Ray digital twin reference radiation field establishment method

Abstract

The embodiment of the invention provides a method for establishing a ray digital twin reference radiation field. The method comprises the following steps: establishing a Monte Carlo model of a ray measurement standard reference radiation field; and establishing a neural network prediction model based on a deep neural network algorithm according to a simulation result of the Monte Carlo model. And acquiring the ray radiation dose rate of the specified coordinate point according to the neural network prediction model, and establishing a ray digital twin reference radiation field. According to the establishment method of the ray digital twin reference radiation field, the radiation dose rate prediction accuracy effect is good.

Description

Technical Field

[0001] The present application relates to the technical field of ray digital twin reference radiation fields, and in particular to a method for establishing a ray digital twin reference radiation field. Background Art

[0002] X-ray and gamma-ray metrology standard reference radiation fields are mainly used for the reproduction and transfer of the values ​​of X-ray and gamma-ray related physical quantities, and play an important role in metrology assurance.

[0003] In recent years, with the increasing demand for the use of X-rays and gamma rays in the field of ionizing radiation metrology, the pressure on related radiation dose rate monitoring has also been increasing.

[0004] However, the radiation dose rate prediction accuracy of the X-ray and gamma-ray metrology standard reference radiation fields in the related art is relatively poor. Summary of the invention

[0005] In view of this, the main purpose of the embodiments of the present application is to provide a method for establishing a ray digital twin reference radiation field with good radiation dose rate prediction accuracy.

[0006] To achieve the above purpose, the technical solution of the embodiment of the present application is implemented as follows:

[0007] The present application provides a method for establishing a ray digital twin reference radiation field, comprising the following steps:

[0008] Establish a Monte Carlo model of the reference radiation field of the radiation metrology standard;

[0009] According to the simulation results of the Monte Carlo model, a neural network prediction model is established based on a deep neural network algorithm;

[0010] The ray radiation dose rate of the specified coordinate point is obtained according to the neural network prediction model, and the ray digital twin reference radiation field is established.

[0011] In one embodiment, the establishment method further includes:

[0012] Establishing a three-dimensional model of the reference radiation field of the radiation metrology standard;

[0013] The three-dimensional model is used as a layer background, and the radiation dose rate of each coordinate point is made to correspond to the three-dimensional model.

[0014] In one embodiment, the Monte Carlo model of establishing a reference radiation field of a ray metrology standard includes:

[0015] A Monte Carlo model is established for the experimental facilities of the radiation metrology standard reference radiation field; wherein the experimental facilities include typical instruments.

[0016] In one embodiment, the material type of the housing of the typical instrument is aluminum or polytetrafluoroethylene.

[0017] In one embodiment, after establishing the Monte Carlo model of the radiation metrology standard reference radiation field, the establishing method further includes:

[0018] A dose determination experiment of the radiation measurement standard reference radiation field is carried out, and the simulation results of the Monte Carlo model are compared with the experimental results of the dose determination experiment to verify the effectiveness of the Monte Carlo model.

[0019] In one embodiment, the establishing of a neural network prediction model based on a deep neural network algorithm according to the simulation results of the Monte Carlo model specifically includes:

[0020] Performing data preprocessing on the simulation results of the Monte Carlo model to establish a neural network data set;

[0021] The computational model is trained, verified and tested based on the neural network data set to establish the neural network prediction model.

[0022] In one embodiment, after establishing the neural network prediction model, the establishment method further includes:

[0023] The prediction result of the neural network prediction model is obtained according to multiple characteristic values; wherein the multiple characteristic values ​​include radiation quality, tube voltage, tube current, x-coordinate, y-coordinate and z-coordinate.

[0024] In one embodiment, the establishment method further includes:

[0025] The prediction result is corrected according to the correction factor, and the error between the dose rate of the ray digital twin reference radiation field and the agreed true value is calculated by the error calculation formula.

