is responsible for sub-control unit management, multi-task scheduling and data exchange, human-computer interaction basic platform, etc. The upper layer is the application of each sub-control unit and the conventional application of the robot. The internal control system 230 can be installed on the robot platform 200 to provide low-voltage power supply to the electronics system 220 so that the electronics system 220 provides working high voltage to the plastic scintillator detector 210. Among them, the internal control system 230 is also used to receive processed signals, and the internal control system 230 communicates wirelessly with the host computer system 100 so that the host computer system 100 receives processed signals.
[0037] The host computer system 100 of the source-seeking robot can be installed with host computer software. The host computer software is a web application using CS architecture, based on the x86 system platform, with the back-end server written in Java language and the web front-end written in Javascript. The main functions of the host computer software are as follows: (1) Environmental modeling based on the laser radar, and forming a three-dimensional point cloud map and a two-dimensional grid map; (2) Based on the environmental map and real-time positioning, the robot is planned to move along the specified route; (3) The interaction with the electronic system is realized, the output high voltage and internal parameter settings of the electronic system are controlled, and the pulse amplitude spectrum of the detector data processing result is displayed online (the host computer system is also used for the online display of the pulse amplitude spectrum of the processed signal) and the online analysis of neutron / photon discrimination.
[0038] In some embodiments, a Gaussian process regression method is used to reconstruct a radiation hotspot distribution map of the area to be found.
[0039] Gaussian process regression is a new machine learning method developed based on Bayesian theory and statistical learning theory. It is suitable for dealing with complex regression problems such as high-dimensionality, small samples and nonlinearity. Compared with neural networks and support vector machines, it has the advantages of easy implementation, adaptive acquisition of hyperparameters, flexible nonparametric inference, and probabilistic output.
[0040] From the perspective of function space, a Gaussian process (GP) is defined to describe the function distribution, and Bayesian inference is performed directly in the function space. GP is a set of any finite random variables that have a joint Gaussian distribution, and its properties are completely determined by the mean function and covariance function, that is,
[0041] m(x)=E[f(x)];
[0042] k(x,x′)=E[(f(x)-m(x))(f(x′)-m(x′))].
[0043] Where x, x'∈Rd are arbitrary random variables. Therefore, GP can be defined as f(x)~GP(m(x),k(x,x')). For the sake of symbolic simplicity, the data is usually preprocessed to make its mean function equal to 0.
[0044] For the regression problem, consider the following model:
[0045] y=f(X)+ε.
[0046] Where: x is the input vector, f is the function value, y is the observation value contaminated by additive noise, and further assume that the noise ε~N(0,σ n 2 ). We can get the prior distribution of the observed value y as
[0047]
[0048] And the joint prior distribution of the observed value y and the predicted value f* is
[0049]
[0050] Where: K(x,x)=Kn=(k ij ) is a symmetric positive definite covariance matrix of order n×n, and the matrix element kij=k(x i ,x j ) is used to measure x i and x j The correlation between K(X,x * )=K(X,x * ) T The test point x * The n×1 order covariance matrix between the input X of the training set; K(x * ,x* ) is the test point x * Its own covariance; I n is the n-dimensional identity matrix.
[0051] From this, we can calculate the posterior distribution of the predicted value f* as
[0052]
[0053] in
[0054]
[0055]
[0056] but That is the test point x * Corresponding predicted value f * The mean and variance of .
[0057] With the known training set and the prior distribution of the dependent variable, the posterior distribution of the predicted point can be obtained. However, in the application of machine learning, the prior distribution of the dependent variable is not known at the beginning. In order to use the Gaussian process for effective prediction, after obtaining the training set, the mean function and covariance function that conform to the training set samples must be selected. As the training set is continuously obtained, the process continues, and the mean function and covariance function continue to change. This process is called Gaussian process training.
[0058] For the fuzzy prior information, the covariance function is parameterized by the hyperparameter θ. The optimal hyperparameter is obtained by the maximum likelihood method.
[0059] Before training, the difference between the mean function and the training set data can be described by a smooth function with Gaussian noise. We do not need to define the exact parameters of this smooth function. We hope to infer all hyperparameters based on the known training set. To this end, calculate the probability of a training set sample appearing given a certain hyperparameter. Assume that the distribution of the dependent variable of the training set satisfies the Gaussian distribution. Therefore, θ is obtained by solving the following optimization problem
[0060] θ=arg min θ L(θ).
