KR20210073468A - Nuclear facility unmanned surveillance and artificial intelligence-based automatic alarm system - Google Patents

Abstract

An automated nuclear facility surveillance and artificial intelligence-based automatic alarm system is disclosed. The automatic alarm system in accordance with the present invention comprises: a radiation detecting unit; an optical imaging device detecting unit converting light into an electrical signal; an analog signal processing unit; a digital signal processing unit comparing multiplexed output signals of the analog signal processing unit with data obtained in a specific time to measure a coincidence coefficient; and a control unit executing an artificial intelligence algorithm (convolutional neural networks) for automatic surveillance through which radioactivity deviated from a specific position is identified by using an NN algorithm and a CNN algorithm, thereby performing automated surveillance of nuclear facilities and therethrough, easily managing nuclear power plants and radioactive waste treatment facilities.

Description

Unmanned monitoring of nuclear facilities and automatic warning system based on artificial intelligence {NUCLEAR FACILITY UNMANNED SURVEILLANCE AND ARTIFICIAL INTELLIGENCE-BASED AUTOMATIC ALARM SYSTEM}

The present invention relates to an automatic radiation warning system, and more particularly, to accurately and automatically measure the distribution of radiation in real time and automatically for detecting unidentified nuclear material using a radiation detector or monitoring a nuclear waste storage and nuclear material leakage accident. It relates to unmanned monitoring of nuclear facilities that can issue an alarm and an automatic alarm system based on artificial intelligence.

The technologies currently used for detection and tracking of radioactive materials include gamma and Compton cameras that measure the two-dimensional distribution of the radiation field, a dose detector that monitors the effective dose in space, and the energy of radiation emitted from radioactive materials. There are spectroscopy to measure and optical image acquisition using a CCD camera.

In the case of a gamma camera, it is possible to detect the location of radioactive materials, but there is a problem of reduced sensitivity due to the use of a collimator (collimator). And there is a limitation in that the production cost increases.

In addition, the Compton camera can measure a larger area than the gamma camera, but there is a structural problem in that the incident radiation has to respond to both the detectors composed of two layers, and the decrease in sensitivity due to the use of an electrical collimator to determine the simultaneous coefficient. Occurs.

In addition, the dose detector is used to generate an alarm when the standard value is exceeded through dose measurement, but because it has only a function of changing the measured radiation into a dose, it is impossible to classify and track the location according to the inflow and outflow of radioactive materials. However, it has a limitation in that it is not possible to distinguish the type and nuclide of the generated radiation.

In addition, spectroscopy can distinguish nuclides of a radiation source using a relatively simple algorithm through the energy of the incident radiation, but similarly to a dose detector, it does not consider the direction of incident radiation, so it cannot be used for tracking radioactive materials. there is

On the other hand, as the domestic nuclear power and radiation use industry develops, safe management of radioactive materials generated and handled in nuclear power plants or radiation-using facilities is essential, but in the case of unmanned monitoring of nuclear facilities, an alarm system using dosimeters installed in major places is required. There is a problem because it is losing.

KR Registered Patent No. 10-1675733 (2016.11.08)

The present invention for solving these problems is to use a monitoring system using a communication terminal attached to a storage container for storage and management of radioactive materials, and artificial intelligence-based radioactive material tracking and alarm functions for unmanned monitoring of nuclear facilities and artificial intelligence. It aims to provide an automatic alarm system based on

Another object of the present invention is to provide a nuclear facility unmanned monitoring and artificial intelligence-based automatic warning system capable of accurately measuring the distribution of radiation in real time and automatically issuing an automatic alarm.

In addition, the present invention combines a sensor network and artificial intelligence to overcome the limitations of the conventional spectroscopy system in which location detection is impossible, and the unmanned monitoring and artificial intelligence of nuclear facilities that can detect the inflow of radioactive materials and the unmanned monitoring of a large area. It is another object to provide an automatic alarm system based.

In order to solve this problem, the automatic alarm system based on unmanned monitoring of nuclear facilities and artificial intelligence of the present invention includes an optical sensor that detects radiation emitted from an object (radioactive material) and converts the converted flash signal into an electrical signal. An analog signal multiplexed with an analog signal processing unit composed of a signal amplifier and comparator for multiplexing and amplifying a plurality of optical sensor output signals from a radiation detection unit and the radiation detection unit, and adjusting a signal rise time, a fall time, a signal width, or an offset voltage. Digital signal processing unit and machine that acquires reaction position, reaction size, and reaction time information through ADC, TDC or FPGA (Xilinx family, Altera family) of the output signal of the processing unit, and measures the simultaneous coefficient by comparing the data acquired within a specific time It operates as a running data processing unit and receives digitized output information from the digital signal processing unit and uses NN (Neural networks), an artificial intelligence algorithm, to determine the radiation response and detection position and to include a data processing unit to track the position. can do.