[0026] In one embodiment, the correction factor includes temperature, humidity, atmospheric pressure and the reading of the monitoring ionization chamber, and the error calculation formula is:

[0027]

[0028] Wherein, E is the error of the dose rate prediction value of the reference radiation field of the ray digital twin, dimensionless; y is the predicted value of the real-time dose rate prediction model at the reference point of the ray digital twin reference radiation field, nGy / s; C T,P is the correction factor for air temperature and pressure; C h is the correction factor for the difference in relative humidity between the reference condition and the measurement condition, which is determined by an empirical formula based on the response of the transfer standard device as a function of relative humidity, C hSmall, so for the generally achievable relative humidity range, assume that C h =1; correction factor for the difference in the supervisory ionization chamber readings between reference conditions and measurement conditions; N is the agreed true value at the reference point of the ray reference radiation field, nGy / s.

[0029] In one embodiment, the establishment method further includes:

[0030] The normalized and corrected prediction results are used to perform point cloud rendering of the ray radiation dose rate based on visualization technology to achieve visualized monitoring of the ray radiation dose rate in the three-dimensional model space.

[0031] The embodiment of the present application provides a method for establishing a digital twin reference radiation field of radiation, comprising the following steps: establishing a Monte Carlo model of a radiation metrology standard reference radiation field. According to the simulation results of the Monte Carlo model, a neural network prediction model is established based on a deep neural network algorithm. According to the neural network prediction model, the radiation dose rate of the specified coordinate point is obtained, and a digital twin reference radiation field of radiation is established. Therefore, on the one hand, through the precise simulation of the Monte Carlo model and the powerful learning ability of the deep neural network algorithm, the radiation dose rate can be predicted more accurately, thereby greatly improving the accuracy of the radiation dose rate prediction. On the other hand, by establishing a digital twin reference radiation field of radiation, real-time dose rate prediction of each point in the radiation metrology standard reference radiation field space can be achieved, which can meet the radiation safety monitoring, personnel training, scientific research and efficient value transfer of the radiation metrology standard reference radiation field, and can improve the accuracy of the radiation field reference value and promote the digital transformation of the metrology standard device. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 This is a flowchart of a method for establishing a ray digital twin reference radiation field according to an embodiment of the present application. DETAILED DESCRIPTION

[0033] In the related technologies, the traditional dose rate monitoring method of X-ray and gamma-ray metrology standard reference radiation fields has some difficulties such as insufficient environmental monitoring, non-intuitive radiation dose information, and incomplete effective measurement and evaluation of staff, which makes it difficult to ensure the efficient implementation of the value transfer work. Therefore, X-ray and gamma-ray metrology standard reference radiation fields are in urgent need of digital and intelligent transformation.

[0034] Among the related technologies, the rapid development of digital twins and artificial intelligence technologies has provided important technical support for the establishment of digital twin reference radiation fields for X-rays and gamma-rays. The establishment of digital twin reference radiation fields can realize real-time dose rate prediction and visual monitoring of each point in the metrological standard reference radiation field space, allowing the "invisible and intangible" reference radiation field to intuitively present the real-time radiation dose rate.

[0035] An embodiment of the present application provides a method for establishing a ray digital twin reference radiation field, see Figure 1 , the method for establishing the radiation dose rate includes the following steps:

[0036] S1: Establish a Monte Carlo model of the reference radiation field of the X-ray metrology standard.

[0037] S2: According to the simulation results of the Monte Carlo model, a neural network prediction model is established based on the deep neural network algorithm.

[0038] S3: Obtain the radiation dose rate of the specified coordinate point according to the neural network prediction model, and establish a radiation digital twin reference radiation field.

[0039] Specifically, the type of rays is not limited.

[0040] For example, the radiation is an X-ray.

[0041] For another example, the ray is a gamma ray.

[0042] The Monte Carlo model refers to a computational method model based on probability statistics theory.

[0043] The modeling content within the Monte Carlo model is determined based on actual conditions.