[0061] First, the negative log-likelihood L(θ) = -logp(y|X,θ) of the conditional probability of the training sample is established, and its partial derivative with respect to the hyperparameter θ is calculated; then, the partial derivative is minimized using optimization methods such as the conjugate gradient method and Newton's method to obtain the optimal solution for the hyperparameter. Here, the negative log-likelihood function L(θ) and its partial derivative with respect to the hyperparameter θ are as follows:
[0062]
[0063] in
[0064]
[0065] In summary, the training process of Gaussian process regression mainly includes two steps: 1) determining the mean function and covariance function; 2) determining the hyperparameters through the maximum likelihood method.
[0066] This embodiment uses Gaussian process regression as a spatial radiation intensity distribution estimation algorithm. In order to predict the radiation intensity value of any point in space, it is necessary to construct a Gaussian process suitable for the gamma radiation distribution.
[0067] First, construct the mean function. In the search area, the location of the radiation source is unknown. Before searching, it is assumed that the radiation source is at the point (x0, y0), and the γ radiation intensity distribution function in an unobstructed environment is used as the mean function of the Gaussian process:
[0068]
[0069] Where x i is the horizontal coordinate value of point i in the environment, y i is the ordinate value of point i in the environment, and I0 is the radiation intensity distribution in the environment. In this function, (x0, y0) is the hyperparameter of the mean function, and the initial value needs to be given before using Gaussian process regression.
[0070] Next, construct the covariance function. Commonly used covariance functions include RBF covariance function
[0071]
[0072] Matern covariance function
[0073]
[0074] Exponential covariance function
[0075]
[0076] Rational Quadratic Covariance Function
[0077]
[0078] Among them, the hyperparameter d is the dimension of the input, variance is the variance, lengthscale is the length scale, r is the Euclidean distance between the input points x and x′, and ν is a positive real number called the smoothness parameter.
[0079] The initial hyperparameters of the covariance function are set to the default values in the pykrige library. Different covariance functions are used to perform Gaussian process regression based on the training set shown in the figure below to obtain the radiation intensity distribution prediction map of the entire environment. The results are as follows: Figures 4 to 8 As shown, Figure 4 is the radiation intensity distribution prediction diagram of the RBF covariance function, Figure 5 is the radiation intensity distribution prediction diagram of the Matern covariance function when ν = 3, Figure 6 is the radiation intensity distribution prediction diagram of the Matern covariance function when ν = 5, Figure 7 It is the radiation intensity distribution prediction diagram of the Exponential covariance function. Figure 8 It is the radiation intensity distribution prediction diagram of the Rational Quadratic covariance function.
[0080] Will Figures 4 to 8 The MSE values of each image and the simulated radiation field are calculated, and the results are shown in Table 1. The smaller the MSE, the better the reconstruction effect.
[0081] Table 1 Mean square error between the predicted graph and the true graph of different Gaussian process regression covariance functions
[0082]
[0083] According to Table 1 and Figures 4 to 8 It can be seen that the Rational Quadratic (RQ) covariance function is the closest to the true field strength distribution in estimating the radiation field distribution, so the RQ covariance function is selected for Gaussian process regression. Next, determine the initial hyperparameter value of the RQ covariance function to completely define the Gaussian process used for regression. There are three hyperparameters in the RQ covariance function: variance, lengthscale, and r. First, determine the lengthscale. Fix variance = 8.2, r = 100, change the value of lengthscale to perform Gaussian process regression, calculate the similarity between the Gaussian process regression prediction graph and the true field strength distribution graph for different lengthscale values, and obtain the relationship curve between lengthscale and MSE value as shown below: Fig. 9 As shown, Fig. 9 It is the relationship curve between lengthscale and MSE value. It can be seen that when lengthscale = 30, Gaussian process regression has the best effect, so the initial lengthscale = 30 is determined.
[0084] Next, we determine variance, fix lengthscale=30, r=100, change the variance value to perform Gaussian process regression, calculate the similarity between the Gaussian process regression prediction map and the actual field strength distribution map for different variance values, and obtain the relationship curve between variance and MSE value as shown in the figure: Fig.10 As shown, Fig.10 This is the relationship curve between variance and MSE value. It can be seen that when variance ≥ 1, there is no significant difference in the Gaussian process regression effect, and the initial variance = 4 is selected.