Therefore, according to the automatic alarm system based on unmanned monitoring of nuclear facilities and artificial intelligence according to an embodiment of the present invention, unmanned monitoring of nuclear facilities can be performed using automatic alarm technology using radiation field distribution measurement and object recognition. This has the effect of facilitating the management of nuclear power plants and radioactive waste treatment facilities.

In addition, according to the automatic alarm system based on unmanned monitoring of nuclear facilities and artificial intelligence of the present invention, since the redeveloped sensor network and the location tracking system using artificial intelligence are used, the worker can directly It can be used for multi-purpose environmental monitoring in difficult-to-access places.

In addition, according to the automatic alarm system based on unmanned monitoring of nuclear facilities and artificial intelligence of the present invention, it is possible to accurately measure the distribution of radiation in real time and issue an automatic alarm, so that unidentified nuclear material detection, nuclear waste storage, and nuclear material leakage accident are possible. has the effect of monitoring

And according to the nuclear facility unmanned monitoring and artificial intelligence-based automatic alarm system of the present invention, unmanned monitoring and radiation of a large area by combining a sensor network and artificial intelligence in order to overcome the limitations of the conventional spectroscopy system in which location detection is impossible. By developing an unmanned radiation monitoring and automatic alarm system that can detect the ingress of substances, the sensitivity and uniformity characteristics in the measurement space are improved, the location tracking accuracy is improved, and the resolution in the outskirts is improved to automatically track and monitor the movement path of objects in real time. is possible

1 is a main configuration diagram of a nuclear facility unmanned monitoring and artificial intelligence-based automatic alarm system according to an embodiment of the present invention;
2 is a diagram illustrating an artificial intelligence-based radiation unmanned monitoring and automatic warning system for unmanned monitoring of nuclear facilities;
Figure 3 is an exemplary diagram of the radioactivity change detection model using the detector data and the position change monitoring model using the CCD image;
4 is an artificial neural network model classification result and CNN model learning result (loss function and positive classification rate graph) diagram;
5 is a flowchart for explaining an artificial intelligence-based radiation monitoring and automatic alarm method for unmanned monitoring of nuclear facilities;
6 is a diagram illustrating area sharing of a radiation detector;
7 is a diagram illustrating motion detection through artificial intelligence learning;
8 is a reference view for explaining the process of constructing a data set;
And
9 is an example screen in which data sets are automatically input to the GUI window.

The terms or words used in the present specification and claims are not to be construed as limited in their ordinary or dictionary meanings, and on the principle that the inventor can appropriately define the concept of the term in order to best describe his invention. It should be interpreted as meaning and concept consistent with the technical idea of the present invention.

Throughout the specification, when a part "includes" a certain element, it means that other elements may be further included, rather than excluding other elements, unless otherwise stated. In addition, terms such as “…unit”, “…group”, “module”, and “device” described in the specification mean a unit that processes at least one function or operation, which is implemented by a combination of hardware and/or software can be

Throughout the specification, the term “and/or” should be understood to include all possible combinations from one or more related items. For example, the meaning of "the first item, the second item and/or the third item" may be presented from the first, second, or third item as well as two or more of the first, second, or third items. A combination of all possible items.

In each step throughout the specification, identification symbols (eg, a, b, c, ...) are used for convenience of description, and identification codes do not limit the order of each step, and each step is Unless the context clearly dictates a specific order, the order may differ from the stated order. That is, each step may occur in the same order as specified, may be performed substantially simultaneously, or may be performed in the reverse order.

Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

First, the present invention automatically monitors the movement path and monitoring of radioactive materials using a signal processing part that acquires and analyzes a signal coming out of a plurality of radiation detectors and detectors, and artificial intelligence technology for automatically tracking and monitoring the movement route in real time. Characterized in tracking and monitoring discrimination

To this end, the present invention is configured to include a radiation detection unit 110 , an analog signal processing unit 120 , a digital signal processing unit 130 , a data processing unit 140 performing an algorithm, and an image processing unit 150 .