[0044] Exemplarily, a Monte Carlo model is established for an experimental facility of a standard reference radiation field of radiation metrology, wherein the experimental facility includes typical instruments.

[0045] Specifically, when it is a Monte Carlo model of the X-ray metrology standard reference radiation field, the modeling content of the Monte Carlo model includes the X-ray irradiation device, shielding body, translation stage, guide rail and wall. Among them, typical instruments and meters at least include X-ray environmental radiation dosimeter, X-ray personal dosimeter, direct reading personal dosimeter, spherical ionization chamber, thermoluminescent personal dosimeter and water tank.

[0046] When it is a Monte Carlo model of the gamma-ray metrology standard reference radiation field, the modeling content of the Monte Carlo model includes the gamma-ray irradiation device, shielding body, cement table, calibration table, guide rails and walls. Among them, typical instruments and meters at least include gamma-ray environmental radiation dosimeter, gamma-ray personal dosimeter, direct reading personal dosimeter, spherical ionization chamber, thermoluminescent personal dosimeter and water tank.

[0047] It should be noted that in order to evaluate the scattered radiation contribution of the instruments to be tested to the radiation dose rate in the X-ray reference radiation field, it is necessary to establish a variety of typical instruments to be calibrated and verified in the X-ray reference radiation field. By placing the typical instruments on the translation stage according to the verification requirements, the position of the translation stage and the calibrated instrument can be moved arbitrarily in the horizontal direction, simulating the steps of changing the agreed true value by changing the distance between the typical instrument to be tested and the irradiation device during calibration.

[0048] It should be noted that the modeling content in the Monte Carlo model is proportional modeling based on the drawings of the reference radiation field of the ray metrology standard.

[0049] The neural network prediction model refers to a model used to predict the radiation dose rate of rays based on a deep neural network algorithm. By learning the intrinsic relationship between a large amount of data simulated by the Monte Carlo model, it can predict the real-time radiation dose rate of rays at specified coordinate points.

[0050] The X-ray digital twin reference radiation field refers to the digital mapping of the X-ray metrology standard reference radiation field using digital technology. It can reflect the status of the radiation field (such as X-ray radiation dose rate and other information) in real time, realize comprehensive and accurate monitoring and analysis of the radiation field, and make the situation of the radiation field intuitive and visual.

[0051] The embodiment of the present application provides a method for establishing a digital twin reference radiation field of radiation, comprising the following steps: establishing a Monte Carlo model of a radiation metrology standard reference radiation field. According to the simulation results of the Monte Carlo model, a neural network prediction model is established based on a deep neural network algorithm. According to the neural network prediction model, the radiation dose rate of the specified coordinate point is obtained, and a digital twin reference radiation field of radiation is established. Therefore, on the one hand, through the precise simulation of the Monte Carlo model and the powerful learning ability of the deep neural network algorithm, the radiation dose rate can be predicted more accurately, thereby greatly improving the accuracy of the radiation dose rate prediction. On the other hand, by establishing a digital twin reference radiation field of radiation, real-time dose rate prediction of each point in the radiation metrology standard reference radiation field space can be achieved, which can meet the radiation safety monitoring, personnel training, scientific research and efficient value transfer of the radiation metrology standard reference radiation field, and can improve the accuracy of the radiation field reference value and promote the digital transformation of the metrology standard device.

[0052] In one embodiment, establishing a ray digital twin reference radiation field includes:

[0053] Establish a three-dimensional model of the reference radiation field of the radiation metrology standard.

[0054] With the three-dimensional model as the layer background, the radiation dose rate of each coordinate point is matched with the three-dimensional model.

[0055] Therefore, the three-dimensional modeling technology is used to reconstruct the metrology standard laboratory. By corresponding each coordinate point in the neural network prediction model to each coordinate point in the three-dimensional model one by one, a digital twin reference radiation field of the ray is established. The originally abstract ray radiation dose rate information can be presented intuitively, and real-time prediction and visual monitoring of the ray radiation dose rate can be achieved. The staff can observe the dose rate at each position of the radiation field in real time through the visual interface, which is convenient for timely discovery of abnormally high-dose areas, so as to take corresponding protective measures or adjust the experimental operation to ensure the safety of personnel and the normal progress of the experiment.

