CN111667571A - Nuclear facility source item three-dimensional distribution rapid reconstruction method, device, equipment and medium - Google Patents
Nuclear facility source item three-dimensional distribution rapid reconstruction method, device, equipment and medium Download PDFInfo
- Publication number
- CN111667571A CN111667571A CN202010513588.6A CN202010513588A CN111667571A CN 111667571 A CN111667571 A CN 111667571A CN 202010513588 A CN202010513588 A CN 202010513588A CN 111667571 A CN111667571 A CN 111667571A
- Authority
- CN
- China
- Prior art keywords
- source item
- source
- dimensional distribution
- radiation field
- activity
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2200/00—Indexing scheme for image data processing or generation, in general
- G06T2200/04—Indexing scheme for image data processing or generation, in general involving 3D image data
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Geometry (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- Computer Graphics (AREA)
- Mathematical Physics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application discloses a method, a device, equipment and a medium for rapidly reconstructing three-dimensional distribution of nuclear facility source items, which comprise the following steps: adopting a Monte Carlo particle transport program to construct a source item geometric model and calculating a radiation field; extracting three-dimensional distribution data of space radiation field data and source item activity from the calculation result, constructing source item position matrixes with the same size according to source item distribution, generating deep learning training samples, and constructing and training a deep neural network model; converting the actual radiation field dose matrix and the source item position matrix, inputting the converted radiation field dose matrix and the source item position matrix into a trained deep neural network model, and predicting three-dimensional distribution data of source item activity; and verifying the error between the output result and the actual value. According to the method and the device, a proper deep neural network is obtained through construction and parameter adjustment, rapid reconstruction of three-dimensional distribution of source item activity not limited to a specific physical model is achieved, and the source item activity value at any specified position can be rapidly obtained through deep learning intelligent inversion reconstruction.
Description
Technical Field
The invention relates to the field of radiation protection and nuclear safety, in particular to a method, a device, equipment and a medium for quickly reconstructing three-dimensional distribution of nuclear facility source items.
Background
The characteristics of radioactive source items are very important precondition for decommissioning or processing nuclear facilities, in the radiation scene of complex source items of actual nuclear facilities, because the activity and three-dimensional distribution of the source items are often unknown and have large uncertainty, and limited by measurement means, it is very difficult to directly measure the composition and quantity of radioactive substances, and the measurement result is only the appearance of the source items in a certain part. Therefore, the source item data is generally obtained by adopting an analytical calculation mode. The three-dimensional radiation field is a database reflecting the real external irradiation distribution in a nuclear facility, required radiation field data can be obtained through measurement, and the activity of a source item is reversely calculated by using the measurement value of the radiation field dosage rate. By analyzing the three-dimensional radiation field, the position of a radiation hot spot can be determined, the equivalent activity three-dimensional distribution condition of radioactive substances in equipment or pipelines can be reconstructed, effective shielding measures are further established, and on-site refined radiation protection optimization analysis is realized.
At present, the source item inversion is carried out by adopting a physical fitting method or a numerical value interpolation method, only the radioactivity activity calculation of a simple fixed model can be processed, the physical correlation requirement between input and output is high, and the three-dimensional source item activity distribution under the conditions of complex source items and strong anisotropy distribution of nuclear facilities cannot be well reconstructed.
Therefore, how to realize intelligent and rapid reconstruction of activity distribution of three-dimensional source items under the conditions of complex source items and strong anisotropic distribution of nuclear facilities is a technical problem to be solved urgently by technical personnel in the field.
Disclosure of Invention
In view of this, the present invention provides a method, an apparatus, a device and a medium for rapidly reconstructing three-dimensional distribution of nuclear facility source items, which can rapidly obtain an activity value of a source item at any specified position through deep learning intelligent inversion reconstruction. The specific scheme is as follows:
a nuclear facility source item three-dimensional distribution rapid reconstruction method comprises the following steps:
adopting a Monte Carlo particle transport program to construct a source item geometric model and calculating a radiation field;
extracting three-dimensional distribution data of space radiation field data and source item activity from the calculation result, constructing source item position matrixes with the same size according to source item distribution, generating deep learning training samples, and constructing and training a deep neural network model;
converting the actual radiation field dose matrix and the source item position matrix, inputting the converted actual radiation field dose matrix and the source item position matrix into the trained deep neural network model, and predicting three-dimensional distribution data of source item activity;
and verifying the error between the output three-dimensional distribution data result of the source item activity and the actual value.
Preferably, in the method for rapidly reconstructing three-dimensional distribution of source items of a nuclear facility according to the embodiment of the present invention, a monte carlo particle transport program is used to construct a geometric source item model, and the calculation of a radiation field and a geometric source item spatial distribution matrix includes:
according to the distribution condition of the source items, carrying out grid division on the source items;
according to the source items divided into a plurality of areas, a Monte Carlo particle transport program is adopted to construct a source item geometric model;
generating source item data with certain distribution by adopting a universal source SDEF card of a Monte Carlo particle transport program;
setting the region except the source item as a non-interested region, setting a cubic space with a set size outside the source item, dividing the cubic space into networks, and performing three-dimensional distribution statistics of the activity of the source item on the divided cubic space;
randomly sampling a plurality of groups of different source item geometric parameters, generating Monte Carlo calculation files in batches, and calling a Monte Carlo particle transport program to calculate the radiation field.
Preferably, in the method for rapidly reconstructing three-dimensional distribution of nuclear facility source items provided in the embodiment of the present invention, before constructing the deep neural network model, the method further includes:
performing three-dimensional gridding on the spatial radiation field data and the three-dimensional distribution data of the source item activity;
coarsening or thinning the three-dimensional distribution data of the space radiation field data and the source item activity after three-dimensional gridding;
taking the processed space radiation field and the constructed source item position matrix as input training samples;
taking the processed three-dimensional distribution data of the activity of the source item as an output training sample;
and adding Gaussian white noise to the output training sample.