[0085] Finally, r is determined, lengthscale = 30, variance = 4 is fixed, and the value of r is changed to perform Gaussian process regression. The similarity between the Gaussian process regression prediction map and the actual field strength distribution map with different r values is calculated, and the relationship curve between r and MSE value is obtained as shown in the figure. Fig.11 As shown, Fig.11 It is the relationship curve between r and MSE value. From the above analysis, we can see that the Gaussian process regression effect is best when r = 4, so the initial r = 4 is determined. Therefore, the mean function of the training process of Gaussian process regression is as follows:
[0086]
[0087] In the formula, x i is the horizontal coordinate value of point i in the source area to be found, y i is the ordinate value of point i in the source area to be found, x0 is the given assumed abscissa value of the radiation source, y0 is the given assumed ordinate value of the radiation source, and I0 is the radiation intensity distribution in the environment.
[0088] Moreover, the covariance function of the training process of Gaussian process regression is as follows:
[0089]
[0090] Where variance is the variance, lengthscale is the length scale, r is the Euclidean distance between the input points x and x′, variance, lengthscale and r are all given initial values, and the value of power is 1.
[0091] Moreover, the given initial value of Variance is greater than or equal to 1, the given initial value of lengthscale is 30, and the given initial value of r is 4.
[0092] In some embodiments, the host computer system generates an environmental map model of the source area to be found, including: generating an environmental point cloud map and a grid map of the source area to be found by laser scanning.
[0093] For example, in an outdoor environment with obstacles of size 18.2m×10.2m, using a 4.0E+04 (neutron / s) 241 The Am-Be source has been used for source search experiments. The environment point cloud generated by laser scanning is shown in the figure below. Fig.12 As shown in Figure 2, the raster map generated by laser scanning is as follows: Fig.13 shown.
[0094] The specific parameter settings of the source-seeking task are shown in Table 2. To reduce the impact of randomness, the robot stays at a new position for 5 seconds each time it arrives at the position, and the average count rate within these 5 seconds is obtained. Since the neutron source has a low emission rate, the neutron count rate after PSD screening is too low, so the unscreened total photon count rate is used as the radiation intensity value input in the source-seeking task (that is, the electrical pulse signal is converted from the sum of the radiation intensity values of the received neutrons and photon rays).
[0095] Table 2 Parameter settings of the path planning algorithm for the physical source search experiment
[0096]
[0097] The whole machine detection system of the source-seeking robot is 241 The pulse amplitude spectrum and PSD discrimination spectrum generated by Am-Be source are as follows Fig.14 As shown, Fig.14 The pulse amplitude spectrum and PSD spectrum measured in the source search mission. The upper part of the figure is the pulse amplitude spectrum, and the lower part is the PSD discrimination spectrum, where the left side is the γ peak and the right side is the neutron peak. It can be seen that the source search robot can effectively distinguish the γ source from the neutron source in the source search mission, which helps to determine the type of radioactive source.
[0098] Fig.15 This is the actual path diagram of the neutron source homing robot. Fig.16 This is the radiation hotspot distribution map reconstructed by the neutron source search robot. The red dot in the figure is the predicted location of the radiation source, and the white dot is the actual location of the radiation source. The robot found the radiation source after moving 79 steps. The coordinates of the actual location of the radiation source in the grid coordinate system are (62,20), and the predicted coordinates of the radiation source are (64,20). The grid side length unit is 20cm, that is, the predicted location of the radiation source differs from the actual location by 40cm, which verifies the effectiveness of the source search algorithm.
[0099] The neutron source locating robot of this embodiment may be composed of a plastic scintillator detector 210 , an electronics system 220 , a robot platform 200 , and the like. Fig.17 This is a top view of the installation of the neutron source locating robot of this embodiment.
[0100] The plastic scintillator detector model is EJ276, the scintillator size is 2 inches in diameter and 2 inches in length, the photomultiplier tube model is R7724, the diameter is 2 inches, and the recommended voltage is -1200V. The plastic scintillator detector and the electronics system are fixed in the robot platform. The plastic scintillator detector receives -1200V high-voltage power supply from the electronics system through the SHV interface, and outputs pulse signals to the electronics system through the BNC signal line from the ANODE signal outlet.
[0101] The electronics system is divided into a main amplifier module and a conditioning module. The main amplifier module is mainly divided into two parts: a high-voltage power supply module and a signal conditioning circuit. The high-voltage power supply module is controlled by the FPGA on the signal acquisition and processing module to step the DAC output of the corresponding amplitude control voltage (0-5V) and input it to its Vadj terminal to generate the high-voltage DC bias (about -1200V) required by the scintillator detector. The signal conditioning circuit buffers and isolates the electrical pulse signal generated by the detector under high-voltage bias through a voltage follower circuit to obtain an analog signal that meets the ADC input conditions.