[0056] Specifically, the method of establishing the three-dimensional model is not limited.

[0057] For example, a 3D scanner is used to scan the radiation metrology standard laboratory from multiple perspectives to achieve 3D data collection, covering all 3D data in the scene. A 3D model of the radiation metrology standard reference radiation field is established based on the 3D point cloud data to generate a point cloud model, and then the point cloud data is processed to remove noise and fill in the missing parts to generate the final 3D model, thus realizing the reconstruction of the metrology standard laboratory.

[0058] Using the three-dimensional model as the layer background and making the radiation dose rate of each coordinate point correspond to the three-dimensional model means that the radiation dose rate of each coordinate point predicted by the neural network prediction model corresponds to the radiation dose rate of the corresponding coordinate point in the three-dimensional model.

[0059] It should be noted that the neural network prediction model and the three-dimensional model together constitute the radiation digital twin reference radiation field. The radiation dose rate of a certain coordinate in the radiation digital twin reference radiation field is the radiation dose rate of the corresponding coordinate in the corresponding radiation metrology standard laboratory.

[0060] In one embodiment, the material type of the shell of a typical instrument is aluminum or polytetrafluoroethylene. Therefore, the stability of aluminum and polytetrafluoroethylene materials ensures the structural integrity of the instrument in the radiation field, and will not be deformed or damaged due to radiation or other environmental factors, thereby ensuring that the internal measuring elements are always in a stable working state, and improving the accuracy and reliability of the measurement results.

[0061] In one embodiment, after establishing the Monte Carlo model of the radiation metrology standard reference radiation field, the establishing method further includes:

[0062] The dose determination experiment of the reference radiation field of the radiation metrology standard was carried out, and the simulation results of the Monte Carlo model were compared with the experimental results of the dose determination experiment to verify the effectiveness of the Monte Carlo model. Therefore, by verifying the effectiveness of the Monte Carlo model to ensure the accuracy of the Monte Carlo model simulation results, the performance of the neural network prediction model was ensured.

[0063] Specifically, a radiation metrology reference radiation field dose determination experiment was carried out based on a graphite cavity ionization chamber and compared with the simulation results of the Monte Carlo model. When the error between the simulation results of the Monte Carlo model and the agreed true value is less than or equal to 15%, the Monte Carlo model is considered valid.

[0064] In one embodiment, according to the simulation results of the Monte Carlo model, a neural network prediction model is established based on a deep neural network algorithm, specifically including:

[0065] The simulation results of the Monte Carlo model are preprocessed to establish a neural network data set.

[0066] The computational model is trained, verified and tested based on the neural network dataset to establish a neural network prediction model.

[0067] Therefore, by preprocessing the simulation results of the Monte Carlo model to obtain more accurate data, and training based on accurate data, the neural network prediction model can more accurately predict the radiation dose rate at different positions and conditions in the reference radiation field of the X-ray metrology standard, reduce prediction errors, and improve the reliability of prediction results.

[0068] A neural network dataset refers to a data set obtained after data preprocessing and used to train, verify, and test computational models.

[0069] Training a computational model refers to using most of the data in a neural network dataset (usually about 70%-80%) to adjust the parameters of the neural network model (such as the connection weights and biases between neurons) so that the model can learn the intrinsic relationship between input features and output targets, and continuously optimize the model's predictive ability to minimize prediction errors.

[0070] Validating a computational model means using a portion of independent data (about 10%-15%) during the training process to evaluate the performance of the model on unseen data and monitor whether the model is overfitting (the model performs well on the training data, but the performance deteriorates on new data). Through the performance indicators of the validation set (such as mean absolute error, etc.), the model's hyperparameters (such as the number of hidden layers, learning rate, etc.) can be adjusted to optimize the model structure and training process, and improve the model's generalization ability.