Referring to the main configuration diagram of a nuclear facility unmanned monitoring and artificial intelligence-based automatic alarm system according to an embodiment of the present invention in FIG. 1 , the radiation detection unit 110 as shown in the figure includes the radiation emitted from the object (radioactive material). is detected, converted into an electrical signal and output.

To this end, the radiation detector 110 detects the radiation emitted from the object (radioactive material) and converts it into a flash signal. The radiation detector 112 and the one-layer or multi-layered flash crystal output from the radiation detector 112 electrically and a photodetector 114 composed of a photosensor that converts and outputs a signal.

The photodetector 114 generates light from the scintillation crystal by radiation emitted from the radioactive material, and converts the light signal into an electrical signal by using the light signal, and operates to convert the light signal into an electrical signal.

Specifically, the radiation detection unit 110 is operated to generate a detection signal from the plurality of radiation detectors 112 and then output the radiation emitted from the nuclear material or the object or the radiation transmitted by being irradiated to the object.

The radiation detection unit 110 generates a difference in radiation counts according to positions detected through a plurality of detectors coupled with a scintillation crystal and an optical sensor, and analyzes it with an artificial intelligence-based algorithm using the count ratio detected by each radiation detector. This is to enable automatic tracking or triggering of alarms.

Referring to the diagram illustrating an artificial intelligence-based radiation monitoring and automatic alarm system for unattended monitoring of nuclear facilities in FIG. 2 , the radiation detection unit 110 can be installed in a number close to or far away from the required place, so each identification code is provided. When the detection signal is transmitted and transmitted together with the identification code, it is possible not only to identify the installed location, but also to easily check the location through the screen in the control center 200 when an event occurs.

Of course, the identification code may be an IP address, a product code, an installation address, or an administrator's name.

Referring to Figure 2, the left figure illustrates the situation before radiation detection, and the right figure illustrates the situation after radiation detection. In each figure, the upper left shows the nuclear waste storage facility monitoring space, and the right shows the airport terminal monitoring space. did it

If you look at the regular radiation field measurement screen of the nuclear waste storage facility before radiation detection, no radioactivity is detected, but if you look at the change in the radiation field on the right, a nuclear material leakage accident has occurred in the left central part of the monitoring space, and an alarm is issued through automatic tracking. It can be seen that this is happening

In addition, the gamma-ray counting rate before the influx of unidentified nuclear material into the monitoring space of the airport terminal before radiation detection is displayed as a graph. have.

The analog signal processing unit 120 is configured to multiplex and then amplify the output signals of the plurality of photodetectors 114 in a configuration including a signal amplification and comparator to adjust the rise time, fall time, signal width, or offset voltage of the output signal. do.

That is, the analog signal processing unit 120 receives the radiation detection signals for each of the plurality of radiation detectors output from the radiation detection unit 110 , and amplifies and shapes the signals.

The digital signal processing unit 130 acquires the reaction position, reaction magnitude, and reaction time information through ADC, TDC or FPGA (Xilinx family, Altera family) of the output signal of the multiplexed analog signal processing unit 120, and acquires it within a specific time It operates to measure the simultaneity coefficient by comparing the collected data.

The data processing unit 140 operates as a machine learning data processing unit and receives digitized output information from the digital signal processing unit 130 and tracks the radiation response and detection position by using an NN (Neural Networks) artificial intelligence algorithm.

The radiation detector 110 constitutes a multi-channel detector, and each detector monitors a certain area and at the same time shares a certain monitoring area with a nearby detector, so the radiation measurement coefficients obtained from a single detector are converted into data for artificial intelligence. It is used as data for algorithm execution. At this time, it is possible to track the position of the radioactive material by the ratio of the radiation measurement coefficient obtained from each detector.

Referring to the diagram illustrating the area sharing of the radiation detectors of FIG. 6 , it can be seen that each radiation detector shares a predetermined area.

The image processing unit 150 operates to identify a material deviated from a specific location using an optical image.

Specifically, it receives CCD images installed in the monitoring area and uses a CNN (Convolution Neural Networks) algorithm to perform an artificial intelligence algorithm for automatic monitoring that identifies radioactivity deviated from a specific location (Convolution neural networks).