Preferably, in the method for rapidly reconstructing three-dimensional distribution of nuclear facility source items provided in the embodiment of the present invention, constructing a deep neural network model specifically includes:
constructing a deep neural network model according to three parts of feature extraction, advanced feature learning and feature combination; the feature extraction part is composed of a combination of a plurality of convolutional layers and downsampling layers and a combination of a cross layer, an upsampling layer and a convolutional layer, the advanced feature learning part is composed of a plurality of fully-connected layers, and the feature combination part is composed of a product calculation kernel.
Preferably, in the method for rapidly reconstructing three-dimensional distribution of source items of nuclear facilities provided in the embodiment of the present invention, in the down-sampling process of the feature extraction part, the size of the convolution kernel is gradually reduced, and the number of the convolution kernel is gradually increased; in the up-sampling process, the size of the convolution kernel is gradually increased, and the number of the convolution kernels is gradually reduced;
and the number of nodes of the hidden layer of the current layer in the advanced feature learning part is not less than that of nodes of the hidden layer of the next layer.
Preferably, in the method for rapidly reconstructing three-dimensional distribution of nuclear facility source items provided in the embodiment of the present invention, training the deep neural network model specifically includes:
setting the proportion among the training set, the verification set and the test set, and selecting the optimal learning rate, the transfer function and the training function;
and training the deep neural network model by adjusting the hyper-parameters according to the convergence condition until the error of the test set meets the expectation, and reaching the condition of terminating training.
Preferably, in the method for reconstructing three-dimensional distribution of source items of a nuclear facility provided in an embodiment of the present invention, the transforming an actual radiation field dose matrix and an actual source item position matrix specifically includes:
judging whether the actual radiation field dose matrix and the source item position matrix are higher than the input matrix resolution of the deep neural network model or not;
if yes, carrying out reduction transformation on the dose matrix;
and if not, carrying out amplification transformation on the dose matrix.
The embodiment of the invention also provides a device for rapidly reconstructing three-dimensional distribution of nuclear facility source items, which comprises:
the simulation calculation module is used for constructing a source item geometric model by adopting a Monte Carlo particle transport program and calculating a radiation field;
the model training module is used for extracting space radiation field data and three-dimensional distribution data of source item activity from the calculation result, constructing a source item position matrix with the same size according to source item distribution, generating a deep learning training sample, and constructing and training a deep neural network model;
the data prediction module is used for inputting the actual radiation field dose matrix and the source item position matrix after transformation processing to the trained deep neural network model and predicting three-dimensional distribution data of the source item activity;
and the data verification module is used for verifying the error between the output three-dimensional distribution data result of the source item activity and the actual value.
The embodiment of the present invention further provides a device for rapidly reconstructing three-dimensional distribution of nuclear facility source items, which includes a processor and a memory, wherein when the processor executes a computer program stored in the memory, the method for rapidly reconstructing three-dimensional distribution of nuclear facility source items provided by the embodiment of the present invention is implemented.
The embodiment of the present invention further provides a computer-readable storage medium for storing a computer program, where the computer program, when executed by a processor, implements the above-mentioned method for rapidly reconstructing three-dimensional distribution of nuclear facility source items according to the embodiment of the present invention.
According to the technical scheme, the method, the device, the equipment and the medium for rapidly reconstructing the three-dimensional distribution of the nuclear facility source items, provided by the invention, comprise the following steps: adopting a Monte Carlo particle transport program to construct a source item geometric model and calculating a radiation field; extracting three-dimensional distribution data of space radiation field data and source item activity from the calculation result, constructing source item position matrixes with the same size according to source item distribution, generating deep learning training samples, and constructing and training a deep neural network model; converting the actual radiation field dose matrix and the source item position matrix, inputting the converted radiation field dose matrix and the source item position matrix into a trained deep neural network model, and predicting three-dimensional distribution data of source item activity; and verifying the error between the output three-dimensional distribution data result of the source item activity and the actual value.
The invention provides a nuclear facility source item three-dimensional distribution fast reconstruction method based on deep learning, aiming at nuclear facilities with unknown source item activity and body distribution conditions, a proper deep neural network is obtained through construction and parameter adjustment, a deep neural network mode analysis method is applied, under the condition that the decoupling of complex physical relations between input and output is not carried out, the fast reconstruction of the three-dimensional distribution of the source item under the conditions of complex source items and strong anisotropic distribution of the nuclear facility by utilizing space radiation field data and a source item position matrix without being limited to a specific physical model is further realized through the deep learning training and generalization of known sample data, and the source item activity value at any specified position can be fast obtained through the deep learning intelligent inversion reconstruction under the condition of limited field measurement means.