[0102] The conditioning module is divided into a clock generation module and a high-speed data acquisition module. The clock module is configured through the PS end of the FPGA to generate synchronous multi-channel differential clock inputs. The high-speed data acquisition module interface is configured and interconnected through the PL end of the FPGA.
[0103] The electronics system receives +24V DC power supply from the robot platform through the J30JZ interface, communicates with the robot platform through the 100M Ethernet interface, and realizes differential transmission in the form of digital code stream signals. The transmission content includes collected data and control instructions.
[0104] The robot platform adopts a universal double-wheeled robot mobile chassis, a planetary swing arm wheel structure, and is driven by four 1000W low-voltage servo motors, and all wheels are active wheels. The power lithium battery has a capacity of 48V105Ah and is equipped with a battery protection system. The brake system uses the driver electric brake in conjunction with the mechanical brake, and the swing arm wheel structure allows the chassis to climb 35° under load conditions. The platform is equipped with an autonomous navigation perception kit based on 16-line laser radar, IMU, and Beidou RTK, and an on-board computing unit to realize an autonomous navigation system with multi-sensor fusion. Sealing rubber rings are made at the joints of the robot interface; the sheet metal shell has good impact resistance; to protect the internal electronic components from damage by the external environment. The chassis material is Q235B, and the wheel spindle is made of 45 steel.
[0105] Regarding the embodiments of the present application, it should also be noted that, in the absence of conflict, the embodiments of the present application and the features therein can be combined with each other to obtain new embodiments.
[0106] The above are only specific implementation methods of the present application, but the protection scope of the present application is not limited thereto. The protection scope of the present application shall be based on the protection scope of the claims.
Claims
1. A neutron source locating robot, characterized in that: include: A robot platform, the robot platform comprising a movable chassis, the movable chassis being used to move in the area to be searched for a source; A plastic scintillator detector, fixed to the robot platform, for receiving neutrons and photon rays when the movable chassis moves, and converting the received neutrons and photon rays into electrical pulse signals; An electronics system, fixed to the robot platform, for receiving the electrical pulse signal converted by the plastic scintillator detector and processing the electrical pulse signal to obtain a processed signal; The host computer system is used to generate an environmental map model of the area to be found, and plan the walking path of the movable chassis according to the environmental map model and the real-time positioning of the robot platform, and control the movable chassis to move along the walking path. The host computer system is also used to receive the processed signal, and estimate the position of the radiation source according to the processed signal, and reconstruct the radiation hotspot distribution map of the area to be found, and is used to determine whether the radiation source is a neutron source or a gamma source.
2. The neutron source locating robot according to claim 1, characterized in that: The Gaussian process regression method is used to reconstruct the radiation hotspot distribution map of the source-to-be-searched area.
3. The neutron source locating robot according to claim 2, characterized in that: The mean function of the training process of the Gaussian process regression is as follows: In the formula, x i is the horizontal coordinate value of point i in the source area to be found, y i is the ordinate value of point i in the source area to be found, x0 is the given assumed abscissa value of the radiation source, y0 is the given assumed ordinate value of the radiation source, and I0 is the radiation intensity distribution.
4. The neutron source locating robot according to claim 2, characterized in that: The covariance function of the training process of the Gaussian process regression is as follows: Where variance is the variance, lengthscale is the length scale, and r is the input points x and x. ′ The Euclidean distance between them, variance, lengthscale and r are all given initial values, and the power value is 1.
5. The neutron source locating robot according to claim 4, characterized in that: The given initial value of Variance is greater than or equal to 1, the given initial value of lengthscale is 30, and the given initial value of r is 4.
6. The neutron source locating robot according to claim 1, characterized in that: The electric pulse signal is converted from the sum of the radiation intensity values of the received neutron and photon rays.
7. The neutron source locating robot according to claim 1, characterized in that: Also includes: An internal control system is installed on the robot platform and is used to provide low voltage power to the electronics system so that the electronics system provides working high voltage to the plastic scintillator detector.
8. The neutron source locating robot according to claim 7, characterized in that: The internal control system is also used to receive the processed signal, and the internal control system communicates wirelessly with the host computer system so that the host computer system receives the processed signal.
9. The neutron source locating robot according to claim 1, characterized in that: The host computer system is also used for online display of the pulse amplitude spectrum of the processed signal.
10. The neutron source locating robot according to claim 1, characterized in that: The host computer system generates an environmental map model of the area to be found, including: generating an environmental point cloud map and a grid map of the area to be found by laser scanning.