[0071] Testing the calculation model means using the remaining independent data (about 10%-15%) to evaluate the performance of the final model after training and verification, and obtain accurate performance indicators of the model in actual application scenarios. The test results reflect the accuracy and reliability of the model's prediction of new data, and are an important basis for judging whether the model meets actual needs.

[0072] By training the computational model, the model with the best performance is obtained as the neural network prediction model of the ray digital twin reference radiation field, so as to predict the real-time radiation dose rate in the space of the digital twin reference radiation field.

[0073] It should be noted that by training the computational model with the help of the neural network data set, multiple prediction models can be obtained, among which the prediction model with the best prediction effect is determined as the neural network prediction model.

[0074] The performance of multiple computational models was evaluated by mean absolute error (MAE).

[0075] In one embodiment, after establishing the neural network prediction model, the establishment method further includes:

[0076] The prediction result of the neural network prediction model is obtained according to multiple characteristic values; wherein the multiple characteristic values ​​include radiation quality, tube voltage, tube current, x-coordinate, y-coordinate and z-coordinate.

[0077] Therefore, by comprehensively considering multiple eigenvalues, the neural network prediction model can comprehensively describe the state of the radiation field and thus predict the radiation dose rate more accurately.

[0078] Specifically, the eigenvalue refers to a variable used to describe information such as the state and position of the reference radiation field of the radiation metrology standard.

[0079] The origin coordinates of the ray digital twin reference radiation field are the center of the irradiation device outlet.

[0080] Six eigenvalues ​​are input into the neural network prediction model, and the output of the model is the predicted value of the radiation dose rate at the specified coordinate point under the input conditions.

[0081] In one embodiment, the establishing method further includes:

[0082] The prediction result is corrected according to the correction factor, and the error between the dose rate of the reference radiation field of the digital twin and the agreed true value is calculated by the error calculation formula.

[0083] This can further improve the accuracy of the prediction of the ray digital twin reference radiation field.

[0084] Specifically, the correction factor refers to a coefficient used to adjust the prediction result to take into account the impact of various environmental factors and measurement conditions on the radiation dose rate measurement.

[0085] The error range is determined based on actual conditions.

[0086] For example, when the absolute value of the error between the dose rate of the ray digital twin reference radiation field and the agreed true value is less than or equal to 15%, the visual monitoring function of the ray digital twin reference radiation field can be realized.

[0087] In one embodiment, the correction factors include temperature, humidity, atmospheric pressure and the reading of the monitoring ionization chamber, and the error calculation formula is:

[0088]

[0089] Where E is the error of the dose rate prediction value of the reference radiation field of the digital twin of the ray, dimensionless; y is the predicted value of the real-time dose rate prediction model at the reference point of the reference radiation field of the digital twin of the ray, nGy / s; C T,P is the correction factor for air temperature and pressure; C h is the correction factor for the difference in relative humidity between the reference condition and the measurement condition, which is determined by an empirical formula for the response of the transfer standard device as a function of relative humidity, C h Small, so for the generally achievable relative humidity range, assume that C h =1; correction factor for the difference in monitoring ionization chamber readings between reference conditions and measurement conditions; N is the agreed true value at the reference point of the reference radiation field, nGy / s.

[0090] Specifically, C T,P The calculation formula is:

[0091] C T,P =(P0×T) / (P×T0)

[0092] Among them, P is the pressure of the air during measurement; T is the temperature of the air during measurement; P0 is the pressure under reference conditions, P0 = 101.3 kPa; T0 is the temperature under reference conditions, T0 = 293.15 K.

[0093] C s The calculation formula is:

[0094] C s =M s / N s

[0095] Among them, M s N is the predicted value of the reading of the monitoring ionization chamber by the reference radiation field of the ray digital twin; s is the reading of the monitoring ionization chamber under test conditions, calculated when M s The unit should be the same as N s The units remain consistent.