Drawings
In order to more clearly illustrate the embodiments of the present invention or technical solutions in related arts, the drawings used in the description of the embodiments or related arts will be briefly introduced below, it is obvious that the drawings in the following description are only embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
Fig. 1 is a flowchart of a method for rapidly reconstructing three-dimensional distribution of a nuclear facility source item according to an embodiment of the present invention;
fig. 2 is a specific flowchart of a method for rapidly reconstructing three-dimensional distribution of a nuclear facility source item according to an embodiment of the present invention;
FIG. 3 is a top view of a cylindrical barrel radionuclide facility source item model provided by an embodiment of the present invention;
FIG. 4 is a front view of a cylindrical barrel radionuclide facility source item model provided by an embodiment of the present invention;
FIG. 5 is a graph of deep neural network training errors provided by embodiments of the present invention;
FIG. 6 is a statistical diagram of the error of the prediction result of the deep neural network according to the embodiment of the present invention;
fig. 7 is a comparison graph of activity distribution and a real value of a source item reconstructed based on deep learning in a first model provided in the embodiment of the present invention;
FIG. 8 is a comparison graph of activity distribution and real values of a source item reconstructed based on deep learning in a second model provided by the embodiment of the invention;
fig. 9 is a comparison graph of activity distribution and a real value of a source item reconstructed based on deep learning in a third model provided by the embodiment of the present invention;
fig. 10 is a schematic structural diagram of a nuclear facility source item three-dimensional distribution fast reconstruction apparatus according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The invention provides a nuclear facility source item three-dimensional distribution rapid reconstruction method, as shown in figure 1, comprising the following steps:
s101, constructing a source item geometric model by adopting a Monte Carlo Particle Transport program (Monte Carlo N Particle Transport Code, MCNP), and calculating a radiation field;
s102, extracting three-dimensional distribution data of space radiation field data and source item activity from a calculation result, constructing source item position matrixes with the same size according to source item distribution, generating Deep Learning (DL) training samples, and constructing and training a Deep Neural Network (DNN) model;
in practical application, spatial radiation field data is used as a first input, a source item position matrix is used as a second input, and three-dimensional distribution data of source item activity is used as an output. It should be understood that deep learning is one kind of machine learning, and machine learning is a necessary path for realizing artificial intelligence; the concept of deep learning is derived from the research of an artificial neural network, and a multilayer perceptron comprising a plurality of hidden layers is a deep learning structure; deep learning forms more abstract high-level representation attribute categories or features by combining low-level features to discover a distributed feature representation of the data;
s103, converting the actual radiation field dose matrix and the source item position matrix, inputting the converted actual radiation field dose matrix and the source item position matrix into a trained deep neural network model, and predicting three-dimensional distribution data of source item activity;
according to the invention, aiming at the strong anisotropy distribution condition of nuclear facility complex source items, a deep neural network formed by combining different types of networks is constructed, and after deep learning, training and generalization of a certain amount of sample data, an actual three-dimensional radiation field dose matrix and a three-dimensional source item position vector are input during prediction, so that intelligent rapid inversion reconstruction of three-dimensional source item activity distribution by three-dimensional radiation field data can be realized;
and S104, verifying the error between the output three-dimensional distribution data result of the source item activity and the actual value.
In the method for rapidly reconstructing three-dimensional distribution of nuclear facility source items provided by the embodiment of the invention, a suitable deep neural network is obtained by constructing and adjusting parameters for a nuclear facility with unknown source item activity and body distribution conditions, a deep neural network mode analysis method is applied, and rapid reconstruction of three-dimensional distribution of source items under the conditions of complex source items and strong anisotropic distribution of the nuclear facility by using space radiation field data and a source item position matrix without being limited to a specific physical model is realized by deep learning training and generalization of known sample data under the condition of not decoupling complex physical relationships between input and output, and the source item activity value at any specified position can be rapidly obtained by performing deep learning intelligent inversion calculation and reconstruction under the condition of limited field measurement means.
In specific implementation, in the method for rapidly reconstructing three-dimensional distribution of source items of a nuclear facility provided in the embodiment of the present invention, step S101 is to construct a geometric source item model by using a monte carlo particle transport program, and to calculate a radiation field and a geometric source item spatial distribution matrix, which may specifically include: firstly, carrying out grid division on source items according to the distribution condition of the source items; secondly, according to the source items divided into a plurality of areas, a Monte Carlo particle transport program is adopted to construct a source item geometric model; then, generating source item data with certain distribution by adopting an MCNP program universal source SDEF card; then, setting the region outside the source item as a non-interested region, setting a cubic space with a set size outside the source item, dividing the cubic space into networks, and carrying out statistics on the three-dimensional distribution of the activity of the source item on the divided cubic space; and finally, randomly sampling a plurality of groups of different source item geometric parameters, generating Monte Carlo calculation files in batches, and calling an MCNP program to calculate the radiation field.
In practical application, as shown in fig. 2, an inversion scene needing to be calculated is determined, whether the type of scene is trained or not is judged, if yes, a radiation field dose matrix actually detected is directly input, and a dose field is converted into neural network input; if not, randomly generating N groups of different source item simulation parameters, generating Monte Carlo calculation files in batches, calling a Monte Carlo particle transport calculation program to perform simulation calculation, and obtaining a calculation result. It should be understood that the calculation result includes data of the activity of the source item partition, the spatial radiation field, and the like. And the source item spatial distribution and spatial radiation field data can be extracted from the calculation result, and a neural network learning training sample is generated and output.
In specific implementation, in the method for rapidly reconstructing three-dimensional distribution of nuclear facility source items provided in the embodiment of the present invention, before the step S102 of constructing the deep neural network model is executed, the training samples may be preprocessed, which specifically includes: carrying out three-dimensional gridding on the spatial radiation field data and the three-dimensional distribution data of the source item activity; coarsening or thinning the three-dimensional distribution data of the three-dimensional gridded spatial radiation field data and the source item activity; taking the processed space radiation field and the constructed source item position matrix as input training samples; taking the processed three-dimensional distribution data of the activity of the source item as an output training sample; gaussian white noise is added to the output training samples.
The grid resolution can be reduced by coarsening the three-dimensional gridded data input and output according to actual needs, or the grid resolution can be improved by thinning the three-dimensional gridded sample; the generalization capability of the network can be improved by adding Gaussian white noise.
In specific implementation, in the method for rapidly reconstructing three-dimensional distribution of nuclear facility source items provided in the embodiment of the present invention, the step S102 of constructing a deep neural network model may specifically include: constructing a deep neural network model according to three parts of feature extraction, advanced feature learning and feature combination; the feature extraction part is composed of a plurality of convolution layers and a combination of down-sampling layers (namely convolution layers and down-sampling layers), a cross-layer, an up-sampling layer and a combination of convolution layers (namely up-sampling layers and convolution layers), the advanced feature learning part is composed of a plurality of full-connection layers, and the feature merging part is composed of a product calculation kernel.
Further, in specific implementation, in the method for rapidly reconstructing three-dimensional distribution of nuclear facility source items provided in the embodiment of the present invention, in the process of downsampling of the feature extraction part, the size of the convolution kernel is gradually reduced, and the number of the convolution kernels is gradually increased; in the up-sampling process, the size of the convolution kernel is gradually increased, and the number of the convolution kernels is gradually reduced; in the advanced feature learning part, the number of nodes of the hidden layer of the current layer is not less than that of nodes of the hidden layer of the next layer. It should be noted that the number of feature extraction layers is not too small; and the advanced characteristic learning part can ensure that the neuron in the hidden layer is 1-2 times of the neuron in the next layer as much as possible.