Convolutional Neural Network (CNN), one of the AI-based algorithms, is an algorithm optimized to identify and classify image features in unstructured data such as 2-dimensional images. An image can be classified into an image that does not deviate from the correct position) and an image that is not correct (radioactivity deviates from a specific position). For this, the AI needs to learn the correct and non-correct images in advance.

In other words, it is to perform artificial intelligence algorithms (convolutional neural networks) for automatic monitoring that identifies radiation deviating from a specific location.

After the learning process is completed with a simulation or a minimum image, it is determined whether there is a movement at a specific location with an artificial intelligence that has previously learned an image received in real time.

Referring to the diagram exemplifying motion detection through artificial intelligence learning of FIG. 7 , if there is no movement through learning, it is displayed as “0” and if it deviates from a specific position, it is displayed as “1” to detect motion through learning. .

In addition, the image processing unit 150 operates to automatically track the detected coefficient ratio of each detector and generate an alarm according to a condition using an AI-based algorithm.

For example, in a 5 X 5 grid pattern, the number of counts detected by each detector is slightly different depending on the location of the defect. The data set is first constructed through the pattern showing the difference.

Referring to the reference diagram for explaining the process of composing the dataset of FIG. 8, it is data that is formed when position 1 is missing, where v1 to v6 represent each detector, and y is information of the corresponding row. It means correct answer data that determines whether or not (in this case, 1) indicates that the is missing (in this case, 0).

Next, in the data configured above, the data is processed as a ratio value for each row, and then the AI-based algorithm is applied to each of the 25 digits to determine whether the corresponding position is missing or not.

That is, if the learned AI algorithm determines that there is a defect in the corresponding location, it will indicate that there is a defect.

A radiation unattended monitoring and automatic alarm method of the present invention will be described using the above-described configuration.

5 is a flowchart for explaining an artificial intelligence-based radiation monitoring and automatic alarm method for unattended monitoring of nuclear facilities. As shown in FIG. 5 , the radiation detection unit 110 irradiates radiation emitted from a nuclear material or an object to the object or to the object. It is operated to generate and output the transmitted radiation after generating a detection signal in a plurality of detectors (S110).

Step S110 consists of first detecting radiation (S112) and outputting a detection signal (S114).

Referring to the drawing illustrating the artificial intelligence-based radiation monitoring and automatic alarm system for unmanned monitoring of nuclear facilities of FIG. 2, the radiation detector 112 is installed as a monitoring space at least one nuclear waste storage facility and an airport terminal, It is determined that the radiation field is constantly measured and the gamma-ray counting rate is monitored before the unidentified nuclear material is introduced, and then the radiation field is changed and the gamma-ray counting rate is changed after the unidentified nuclear material is introduced (S112).

That is, the step (S112) of detecting the emitted radiation by the radiation detector 112 and the output signal of the single-layer or multi-layered scintillation crystal output from the radiation detector 112 using the photodetector 114 composed of an optical sensor is converted into an electrical signal. converted to and outputted in step S114.

When the detection signal is generated and output by the radiation detector in step S110, the analog signal processing unit 120 receives a plurality of radiation detection signals for each channel output from the radiation detection unit 110, and amplifies and shapes the signal (S120).

In step S130, the digital signal processing unit 130 receives the analog signal output from the analog signal processing unit 120, converts it into a digital value using ADC or TDC, and outputs energy and position information in response to the received signal. do.

When energy and position information are output from the digital signal processing unit 130, the data processing unit 140 receives the digitized output information from the digital signal processing unit 130 and uses a neural networks algorithm to determine the radiation response and detection position. NN (Neural networks) artificial intelligence algorithm and CCD image for automatic monitoring are performed using a CNN (Convolution neural networks) algorithm for automatic monitoring to identify radioactivity deviated from a specific location (S140) ).

Whether it is possible to determine the detection location and track the location using energy and location information was written in detail in the previous question, but after applying the algorithm, it is possible to determine the detection location and track the location through the GUI (S150).

That is, when the detection signal is transmitted along with the identification code, it is possible not only to identify the installed location, but also to easily check the location through the screen in the control center 200 when an event occurs.

The learned algorithm is stored, and then there are new datasets that are created in real time. Corresponding data sets are automatically input to the GUI window, and in the GUI window, movement can also be detected in parallel while determining the position of the defect through data input in real time.

9 exemplifies a screen in which data sets are automatically input to the GUI window.

The CNN also learns the features (edge, line, point, object ...) of the image through the layers present in the middle of the model structure shown in FIG. Characteristics are extracted and learned, and the video being learned is divided into a correct answer image and a non-correct answer image.