[0096] In one embodiment, the establishing method further includes:

[0097] The normalized and corrected prediction results are used for point cloud rendering of radiation dose rate based on visualization technology to realize visual monitoring of radiation dose rate in the three-dimensional model space.

[0098] Therefore, based on point cloud rendering, the changes in color and transparency are used to intuitively display the "invisible and intangible" radiation dose rate, so that the "invisible and intangible" radiation dose rate can be accurately and intuitively monitored, thereby improving the work efficiency of environmental monitoring, safety management, personnel training and scientific research of the radiation metrology standard reference radiation field, and improving the accuracy of the radiation field reference value, providing strong support for radiation workers to promptly detect abnormal changes in radiation dose, so as to take rapid measures and reduce potential safety risks.

[0099] In a specific embodiment, based on the neural network prediction model of the X-ray and gamma-ray reference radiation field, the real-time radiation dose rate of each point in the space of the X-ray and gamma-ray metrology standard laboratory is calculated. Before each prediction, the coordinates of each point evenly distributed in the laboratory are input into the model for calculation, and the calculation results are output. In the X-ray and gamma-ray metrology standard laboratory, graded coordinate point data is used, with a number of 2 million to 2.8 million. The coordinate points generated above are input into the neural network prediction model of the X-ray and gamma-ray reference radiation field in the form of a list, and the prediction results, i.e., the real-time radiation dose rate of the corresponding coordinate points, can be obtained. Based on visualization technology, the real-time dose rate of the X-ray and gamma-ray digital twin reference radiation field is visualized and monitored. In order to obtain the intensity value of the point cloud in the shader, the prediction results need to be normalized, and the prediction and normalization process takes up to about 30 seconds. Based on visualization technology, the point cloud rendering technology of the real-time dose rate of the digital radiation field is studied, and the visual display of the changes in color and transparency is realized to realize the visual monitoring of the real-time dose rate of the X-ray and gamma-ray digital twin reference radiation field.

[0100] In a specific embodiment, the real-time dose rate of the X-ray and gamma-ray digital twin reference radiation field is predicted by a computer processor.

[0101] Specifically, when using the X-ray and gamma-ray digital twin reference radiation field, first enter the login interface and enter the user name and password. After successful login, enter the calculation and correction interface, and the user enters the characteristic values ​​to be predicted, that is, the radiation quality, tube voltage and tube current of the X-ray digital twin reference radiation field. The computer processor can correct the predicted values ​​of the real-time dose rate prediction model of the X-ray and gamma-ray reference radiation field. The correction factors are the output values ​​of the temperature, atmospheric pressure, humidity and the monitored ionization chamber, among which the real-time temperature, humidity and atmospheric pressure need to be manually entered. In addition, the X-ray and gamma-ray digital twin reference radiation field is equipped with an interface for the X-ray and gamma-ray metrology standard reference radiation field monitored ionization chamber. After the monitored ionization chamber is connected to the electrometer, the electrometer is connected to the terminal of the computer processor, and the computer processor can obtain the reading of the monitored ionization chamber in real time, and then correct the predicted value of the neural network. After the above information is entered, calculation and correction are performed, and the X-ray and gamma-ray digital twin reference radiation field calls the X-ray and gamma-ray reference radiation field real-time dose rate prediction model to predict the real-time radiation dose rate of multiple coordinate points, and corrects the prediction results. After completion, the real-time radiation dose rate will be visualized and displayed in the three-dimensional model of the X-ray and gamma-ray reference radiation field. After each prediction of the X-ray and gamma-ray digital twin reference radiation field is completed, the prediction data will be automatically saved as historical data, and authorized users can view and delete it at any time.

[0102] In the description of the present application, the description with reference to the terms "in one embodiment", "in some embodiments", "in a specific embodiment", or "exemplary" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the embodiments of the present application. In the present application, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in a suitable manner in any one or more embodiments or examples. In addition, those skilled in the art may combine different embodiments or examples described in the present application and the features of different embodiments or examples without contradiction.