In specific implementation, in the method for rapidly reconstructing three-dimensional distribution of nuclear facility source items provided in the embodiment of the present invention, the step S102 of training the deep neural network model may specifically include: setting the proportion among the training set, the verification set and the test set, and selecting the optimal learning rate, the transfer function and the training function; and training the deep neural network model by adjusting the hyper-parameters according to the convergence condition until the error of the test set meets the expectation, and reaching the condition of terminating training.
In practical application, the ratio among the training set, the verification set and the test set can be set to be 8:1: 1; the learning rate is not suitable to be too large and can be less than 0.01, and the transfer function can use a ReLU or ELU function for a non-output layer to obtain a better training effect; the training function may select either the SGD or Adam functions.
In specific implementation, in the method for rapidly reconstructing three-dimensional distribution of source items of a nuclear facility provided in the embodiment of the present invention, the step S103 performs transformation processing on an actual radiation field dose matrix and a source item position matrix, which may specifically include: judging whether the actual radiation field dose matrix and the source item position matrix are higher than the input matrix resolution of the deep neural network model or not; if yes, carrying out reduction transformation on the dose matrix; and if not, carrying out amplification transformation on the dose matrix.
In practical application, inputting an actual radiation field dose matrix and a source item position matrix (in the matrix, 1 represents that a radioactive source exists in a corresponding position, and 0 represents that no radioactive source exists), if the resolution of the matrix is higher than the inherent input matrix of the neural network, performing reduction transformation, and conversely performing amplification transformation, wherein in the process, whether anti-aliasing transformation is performed or not can be selected according to actual conditions, the trained deep neural network parameters are loaded, and the activity distribution of source items is predicted by using the deep neural network.
The following describes in detail the method for rapidly reconstructing three-dimensional distribution of nuclear facility source items provided by the embodiment of the present invention, taking a cylindrical barrel radioactive nuclear facility source item as an example:
step one, constructing a source item model of the radioactive nuclear facility of the cylindrical barrel body, dividing the barrel body into a plurality of grid areas according to the distribution condition of the source items, in the embodiment, dividing the source items into 28 areas, and modeling by adopting an MCNP program, as shown in fig. 3 and 4: the barrel body is provided with a cover and a bottom at the lower part, the bottom and the cover are divided into 3 layers in the radial direction and 2 layers in the axial direction; the middle barrel body part is divided into 2 layers in the radial direction, 2 layers in the axial direction and 4 equal parts in the angular direction; the material in the barrel is assumed to be water, and the material in the barrel body is assumed to be iron; the outside of the barrel is a cubic space with air as a medium and is divided into 40 multiplied by 20 grids;
step two, generating certain distributed source item data in 28 cells by adopting an MCNP program universal source SDEF card, wherein the method comprises the following steps: in the radial direction of the barrel body, the activity of the source item is decreased from inside to outside, and specifically, the probability of generating particles by the inner grid cells is greater than that of the outer grid cells; in the axial direction of the barrel body, the activity of the source item is reduced from bottom to top, and the specific description is that the probability of generating particles by the lower grid cell is greater than that of the upper grid cell; in each grid cell, the activity of the source item is uniformly distributed, namely, the radial sampling is distributed according to a quadratic power function rule, and the axial sampling is uniformly distributed;
step three, constructing a model with the same size as the step one and the step two, setting a region outside the barrel as a non-interested region, setting a cubic space with a fixed size outside the barrel, dividing the cubic space into 20 multiplied by 20 grids, and counting the three-dimensional distribution of the activity of the source items;
sampling geometric parameters of the source item, randomly extracting 5000 groups of samples, generating calculation files in batches, and calculating the radiation field by adopting an MCNP program;
and fifthly, extracting data such as three-dimensional distribution of activity of source items and a spatial radiation field from the calculation output result, constructing source item position matrixes with the same size according to the distribution of the source items, generating a neural network learning training sample, preprocessing the sample, three-dimensionally gridding the spatial radiation field data with the size of 5000 multiplied by 32000 into 5000 multiplied by 40 multiplied by 20, scaling the three-dimensional gridding into 20 multiplied by 10, and constructing the source item position matrix with the size of 10 multiplied by 10 simultaneously to obtain the input sample. The activity data size of the source item partition is 5000 × 8000, the three-dimensional gridding is 5000 × 20 × 20 × 20 and is scaled to 10 × 10 × 10 to serve as an output sample, and Gaussian white noise is added (the sigma parameter of the noise is 0.0033 percent of the sample);
step six, constructing a deep neural network model, determining hyper-parameters, setting a training set and a testing set, and selecting parameters such as an optimal learning rate, a transfer function, a training function and the like: the input data 1 is a three-dimensional radiation field with the size of (20,20,10), the input data 2 is a source item distribution matrix with the size of (10,10, 10); convolutional layer 1, which contains 8 convolutional kernels, the size of the convolutional kernel is (3,3,3), the convolutional padding is "similarity (same)", and the activation function is ELU; convolutional layer 2, which contains 8 convolutional kernels, the size of the convolutional kernel is (3,3,3), the convolutional padding is "similarity (same)", and the activation function is ELU; pooling layer 1, maximum pooling method, mask size (5,5,1), step size 1; convolutional layer 3, which contains 16 convolutional kernels, the size of the convolutional kernel is (3,3,3), the convolutional padding is "similarity (same)", and the activation function is ELU; convolutional layer 4, which contains 16 convolutional kernels, the size of the convolutional kernel is (3,3,3), the convolutional padding is "similarity (same)", and the activation function is ELU; pooling layer 2, maximum pooling method, mask size (5,5,1), step size 1; convolutional layer 5, which contains 64 convolutional kernels, the size of the convolutional kernel is (2,2,2), the convolutional padding is "similarity (same)", and the activation function is ELU; convolutional layer 6, which contains 64 convolutional kernels, the size of the convolutional kernel is (2,2,2), the convolutional padding is "similarity (same)", and the activation function is ELU; pooling layer 3, maximum pooling method, mask size (5,5,1), step size 1; convolutional layer 7, which contains 64 convolutional kernels, the