After that, when a real-time image is entered into the algorithm as a test image in the learned algorithm, the test image is classified into correct and non-correct answers through the already learned algorithm, and motion is detected and tracked.

Referring to the example diagram of the radioactivity change detection model using the detector data of FIG. 3 and the radioactivity position change monitoring model using the CCD image, looking at the artificial neural network model on the left, the deep neural network uses one input layer and one output layer to collect data learn and recognize or infer patterns. The number of radiation signals determined by the detector is applied to the model as input layer data. Through this, the result of determining in which area the radioactive material is detected is calculated as a value in the output layer.

The right figure is an example of a CNN (Convolutional Neural Network) structure useful for recognizing images and finding patterns. It receives an image as input data, performs a convolution operation through the pixels of the image, recognizes the image, and finds a pattern. The CCD image is used as the input image of the CNN model, and the CNN model is used to detect the position change of the drum containing the radioactive material using the CCD image. This means that the model can learn the pattern and detect the movement when the position of the drum moves from the existing position through the learning of the model.

4 is a diagram of an artificial neural network model classification result and a CNN model learning result (loss function and positive classification rate graph), the graph is a graph showing whether the model is trained stably and accurately. The left graph is a loss function graph, and the right graph is a correct classification graph. In both graphs, the X axis represents the number of learning times.

The loss function graph shows how much the model's predicted value differs from the actual value. The loss function graph of a well-trained model shows a decrease as the number of training increases, like an exponential function. The AI developed in the present invention also shows that the difference between the predicted value and the actual value decreases as the number of times of learning increases, indicating that it is a well-learned model.

The positive classification rate graph is a graph showing the target value of the present invention, and it can be seen that the target value (correction rate) increases as the number of times increases because the model is well trained. And as the number of learning exceeds 200, it can be seen that the target value converges to a specific value, indicating that the model is stabilized.

As described above, in the present invention, the number of sensor networks optimized according to the measurement space is operated, the response characteristics according to the radiation response characteristics and the size and position of the detector are analyzed, and the position of the system is established as a database by DB learning using artificial intelligence. resolution can be improved.

In addition, we intend to develop an automatic alarm system because it has excellent scalability to various monitoring spaces and can detect targets in a short time. The existing radiation monitoring system provided only a simple fusion image of a radiation image and an optical image, but in the present invention, an artificial intelligence-based object recognition algorithm is applied to an optical image to automatically track the movement path of an object as well as a fusion image of radiation and optics. and monitoring.

Although it is possible to detect the location of radioactive materials in the existing monitoring system, there is a problem of reduced sensitivity due to the use of a collimator (collimator), and since the sensitivity and effective field of view are proportional to the area of the detector, the detector is enlarged and manufactured to monitor a large area In order to overcome the limitations of increasing cost, unmanned monitoring in a large area by combining a sensor network with multiple sensors and artificial intelligence that can analyze the characteristics of a target in a short time to provide excellent scalability of various monitoring spaces And the radiation unmanned monitoring and automatic alarm system that can detect the inflow of radioactive material is designed.

In addition, the existing radiation monitoring system provided only a simple fusion image of the radiation image and the optical image, but the proposed unmanned monitoring and automatic alarm system applies an artificial intelligence-based object recognition algorithm to the optical image to automatically It is possible to track and monitor the movement path of an object.

And because the present invention is an unmanned radiation monitoring and automatic alarm system that can detect the inflow of radioactive materials and unmanned monitoring of a large area by combining a sensor network and artificial intelligence, the sensitivity uniformity characteristics in the measurement space are improved and location tracking is improved. It is possible to automatically track and monitor the movement path of an object in real time by improving the accuracy and resolution at the outskirts.

Although the present invention has been described in detail with respect to the described embodiments, it is obvious to those skilled in the art that various changes and modifications are possible within the scope of the technical spirit of the present invention, and it is natural that such variations and modifications belong to the appended claims.