[0103] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application are included in the protection scope of the present application.

Claims

1. A method for establishing a ray digital twin reference radiation field, characterized in that: The steps include: Establish a Monte Carlo model of the reference radiation field of the radiation metrology standard; According to the simulation results of the Monte Carlo model, a neural network prediction model is established based on a deep neural network algorithm; The ray radiation dose rate of the specified coordinate point is obtained according to the neural network prediction model, and the ray digital twin reference radiation field is established.

2. The method for establishing a ray digital twin reference radiation field according to claim 1, characterized in that: The establishing of the ray digital twin reference radiation field comprises: Establishing a three-dimensional model of the reference radiation field of the radiation metrology standard; The three-dimensional model is used as a layer background, and the radiation dose rate of each coordinate point is made to correspond to the three-dimensional model.

3. The method for establishing a ray digital twin reference radiation field according to claim 1, characterized in that: The Monte Carlo model for establishing a reference radiation field of a ray metrology standard comprises: A Monte Carlo model is established for the experimental facilities of the radiation metrology standard reference radiation field; wherein the experimental facilities include typical instruments.

4. The method for establishing a ray digital twin reference radiation field according to claim 3, characterized in that: The material type of the housing of the typical instrument is aluminum or polytetrafluoroethylene.

5. The method for establishing a ray digital twin reference radiation field according to claim 1, characterized in that: After establishing the Monte Carlo model of the ray metrology standard reference radiation field, the establishing method further includes: A dose determination experiment of the radiation measurement standard reference radiation field is carried out, and the simulation results of the Monte Carlo model are compared with the experimental results of the dose determination experiment to verify the effectiveness of the Monte Carlo model.

6. The method for establishing a ray digital twin reference radiation field according to claim 2, characterized in that: According to the simulation results of the Monte Carlo model, a neural network prediction model is established based on a deep neural network algorithm, specifically including: Performing data preprocessing on the simulation results of the Monte Carlo model to establish a neural network data set; The computational model is trained, verified and tested based on the neural network data set to establish the neural network prediction model.

7. The method for establishing a ray digital twin reference radiation field according to claim 6, characterized in that: After establishing the neural network prediction model, the establishment method further includes: The prediction result of the neural network prediction model is obtained according to multiple characteristic values; wherein the multiple characteristic values ​​include radiation quality, tube voltage, tube current, x-coordinate, y-coordinate and z-coordinate.

8. The method for establishing a ray digital twin reference radiation field according to claim 7, characterized in that: The establishment method also includes: The prediction result is corrected according to the correction factor, and the error between the dose rate of the ray digital twin reference radiation field and the agreed true value is calculated by the error calculation formula.

9. The method for establishing a ray digital twin reference radiation field according to claim 8, characterized in that: The correction factors include temperature, humidity, atmospheric pressure and the reading of the monitoring ionization chamber, and the error calculation formula is: Wherein, E is the error of the dose rate prediction value of the reference radiation field of the ray digital twin, dimensionless; y is the predicted value of the real-time dose rate prediction model at the reference point of the ray digital twin reference radiation field, nGy / s; C T,P is the correction factor for air temperature and pressure; C h is the correction factor for the difference in relative humidity between the reference condition and the measurement condition, which is determined by an empirical formula based on the response of the transfer standard device as a function of relative humidity, C h Small, so for the generally achievable relative humidity range, assume that C h =1; correction factor for the difference in the supervisory ionization chamber readings between reference conditions and measurement conditions; N is the agreed true value at the reference point of the ray reference radiation field, nGy / s.

10. The method for establishing a ray digital twin reference radiation field according to claim 8, characterized in that: The establishment method also includes: The normalized and corrected prediction results are used to perform point cloud rendering of the ray radiation dose rate based on visualization technology to achieve visualized monitoring of the ray radiation dose rate in the three-dimensional model space.

Images