size of the convolutional kernel is (2,2,2), the convolutional padding is "similarity (same)", and the activation function is ELU; convolutional layer 8, which contains 64 convolutional kernels, the size of the convolutional kernel is (2,2,2), the convolutional padding is "similarity (same)", and the activation function is ELU; pooling layer 4, maximum pooling method, mask size (5,5,1), step size 1; convolutional layer 9, which contains 256 convolutional kernels, the size of which is (2,2,2), the convolutional padding is "similarity (same)", and the activation function is ELU; a convolutional layer 10, which contains 1024 convolutional kernels, the size of the convolutional kernel is (2,2,2), the convolutional padding is "similarity (same)", and the activation function is ELU; convolutional layer 11, which contains 256 convolutional kernels, the size of which is (2,2,2), the convolutional padding is "similarity (same)", and the activation function is ELU; an upsampling layer 1, upsampling the convolutional layer 11 by an upsampling size (2,2, 1); a cross-layer 1 connecting the convolution layer 8 and the output of the up-sampling layer 1; convolutional layer 12, which contains 64 convolutional kernels, the convolutional kernel size is (2,2,2), convolutional padding is "similarity (same)", and the activation function is ELU; convolutional layer 13, which contains 64 convolutional kernels, the size of the convolutional kernel is (2,2,2), the convolutional padding is "similarity (same)", and the activation function is ELU; an upsampling layer 2, upsampling the convolutional layer 13 by an upsampling size (2,2, 1); a cross-layer 2 connecting the convolution layer 4 and the output of the up-sampling layer 2; a convolutional layer 14, which contains 32 convolutional kernels, the size of the convolutional kernel is (2,2,2), the convolutional padding is "similarity (same)", and the activation function is ELU; a convolutional layer 15, which contains 32 convolutional kernels, the size of the convolutional kernel is (2,2,2), the convolutional padding is "similarity (same)", and the activation function is ELU; pooling layer 5, maximum pooling method, mask size (4,4,1), step size 1; a convolutional layer 16, which contains 16 convolutional kernels, the size of the convolutional kernel is (3,3,3), the convolutional padding is "similarity (same)", and the activation function is ELU; pooling layer 6, maximum pooling method, mask size (4,4,1), step size 1; convolutional layer 17, which contains 8 convolutional kernels, the size of which is (3,3,3), the convolutional padding is "similarity (same)", and the activation function is ELU; a parameter vectorization layer for converting the output of the convolution layer 17 into a one-dimensional vector; a fully-connected hidden layer 1, which comprises 3000 neurons and has an activation function of ReLU; a fully-connected hidden layer 2, which comprises 1500 neurons and has an activation function of ReLU; output layer 1, size 1000, no activation function; the characteristic merging layer multiplies the input data 2 and the output 1 one by one according to elements to obtain final output; selecting an Adam algorithm as a training function (the learning rate is 1E-4, and beta is 0.5), selecting a mean square error function (MAPE) as a loss function, setting the batch processing size to be 256, setting the proportion of a cross validation set to be 9:0:1, and setting the training iteration number to be 1000;
and step seven, repeatedly adjusting the proper hyper-parameter training neural network until the error of the test set meets the expectation, the final average absolute relative error of the training set is 0.60 percent, and the training termination condition is reached. The training error variation is shown in fig. 5, and the prediction error distribution is shown in fig. 6;
inputting an actual radiation field dose matrix and a source item position matrix, converting the input size into the input size required by the neural network through scaling, loading the trained deep neural network, and predicting the activity of the source item;
and step nine, verifying the error between the three-dimensional source item distribution reconstruction result and the actual value. In engineering, the source term calculation error is considered acceptable within 30%.
Three groups of models were selected for validation as follows:
model one: inner diameter of barrel body: 76.9183cm, barrel outside diameter: 98.9167cm, height in the barrel body: 200cm, height outside the barrel body: 242.530cm, comparing the activity distribution of the three-dimensional source item reconstructed based on deep learning with the real value, and obtaining the activity unit: bq, as shown in Table one and FIG. 7. The average absolute error is 5.11%, and the maximum absolute error is 25.31%.
Model two: inner diameter of barrel body: 75.0496cm, barrel outside diameter: 85.0952cm, height in the barrel body: 200cm, height outside the barrel body: 239.742cm, comparing the activity distribution of the three-dimensional source item reconstructed based on deep learning with the real value, and obtaining the activity unit: bq, as shown in Table two and FIG. 8. The average absolute error is 7.80%, and the maximum absolute error is 25.34%.
Watch two
And (3) model III: inner diameter of barrel body: 72.8702cm, barrel outside diameter: 84.4816cm, height in the barrel body: 200cm, height outside the barrel body: 242.531cm, comparing the activity distribution of the three-dimensional source item reconstructed based on deep learning with the real value, and obtaining the activity unit: bq, as shown in Table 3 and FIG. 9. The average absolute error is 4.39%, and the maximum absolute error is 12.27%.
Watch III
As can be seen from the verification and comparison results, the source term calculation errors of the three models are within 30%, which indicates that the output prediction result is acceptable.
Based on the same inventive concept, the embodiment of the invention also provides a device for rapidly reconstructing three-dimensional distribution of nuclear facility source items, and as the principle of solving the problems of the device is similar to that of the method for rapidly reconstructing three-dimensional distribution of nuclear facility source items, the implementation of the device can refer to the implementation of the method for rapidly reconstructing three-dimensional distribution of nuclear facility source items, and repeated parts are not described again.
In specific implementation, the apparatus for rapidly reconstructing three-dimensional distribution of nuclear facility source items provided in the embodiment of the present invention, as shown in fig. 10, specifically includes:
the simulation calculation module 11 is used for constructing a source item geometric model by adopting a Monte Carlo particle transport program and calculating a radiation field;
the model training module 12 is used for extracting the three-dimensional distribution data of the spatial radiation field data and the source item activity from the calculation result, constructing a source item position matrix with the same size according to the source item distribution, generating a deep learning training sample, and constructing and training a deep neural network model;
the data prediction module 13 is configured to transform the actual radiation field dose matrix and the source item position matrix, and then input the transformed actual radiation field dose matrix and the source item position matrix to the trained deep neural network model to predict three-dimensional distribution data of the source item activity;
and the data verification module 14 is used for verifying the error between the output three-dimensional distribution data result of the activity of the source item and the actual value.