110: radiation detector 120: analog signal processing unit
130: digital signal processing unit 140: control unit
150: image processing unit

Claims (12)

a radiation detection unit including an optical sensor that detects radiation emitted from an object (radioactive material) and converts the converted flash signal into an electrical signal;
an analog signal processing unit comprising a signal amplifier and comparator for multiplexing and amplifying a plurality of optical sensor output signals from the radiation detection unit and adjusting a signal rise time, a fall time, a signal width, or an offset voltage;
Digital that measures the simultaneous coefficient by comparing the multiplexed analog signal processing unit output signal through ADC, TDC or FPGA (Xilinx family, Altera family) to acquire response position, response size, and response time information, and compare the data acquired within a specific time signal processing unit; and
a data processing unit that operates as a machine learning data processing unit and tracks the position by receiving digitized output information from the digital signal processing unit and determining the radiation response and detection position using an artificial intelligence algorithm NN (Neural networks);
A nuclear facility unmanned monitoring and artificial intelligence-based automatic warning system, including
The method according to claim 1,
The flash signal is
A nuclear facility unmanned monitoring and artificial intelligence-based automatic warning system, characterized in that it is an output signal of a single or multi-layered flash crystal.
3. The method according to claim 2,
The radiation detector
A photodetector comprising a radiation detector that detects radiation emitted from an object and converts it into a flash signal, and an optical sensor that converts the output signal of the single-layer or multi-layered flash crystal output from the radiation detector into an electrical signal and outputs it Unmanned monitoring of nuclear facilities and automatic warning system based on artificial intelligence.
4. The method according to claim 3,
The radiation detector
Through multiple radiation detectors combined with scintillation crystals and optical sensors, a difference in radiation counts occurs depending on the detected location, and the count ratio detected by each radiation detector is analyzed with an artificial intelligence-based algorithm to automatically track or alert. Unmanned monitoring of nuclear facilities and automatic warning system based on artificial intelligence.
5. The method according to claim 4,
The radiation detector
Install a number of units close to or far away from the required place, each with an identification code, and when transmitting a detection signal, transmit the identification code along with the identification code to the control center so that the installed location is displayed, and when an event occurs, the location is determined through the screen at the control center Unmanned monitoring of nuclear facilities and automatic warning system based on artificial intelligence for easy identification.
The method according to claim 1,
an image detection unit that receives CCD images, identifies radioactivity deviating from a specific location, and automatically tracks and alarms the detected count ratio of each radiation detector according to conditions using an artificial intelligence-based artificial intelligence algorithm (convolution neural networks);
A nuclear facility unmanned monitoring and artificial intelligence-based automatic warning system further comprising a.
7. The method of claim 6,
The artificial intelligence algorithm (convolution neural networks) is
It is an algorithm that identifies and classifies the features of images in unstructured data such as 2-dimensional images. It is an image with correct answers (does not deviate from a specific position) and non-correct images (radioactivity is present from a specific position). Nuclear facility unmanned monitoring and artificial intelligence-based automatic warning system, characterized in that it is an artificial intelligence algorithm for automatic monitoring that identifies radioactivity deviating from a specific location by learning the correct and non-correct images in advance to classify them.
8. The method of claim 7,
The artificial intelligence algorithm is
In a deep neural network, the location of the radioactive material is determined by using an artificial neural network model that learns data through one input layer and one output layer, recognizes or infers patterns, and applies the number of radiation signals determined by the radiation detector to the model as input layer data. A nuclear facility unmanned monitoring and artificial intelligence-based automatic warning system, characterized in that the result of determining whether it is detected in the output layer is calculated as a value.
9. The method of claim 8,
The artificial intelligence network algorithm is
In order to recognize the image and find the pattern, it receives the CCD image as input data, performs a convolution operation through the pixels of the image, recognizes the image, and finds the pattern. A nuclear facility unmanned monitoring and artificial intelligence-based automatic warning system that learns patterns and operates to detect movement when there is movement.
10. The method of claim 9,
The artificial intelligence network algorithm is
The artificial neural network model classification result and the CNN model training result
A nuclear facility unmanned monitoring and artificial intelligence-based automatic warning system that displays a loss function graph that shows how much the predicted value of the model differs from the actual value and a correct classification graph that shows the target value.
11. The method of claim 10,
The artificial intelligence network algorithm is
As described above, in the present invention, the number of sensor networks optimized according to the measurement space is operated, the response characteristics according to the radiation response characteristics and the size and position of the detector are analyzed, and the position of the system is established as a database by DB learning using artificial intelligence. resolution can be improved.
The method according to claim 1,
The data processing unit
A nuclear facility unmanned monitoring and artificial intelligence-based automatic alarm system, characterized in that the detected counting ratio of each detector is automatically tracked and alerted according to conditions using an artificial intelligence-based algorithm.

Images