In the nuclear facility source item three-dimensional distribution fast reconstruction device provided by the embodiment of the invention, the nuclear facility three-dimensional source item distribution fast reconstruction method can be constructed by adopting the deep neural network through the interaction of the four modules, deep learning training and neural network generalization are carried out by adopting a known data sample, and fast inversion calculation and reconstruction of nuclear facility three-dimensional source item activity distribution by utilizing three-dimensional space radiation field data and a source item position matrix can be realized.
For more specific working processes of the modules, reference may be made to corresponding contents disclosed in the foregoing embodiments, and details are not repeated here.
Correspondingly, the embodiment of the invention also discloses a device for rapidly reconstructing the three-dimensional distribution of the nuclear facility source items, which comprises a processor and a memory; when the processor executes the computer program stored in the memory, the method for rapidly reconstructing three-dimensional distribution of the nuclear facility source items disclosed in the foregoing embodiments is implemented.
For more specific processes of the above method, reference may be made to corresponding contents disclosed in the foregoing embodiments, and details are not repeated here.
Further, the present invention also discloses a computer readable storage medium for storing a computer program; the computer program is executed by a processor to realize the three-dimensional distribution rapid reconstruction method of the nuclear facility source item disclosed in the foregoing.
For more specific processes of the above method, reference may be made to corresponding contents disclosed in the foregoing embodiments, and details are not repeated here.
The embodiments are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same or similar parts among the embodiments are referred to each other. The device, the equipment and the storage medium disclosed by the embodiment correspond to the method disclosed by the embodiment, so that the description is relatively simple, and the relevant points can be referred to the method part for description.
Those of skill would further appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the various illustrative components and steps have been described above generally in terms of their functionality in order to clearly illustrate this interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.
The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. A software module may reside in Random Access Memory (RAM), memory, Read Only Memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.
To sum up, a method, an apparatus, a device and a medium for three-dimensional distribution fast reconstruction of nuclear facility source items provided by the embodiments of the present invention include: adopting a Monte Carlo particle transport program to construct a source item geometric model and calculating a radiation field; extracting three-dimensional distribution data of space radiation field data and source item activity from the calculation result, constructing source item position matrixes with the same size according to source item distribution, generating deep learning training samples, and constructing and training a deep neural network model; converting the actual radiation field dose matrix and the source item position matrix, inputting the converted radiation field dose matrix and the source item position matrix into a trained deep neural network model, and predicting three-dimensional distribution data of source item activity; and verifying the error between the output three-dimensional distribution data result of the source item activity and the actual value. According to the method, a proper deep neural network is obtained by constructing and adjusting parameters aiming at the nuclear facility with unknown source item activity and body distribution conditions, a deep neural network mode analysis method is applied, under the condition that the decoupling of complex physical relations between input and output is not carried out, the rapid reconstruction of the three-dimensional distribution of the source item activity under the conditions of complex source items and strong anisotropic distribution of the nuclear facility by utilizing space radiation field data and a source item position matrix which are not limited to a specific physical model is realized through the deep learning training and generalization of known sample data, and the source item activity value at any specified position can be rapidly obtained through the deep learning intelligent inversion reconstruction under the condition that the field measurement means is limited.
Finally, it should also be noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
The method, the device, the equipment and the medium for rapidly reconstructing three-dimensional distribution of nuclear facility source items provided by the invention are described in detail, a specific example is applied in the text to explain the principle and the implementation mode of the invention, and the description of the embodiment is only used for helping to understand the method and the core idea of the invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present invention.
Claims (10)
1. A nuclear facility source item three-dimensional distribution rapid reconstruction method is characterized by comprising the following steps:
adopting a Monte Carlo particle transport program to construct a source item geometric model and calculating a radiation field;
extracting three-dimensional distribution data of space radiation field data and source item activity from the calculation result, constructing source item position matrixes with the same size according to source item distribution, generating deep learning training samples, and constructing and training a deep neural network model;
converting the actual radiation field dose matrix and the source item position matrix, inputting the converted actual radiation field dose matrix and the source item position matrix into the trained deep neural network model, and predicting three-dimensional distribution data of source item activity;
and verifying the error between the output three-dimensional distribution data result of the source item activity and the actual value.
2. The nuclear facility source item three-dimensional distribution rapid reconstruction method according to claim 1, wherein a Monte Carlo particle transport program is adopted to construct a source item geometric model, and calculation of a radiation field and a source item space geometric distribution matrix is performed, specifically comprising:
according to the distribution condition of the source items, carrying out grid division on the source items;
according to the source items divided into a plurality of areas, a Monte Carlo particle transport program is adopted to construct a source item geometric model;
generating source item data with certain distribution by adopting a universal source SDEF card of a Monte Carlo particle transport program;
setting the region except the source item as a non-interested region, setting a cubic space with a set size outside the source item, dividing the cubic space into networks, and performing three-dimensional distribution statistics of the activity of the source item on the divided cubic space;
randomly sampling a plurality of groups of different source item geometric parameters, generating Monte Carlo calculation files in batches, and calling a Monte Carlo particle transport program to calculate the radiation field.
3. The method for rapidly reconstructing three-dimensional distribution of nuclear facility source items according to claim 2, wherein before constructing the deep neural network model, the method further comprises:
performing three-dimensional gridding on the spatial radiation field data and the three-dimensional distribution data of the source item activity;
coarsening or thinning the three-dimensional distribution data of the space radiation field data and the source item activity after three-dimensional gridding;
taking the processed space radiation field and the constructed source item position matrix as input training samples;
taking the processed three-dimensional distribution data of the activity of the source item as an output training sample;
and adding Gaussian white noise to the output training sample.
4. The method for rapidly reconstructing three-dimensional distribution of nuclear facility source items according to claim 3, wherein the constructing of the deep neural network model specifically comprises:
constructing a deep neural network model according to three parts of feature extraction, advanced feature learning and feature combination; the feature extraction part is composed of a combination of a plurality of convolutional layers and downsampling layers and a combination of a cross layer, an upsampling layer and a convolutional layer, the advanced feature learning part is composed of a plurality of fully-connected layers, and the feature combination part is composed of a product calculation kernel.
5. The method for rapidly reconstructing three-dimensional distribution of nuclear facility source items according to claim 4, wherein in the characteristic extraction part, the sizes of convolution kernels are gradually reduced, and the number of the convolution kernels is gradually increased in the process of downsampling; in the up-sampling process, the size of the convolution kernel is gradually increased, and the number of the convolution kernels is gradually reduced;
and the number of nodes of the hidden layer of the current layer in the advanced feature learning part is not less than that of nodes of the hidden layer of the next layer.
6. The method for three-dimensional distributed rapid reconstruction of nuclear facility source items according to claim 5, wherein training the deep neural network model specifically comprises:
setting the proportion among the training set, the verification set and the test set, and selecting the optimal learning rate, the transfer function and the training function;
and training the deep neural network model by adjusting the hyper-parameters according to the convergence condition until the error of the test set meets the expectation, and reaching the condition of terminating training.
7. The method for reconstructing three-dimensional distribution of source items of a nuclear facility according to claim 6, wherein transforming the actual radiation field dose matrix and the source item position matrix specifically comprises:
judging whether the actual radiation field dose matrix and the source item position matrix are higher than the input matrix resolution of the deep neural network model or not;
if yes, carrying out reduction transformation on the dose matrix;
and if not, carrying out amplification transformation on the dose matrix.
8. A nuclear facility source item three-dimensional distribution rapid reconstruction device is characterized by comprising:
the simulation calculation module is used for constructing a source item geometric model by adopting a Monte Carlo particle transport program and calculating a radiation field;
the model training module is used for extracting space radiation field data and three-dimensional distribution data of source item activity from the calculation result, constructing a source item position matrix with the same size according to source item distribution, generating a deep learning training sample, and constructing and training a deep neural network model;
the data prediction module is used for inputting the actual radiation field dose matrix and the source item position matrix after transformation processing to the trained deep neural network model and predicting three-dimensional distribution data of the source item activity;
and the data verification module is used for verifying the error between the output three-dimensional distribution data result of the source item activity and the actual value.
9. A three-dimensional distribution fast reconstruction apparatus of a nuclear facility source item, comprising a processor and a memory, wherein the processor implements the three-dimensional distribution fast reconstruction method of the nuclear facility source item according to any one of claims 1 to 7 when executing a computer program stored in the memory.
10. A computer-readable storage medium for storing a computer program, wherein the computer program, when executed by a processor, implements the method for three-dimensional distributed fast reconstruction of nuclear facility source items according to any of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010513588.6A CN111667571B (en) | 2020-06-08 | 2020-06-08 | Nuclear facility source item three-dimensional distribution rapid reconstruction method, device, equipment and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010513588.6A CN111667571B (en) | 2020-06-08 | 2020-06-08 | Nuclear facility source item three-dimensional distribution rapid reconstruction method, device, equipment and medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111667571A true CN111667571A (en) | 2020-09-15 |
| CN111667571B CN111667571B (en) | 2021-09-17 |
Family
ID=72385791
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010513588.6A Active CN111667571B (en) | 2020-06-08 | 2020-06-08 | Nuclear facility source item three-dimensional distribution rapid reconstruction method, device, equipment and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111667571B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113205596A (en) * | 2021-05-27 | 2021-08-03 | 清华大学 | Radiation field inversion reconstruction method based on curved surface mean curvature and diffusion equation |
| CN114861538A (en) * | 2022-05-07 | 2022-08-05 | 中核四川环保工程有限责任公司 | Post-processing plant retired radiation field estimation method and system based on random forest algorithm |
| WO2022198686A1 (en) * | 2021-03-26 | 2022-09-29 | Shanghaitech University | Accelerated neural radiance fields for view synthesis |
| CN116976202A (en) * | 2023-07-12 | 2023-10-31 | 清华大学 | Fixed complex source item distribution inversion method and device based on deep neural network |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0833616A (en) * | 1994-07-22 | 1996-02-06 | Nec Corp | Device and method for transmitting living body internal state |
| CN107103316A (en) * | 2010-11-04 | 2017-08-29 | 数字标记公司 | Method and system based on smart mobile phone |
| CN107330187A (en) * | 2017-06-30 | 2017-11-07 | 哈尔滨工程大学 | A kind of elimination of nuclear facilities radiation field dosage distributed simulation method |
| CN109101704A (en) * | 2018-07-24 | 2018-12-28 | 南华大学 | Depleted nuclear fuel is used as the feasibility assessment method of Marine Nuclear Power Plant cabin protective materials |
| CN109471999A (en) * | 2018-09-25 | 2019-03-15 | 中国辐射防护研究院 | A kind of the gamma radiation field data correction calculation method and system of non-homogeneous source item distribution |
| WO2019103354A1 (en) * | 2017-11-24 | 2019-05-31 | 주식회사 레이 | Correction method and correction apparatus for scattered rays of dental cone beam ct |
| CN110133703A (en) * | 2019-05-15 | 2019-08-16 | 中国辐射防护研究院 | A kind of gamma-rays Dosimetry based on BP neural network and Monte carlo algorithm |
| CN110210555A (en) * | 2019-05-29 | 2019-09-06 | 西南交通大学 | Rail fish scale hurt detection method based on deep learning |
| CN111159149A (en) * | 2019-12-13 | 2020-05-15 | 国网浙江省电力有限公司紧水滩水力发电厂 | River flow prediction method based on three-dimensional convolutional neural network |
| CN111175803A (en) * | 2020-01-12 | 2020-05-19 | 兰州大学 | Borehole core facility retired waste radioactive multi-parameter detection system and method |
-
2020
- 2020-06-08 CN CN202010513588.6A patent/CN111667571B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0833616A (en) * | 1994-07-22 | 1996-02-06 | Nec Corp | Device and method for transmitting living body internal state |
| CN107103316A (en) * | 2010-11-04 | 2017-08-29 | 数字标记公司 | Method and system based on smart mobile phone |
| CN107330187A (en) * | 2017-06-30 | 2017-11-07 | 哈尔滨工程大学 | A kind of elimination of nuclear facilities radiation field dosage distributed simulation method |
| WO2019103354A1 (en) * | 2017-11-24 | 2019-05-31 | 주식회사 레이 | Correction method and correction apparatus for scattered rays of dental cone beam ct |
| CN109101704A (en) * | 2018-07-24 | 2018-12-28 | 南华大学 | Depleted nuclear fuel is used as the feasibility assessment method of Marine Nuclear Power Plant cabin protective materials |
| CN109471999A (en) * | 2018-09-25 | 2019-03-15 | 中国辐射防护研究院 | A kind of the gamma radiation field data correction calculation method and system of non-homogeneous source item distribution |
| CN110133703A (en) * | 2019-05-15 | 2019-08-16 | 中国辐射防护研究院 | A kind of gamma-rays Dosimetry based on BP neural network and Monte carlo algorithm |
| CN110210555A (en) * | 2019-05-29 | 2019-09-06 | 西南交通大学 | Rail fish scale hurt detection method based on deep learning |
| CN111159149A (en) * | 2019-12-13 | 2020-05-15 | 国网浙江省电力有限公司紧水滩水力发电厂 | River flow prediction method based on three-dimensional convolutional neural network |
| CN111175803A (en) * | 2020-01-12 | 2020-05-19 | 兰州大学 | Borehole core facility retired waste radioactive multi-parameter detection system and method |
Non-Patent Citations (3)
| Title |
|---|
| BI Y. ET AL.: "Study on fast radiation transport calculation methodology in decommissioning of research reactor", 《AT. ENERGY SCI. TECHNOL.》 * |
| 宋英明，等: "核设施退役过程中的辐射场重构与拆除路径优化", 《核技术》 * |
| 罗文，等: "核设施退役作业场景动态三维辐射场模拟计算与可视化", 《科学技术与工程》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2022198686A1 (en) * | 2021-03-26 | 2022-09-29 | Shanghaitech University | Accelerated neural radiance fields for view synthesis |
| CN113205596A (en) * | 2021-05-27 | 2021-08-03 | 清华大学 | Radiation field inversion reconstruction method based on curved surface mean curvature and diffusion equation |
| CN113205596B (en) * | 2021-05-27 | 2022-05-31 | 清华大学 | Radiation field inversion reconstruction method based on curved surface mean curvature and diffusion equation |
| CN114861538A (en) * | 2022-05-07 | 2022-08-05 | 中核四川环保工程有限责任公司 | Post-processing plant retired radiation field estimation method and system based on random forest algorithm |
| CN114861538B (en) * | 2022-05-07 | 2024-05-07 | 中核四川环保工程有限责任公司 | Method and system for estimating retired radiation field of post-treatment plant based on random forest algorithm |
| CN116976202A (en) * | 2023-07-12 | 2023-10-31 | 清华大学 | Fixed complex source item distribution inversion method and device based on deep neural network |
| CN116976202B (en) * | 2023-07-12 | 2024-03-26 | 清华大学 | Fixed complex source item distribution inversion method and device based on deep neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111667571B (en) | 2021-09-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111667571B (en) | Nuclear facility source item three-dimensional distribution rapid reconstruction method, device, equipment and medium | |
| CN111666718B (en) | Intelligent inversion method, device and equipment for nuclear facility source activity and storage medium | |
| Vallecorsa et al. | 3D convolutional GAN for fast simulation | |
| Ebiwonjumi et al. | Machine learning of LWR spent nuclear fuel assembly decay heat measurements | |
| Hoseyni et al. | A Bayesian ensemble of sensitivity measures for severe accident modeling | |
| Donà et al. | Summing bulk quantum numbers with Monte Carlo in spin foam theories | |
| Yau et al. | On the generalizability of artificial neural networks in spin models | |
| Khattak et al. | High energy physics calorimeter detector simulation using generative adversarial networks with domain related constraints | |
| Sharma et al. | House price prediction using machine learning algorithm | |
| Yang et al. | A data-enabled physics-informed neural network with comprehensive numerical study on solving neutron diffusion eigenvalue problems | |
| CN108765137A (en) | A kind of credit demand prediction technique and system, storage medium | |
| CN116595885B (en) | Lightweight shielding deep coupling optimization method for small reactor | |
| CN111105014A (en) | Data processing method, equipment and computer readable storage medium | |
| Wolberg | Designing quantitative experiments: prediction analysis | |
| Ebiwonjumi et al. | Bayesian method and polynomial chaos expansion based inverse uncertainty quantification of spent fuel using decay heat measurements | |
| Abrate et al. | A non-intrusive reduced order model for the characterisation of the spatial power distribution in large thermal reactors | |
| Figueroa et al. | Nuclear archaeology based on measurements of reprocessing waste: first experimental results | |
| Angelo | Representing nuclear criticality excursion experiment data by an artificial neural network | |
| Proctor | Elements of high-order predictive model calibration algorithms with applications to large-scale reactor physics systems | |
| Bouhadida et al. | Neutron spectrum unfolding using two architectures of convolutional neural networks | |
| Bouhadida et al. | Neutron spectra reconstruction based on an artificial neural network trained with a large built dataset | |
| Forget | A three dimensional heterogeneous coarse mesh transport method for reactor calculations | |
| Krivtchik et al. | Development of depletion code surrogate models for uncertainty propagation in scenario studies | |
| Viljoen | Machine learning for particle identification & deep generative models towards fast simulations for the Alice Transition Radiation Detector at CERN | |
| Bartel | Analysis and Improvement of the bRAPID Algorithm and its Implementation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |