Inference device, inference system, and facility maintenance system

Abstract

To provide an inference device capable of rapidly and accurately inferring the operating state of a radiation detection device.SOLUTION: An inference device 40 comprises: data acquisition units 41A, 42A which acquire, as learning data, a state signal S output from a radiation detection device 10 that indicates the operating state of the radiation detection device 10; and an inference unit 42B which uses a learned model that is constructed by machine learning based on a feature amount of the state signal S and infers the operating state of the radiation detection device 10 to output the operating state of the radiation detection device 10 on the basis of the state signal S acquired by the data acquisition unit 42A.SELECTED DRAWING: Figure 6

Description

[Technical Field]

[0001] The present disclosure relates to an inference device, an inference system, and an equipment maintenance system. [Background technology]

[0002] Radiation monitoring equipment as an equipment maintenance system is installed within the premises of a nuclear power plant or its surrounding area, and is used to manage radiation from radioactive materials and to detect abnormalities occurring in the nuclear power plant early. Due to the importance of radiation monitoring equipment, reliable monitoring performance for nuclear power plants is required. In addition, since radiation monitoring equipment has a large number of components such as radiation detectors, and the signals output from each component are complex and numerous, there is also a demand for simplification of the maintenance method for radiation monitoring equipment. Conventionally, radiation monitoring devices have been disclosed as equipment maintenance systems, such as those described below, which accurately determine the soundness of radiation monitoring devices and simplify the determination process.

[0003] That is, conventional radiation monitoring devices are basically configured with a computer that calculates the statistics of detector signals and a multi-pulse height analyzer that performs spectral measurements. They detect abnormal fluctuations in measured values and simultaneously measure spectral changes according to a sequence appropriate to the type of detector to confirm the integrity of the device. A waveform observation device is also added to the configuration to check for abnormalities in the signal itself. This allows for early detection of abnormalities by identifying whether the abnormality is due to the measurement target (radiation) or abnormal fluctuations in the device's characteristics (see, for example, Patent Document 1). [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 6-324158 Summary of the Invention [Problem to be solved by the invention]

[0005] The conventional radiation monitoring devices described above measure changes in the spectrum according to a sequence corresponding to the type of detectors to confirm the soundness of the device. However, when there are a large number of detectors, or when the signals output from each detector are numerous and complex, determining the soundness of the radiation monitoring device becomes complicated, and there are problems such as an inability to accurately determine the soundness or the time required to make the determination. The present disclosure discloses technology for solving the above-mentioned problems, and aims to provide an inference device, an inference system, and an equipment maintenance system that can simply and accurately infer the health of a radiation detection device. [Means for solving the problem]

[0006] The inference device of the present disclosure comprises: a data acquisition unit that acquires, as learning data, a status signal that is output from the radiation detection device and indicates an operating status of the radiation detection device; an inference unit that outputs the operating status of the radiation detection device based on the status signal acquired by the data acquisition unit, using a trained model that infers the operating status of the radiation detection device, the trained model being constructed by machine learning based on the feature amount of the status signal; It is something. The inference system of the present disclosure also includes: An inference device configured as described above; a display unit that acquires the operating status of the radiation detection device output by the inference device via a network and displays the acquired operating status, It is something. In addition, the equipment maintenance system of the present disclosure includes: a plurality of the radiation detection devices; a data collection unit that collects the status signals output from the plurality of radiation detection devices; The inference system configured as described above; It is something. [Effects of the Invention]

[0007] According to the inference device, inference system, and equipment maintenance system disclosed herein, an inference device, inference system, and equipment maintenance system can be obtained that can simply and accurately infer the health of a radiation detection device. [Brief explanation of the drawings]

[0008] [Figure 1] 1 is a block diagram showing a schematic configuration of a monitoring device according to a first embodiment. [Figure 2] 5 is a diagram showing an example of a status signal indicating the operating status of the radiation detection apparatus in accordance with the first embodiment. FIG. [Figure 3] 1 is a diagram showing a schematic configuration of a learning device provided in the inference device according to the first embodiment. [Figure 4] FIG. 2 is a diagram showing a neural network of the inference device according to the first embodiment. [Figure 5] FIG. 3 is a diagram showing a processing flow performed by the learning device according to the first embodiment. [Figure 6] 2 is a diagram showing a schematic configuration of an inference function unit included in the inference device according to the first embodiment. FIG. [Figure 7] 10 is a diagram showing a processing flow performed by an inference function unit according to the first embodiment. FIG. [Figure 8] 3 is a diagram for explaining the estimation of an abnormality factor performed by the inference device according to the first embodiment. FIG. [Figure 9] Fig. 9A is a diagram showing an example of a pulse height spectrum as a status signal according to Embodiment 1. Fig. 9B is a diagram showing an example of the shape of a changed pulse height spectrum. [Figure 10] FIG. 2 is a diagram illustrating a hardware configuration of a control device of the estimation device according to the first embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0009] Embodiment 1 FIG. 1 is a block diagram showing a schematic configuration of a monitoring device 100 as a facility maintenance system according to the first embodiment. The monitoring device 100 monitors the operating states of the radiation detection devices 10A, 10B, 10C, and 10D to determine their soundness.

[0010] The monitoring device 100 includes a plurality of radiation detection devices 10A, 10B, 10C, and 10D installed within the nuclear plant premises or the surrounding area, a data collection system 20 that collects status signals S output from each of the radiation detection devices 10A, 10B, 10C, and 10D, and an inference system 50 that infers and outputs the operating status of each of the radiation detection devices 10A, 10B, 10C, and 10D based on the status signals S. In the following description, when there is no need to distinguish between the radiation detection devices 10A, 10B, 10C, and 10D, they will simply be referred to as the radiation detection device 10.

[0011] First, the configuration of the radiation detection device 10 will be described. The radiation detection devices 10A, 10B, 10C, and 10D will be described assuming that they have the same configuration. The radiation detection device 10 includes a detection unit 11, an amplifier unit 14, a sampling unit 12, a signal processing unit 13, and a temperature sensor 15 as components that make up the radiation detection device 10.

[0012] When radiation emitted from a radioactive substance (not shown) is incident on the detection unit 11, the detection unit outputs a detection signal having a pulse height corresponding to the energy value of the incident radiation, and for example, a scintillation detector or a semiconductor detector is used. The amplifier 14 amplifies the minute detection signal output from the detector 11 and transmits it to the signal processor 13 .

[0013] The sampling unit 12 collects water, air, and gas as media containing radioactive materials, and is configured with a pump, valve, control device (PLC: programmable logic controller), and temperature controller for collecting the media.

[0014] The signal processing unit 13 has a function of converting the received detection signal into a radiation level signal indicating the radiation energy value. The signal processing unit 13 also has a function of operating the sampling unit 12, a diagnostic function of diagnosing whether or not there is an abnormality in the sampling unit 12, and a self-diagnostic function of diagnosing whether or not there is an abnormality in the hardware and software that constitute the signal processing unit 13 itself.

[0015] The signal processing unit 13 outputs a radiation level signal indicating the energy value of the radiation as a status signal S indicating the operating status of the radiation detection device 10. This status signal S will be explained below.

[0016] FIG. 2 is a diagram showing an example of the status signal S indicating the operating status of the radiation detection device 10. As shown in FIG. As shown in FIG. 2, the signal processing unit 13 outputs, as a status signal S indicating the operating state of the radiation detection device 10, not only the radiation level signal but also the pulse height in the energy pulse height distribution indicating the count for each energy value of the detection signal from the detection unit 11, the detector signal waveform (raw waveform), and detection signals from an alarm, a DC power supply, a DC power supply fan, a flow meter, a pressure gauge, a thermometer, etc., which are components that make up the signal processing unit 13 and the radiation detection device 10.

[0017] In this way, the status signal S output by the radiation detection device 10 includes a radiation level signal measured by the radiation detection unit 11 reacting to radiation present within the nuclear plant premises or surrounding area, detection signals from each detector, etc., which are components that make up the radiation detection device 10, etc.

[0018] The data collection system 20 collects the status signals S output from the radiation detection devices 10A, 10B, 10C, and 10D, and stores them in an internal data lake.

[0019] Next, the operation of the inference system 50 will be described. The inference system 50 is connected to the data collection system 20 via a network 30, which may be the Internet or a dedicated line. The inference system 50 includes an inference device 40 that determines the health of the radiation detection device 10 using the status signal S accumulated by the data collection system 20, and a display unit 35 that displays the results determined and output by the inference device 40.

[0020] The inference device 40 has a status monitoring function, an abnormality detection function, a judgment function, an abnormality cause estimation function, and an inspection management function for monitoring the operating status of the radiation detection device 10 using the status signal S collected by the data collection system 20.

[0021] The status monitoring function is a function for constantly monitoring the operating status of the radiation detection device 10 . The abnormality detection function is a function that detects that the radiation detection device 10 is behaving differently from its normal operating state and may be malfunctioning. The determination function is a function for determining whether the cause of the abnormality is due to a change in the surrounding environment in which the radiation detection device 10 is installed, or due to a component that configures the radiation detection device 10. The abnormality factor estimation function is a function that, when the abnormality factor is in a component, identifies which component among the multiple components that make up the radiation detection device 10 the abnormality factor is. The inspection management function is a function that identifies the possible extension period from the manufacturer's recommended inspection period and the parts that require inspection based on the past inspection cycles and inspection periods of the components and the current status information of the radiation detection device 10.

[0022] As described above, the inference device 40 of this embodiment has the above-mentioned functions, determines the soundness of the radiation detection device 10 to be monitored, and infers its operating state. The operating status of the radiation detection device 10 inferred by the inference device 40 is displayed on the display unit 35 via the network 30. This enables power company maintenance personnel and manufacturer personnel to monitor the radiation detection device 10 from remote locations and formulate maintenance plans.

[0023] The configuration of the inference device 40 will now be described. FIG. 3 is a diagram showing a schematic configuration of a learning device 41 included in an inference device 40 according to this embodiment. FIG. 4 is a diagram showing the neural network of the inference device 40 of this embodiment. FIG. 5 is a diagram showing a processing flow performed by learning device 41 of this embodiment. FIG. 6 is a diagram showing a schematic configuration of the inference function unit 42 included in the inference device 40 of this embodiment. FIG. 7 is a diagram showing a processing flow performed by the inference function unit 42 of this embodiment.

[0024] The inference device 40 includes a learning device 41 shown in FIG. 3, an inference function unit 42 shown in FIG. 6, and a trained model storage unit 41C shown in both FIGS. The inference device 40 configured in this manner infers and outputs the operating state of the radiation detection device 10 by machine learning using AI (Artificial Intelligence) as follows. First, a case where inference device 40 performs machine learning through supervised learning will be described. Here, supervised learning refers to a technique in which a learning device is provided with a set of input and result (label) data, which allows the device to learn the features of the learning data and infer the result from the input.

[0025] <Learning Phase> The processing performed by the learning device 41 included in the inference device 40 will be described below. As shown in FIG. 3, the learning device 41 includes a data acquisition unit 41A and a model generation unit 41B.

[0026] The data acquisition unit 41A acquires the status signal S output from the radiation detection device 10 as learning data as an input 1. Furthermore, the data acquiring unit 41A acquires the operating state (correct answer) of the radiation detection device 10 corresponding to the feature amount of the state signal S as learning data as the input 2. The learning data (correct answer) as input 2 is, for example, a status signal S (correct answer) indicating a normal state of the radiation detection device 10, a status signal S (correct answer) indicating an abnormal state of the radiation detection device 10, or a status signal S (correct answer) indicating an abnormal symptom of the radiation detection device 10.

[0027] The model generation unit 41B learns the output based on the learning data created based on the combination of the input 1 and input 2 (correct answer) output from the data acquisition unit 41A. That is, a learned model that infers the optimal operating state of the radiation detection device 10 is generated from the feature amounts of the input 1 and input 2 (correct answer) that indicate the state signal S of the radiation detection device 10. Here, the learning data is data in which input 1 and input 2 (correct answer) are associated with each other.

[0028] As an example of the learning algorithm used by the model generating unit 41B, a case where a neural network is applied will be described. The model generation unit 41B learns the output by so-called supervised learning in accordance with the neural network model. A neural network consists of an input layer consisting of multiple neurons, an intermediate layer (hidden layer) consisting of multiple neurons, and an output layer consisting of multiple neurons. The intermediate layer may be one layer, or two or more layers.

[0029] For example, in a three-layer neural network like the one shown in Figure 4, when multiple inputs are input to the input layer (X1-X3), the values are multiplied by weight W1 (w11-w16) and input to the middle layer (Y1-Y2), and the result is further multiplied by weight W2 (w21-w26) and output from the output layer (Z1-Z3). This output result changes depending on the values of weights W1 and W2.

[0030] In the inference device 40 of this embodiment, as described above, the neural network learns the output through supervised learning in accordance with the learning data created based on the combination of input 1 and input 2 (correct answer) acquired by the data acquisition unit 41A. In other words, a neural network learns by inputting input 1 into the input layer and adjusting the weights W1 and W2 so that the result output from the output layer approaches input 2 (the correct answer).

[0031] The model generation unit 41B generates and outputs a trained model by executing the above-described learning. The trained model storage unit 41C stores the trained model output from the model generation unit 41B.

[0032] Next, the learning process performed by the learning device 41 will be described with reference to FIG. In step b1, the data acquisition unit 41A acquires input 1 and input 2 (correct answer). Note that input 1 and input 2 (correct answer) are not limited to being acquired simultaneously, and as long as input 1 and input 2 (correct answer) are entered in association with each other, the data for input 1 and input 2 (correct answer) may be acquired at different times.

[0033] In step b2, the model generation unit 41B learns the output through supervised learning in accordance with the learning data created based on the combination of input 1 and input 2 (correct answer) acquired by the data acquisition unit 41A, and generates a learned model. In step b3, the trained model storage unit stores the trained model generated by the model generation unit.

[0034] <Utilization phase> The processing performed by the inference function unit 42 included in the inference device 40 will be described below. As shown in FIG. 6, the inference function unit 42 includes a data acquisition unit 42A and an inference unit 42B. The data acquisition unit 42A acquires the state signal S from the radiation detection device 10 as an input 1. The inference unit 42B infers the output obtained by using the trained model 41C. That is, by inputting the input 1 acquired by the data acquisition unit 42A to this trained model, the operating state of the radiation detection device 10 inferred from the input 1 can be output.

[0035] In this embodiment, the operating status of the radiation detection device 10 is described as being output using a trained model trained in the model generation unit 41B, but the trained model may also be obtained from outside the inference device 40, and the operating status may be output based on this trained model.

[0036] Next, the processing by the inference function unit 42 will be described with reference to FIG. In step c1, the data acquiring unit 42A acquires the input 1 as the state signal S. In step c2, the inference unit 42B inputs the input 1 to the learned model stored in the learned model storage unit 41C, and obtains an output indicating the operating state of the radiation detection device 10. In step c3, the inference unit 42B outputs the output obtained by the trained model.

[0037] The model generating unit 41B may learn the output in accordance with learning data created for a plurality of radiation detection devices 10. Furthermore, the model generation unit 41B may acquire learning data from a plurality of radiation detection devices 10 used within the same site of the nuclear plant, or may learn the output by using, as learning data, status signals S collected from a plurality of radiation detection devices 10 operating independently at different sites of the nuclear plant.

[0038] Furthermore, it is also possible to add or remove other radiation detection devices 10 other than the radiation detection devices 10A, 10B, 10C, and 10D that collect learning data to or from the monitoring targets during monitoring. Furthermore, a learning device that has learned the output of a certain radiation detection device 10A may be applied to other radiation detection devices 10B, 10C, and 10D, and the output of the other radiation detection devices 10B, 10C, and 10D may be re-learned and updated.

[0039] In addition, the learning algorithm used in the model generation unit can be deep learning, which learns to extract the features themselves, or machine learning can be performed according to other known methods, such as genetic programming, functional logic programming, and support vector machines.

[0040] By using the machine learning described above, the inference device 40 of this embodiment can easily and accurately determine the health of the radiation detection device 10 and accurately monitor its operating status without increasing costs, even if the radiation detection device 10 has a large number of components or if the signals output from each component are complex and numerous.

[0041] The inference device 40 is not limited to a configuration independent of the radiation detection device 10, such as being connected to the radiation detection device 10 via the network 30 as shown in Fig. 1. The inference device 40 may be built into the radiation detection device 10, or may exist on a cloud server.

[0042] Furthermore, the data acquisition unit 41A of the learning device 41 may acquire the operating status of each of the multiple components, such as the detectors, that make up the radiation detection device 10 for each component as a status signal S. The inference device 40 may then derive features for each component and output the operating status of each component using a trained model constructed by machine learning based on each feature. This allows the cause of the abnormality to be identified at the component level of the radiation detection device 10. Furthermore, if an abnormality occurs in the radiation monitoring device, the location of the failure in the radiation detection device 10 can be estimated before the failure occurs. This makes it possible to take preventative measures and intensively monitor locations where failure is likely to occur, and if a failure does occur, it becomes possible to quickly identify the cause.

[0043] In addition, when the inference device 40 outputs an abnormality or an abnormal symptom of the radiation detection device 10 as the operating status of the radiation detection device 10, it may select and output the component from among the multiple components for which the abnormality or abnormal symptom has been inferred. This allows power company maintenance personnel and manufacturer personnel to quickly identify abnormalities or components where abnormalities may occur, thereby making maintenance work more efficient.

[0044] The data acquisition unit 41A may also acquire a plurality of status signals S at set time intervals. The inference device 40 may then output the operating status of the radiation detection device 10 using a trained model constructed by machine learning based on feature amounts derived in accordance with the plurality of acquired status signals S. In this way, the analysis results of the feature quantities indicated by the time-series data are used as objective indicators in formulating a maintenance plan, which allows for understanding the deterioration trend of the radiation detection device 10. In this way, it becomes possible to detect signs of abnormalities in the device before they occur, enabling predictive maintenance that allows for countermeasures to be taken before an abnormality occurs.

[0045] Furthermore, if time-series data of the features of each component, such as a detector, is used, it becomes possible to grasp the deterioration trend at the component level. This allows for the setting of whether or not each component needs to be inspected and replaced, enabling efficient maintenance activities. Furthermore, the inference device 40 may use maintenance data that records the past inspection cycles of each component to select and output components that require inspection from among multiple components, thereby further improving the efficiency of maintenance work.

[0046] Furthermore, the inference device 40 may use a trained model constructed by supervised learning that associates abnormality factors with feature quantities when the radiation detection device 10 is in an abnormal operating state. This allows, for example, when an abnormality in pressure is detected as the status signal S shown in Fig. 2, to identify the abnormality factor that caused this pressure abnormality.

[0047] FIG. 8 is a diagram for explaining the estimation of the abnormality cause performed by the inference device 40. In FIG. In addition, as shown in FIG. 8, in estimating the cause of an abnormality, a method for improving estimation accuracy may be used by adding design data, past trouble information, etc. as input data, such as a tree analysis (FT diagram: Fault tree) that analyzes the causal relationship between an abnormality in the operating state of the radiation detection device and the impact of the abnormality.

[0048] In addition, the inference device 40 may use, as the state signal S, the learned model constructed by supervised learning in which abnormality information is attached to feature quantities when environmental state information indicating the environment in which the radiation detection device 10 is installed indicates an abnormal environment. Because the detection signal output by the detection unit 11 has a minute level, the radiation level signal may fluctuate due to changes in the installation environment caused by temperature, external noise, etc., and may exhibit unusual behavior. By detecting an abnormality in the environmental status information indicating the environment in which the radiation detection device 10 is installed, the inference device 40 can confirm, for example, that the cause of the abnormality is not the radiation detection device 10 but the external environment. This makes it possible to prevent unnecessary inspections of the radiation detection device 10 and the resulting decrease in operation rate.

[0049] Although the above description has been given of an example in which the inference device 40 performs supervised learning, the present invention is not limited to this. The inference device 40 may output the operating status of the radiation detection device 10 using a trained model constructed by unsupervised learning that learns the feature quantities of the status signal S from the radiation detection device 10 in a normal state. This allows the operating state of the radiation detection device 10 to be determined based on a wide range of determination indices such as the structure and tendency of the acquired status signal S, even when there is no direct index for detecting an abnormality, for example.

[0050] Furthermore, with the inference system 50 of this embodiment, both the power maintenance personnel and the manufacturer's personnel who are familiar with the device structure can remotely check the details of the device status of the radiation detection device 10 via the network 30. This allows them to quickly determine how to deal with the problem and take action, leading to early resolution of the abnormality.

[0051] Hereinafter, a case will be described in which a pulse-height spectrum indicating the count for each energy value of the detection signal output by the detection unit 11 included in the radiation detection device 10 is used as the status signal S. FIG. 9A is a diagram showing an example of a pulse height spectrum as the status signal S. In FIG. FIG. 9B is a diagram showing an example of the changed shape of the wave height spectrum.

[0052] As shown in Figure 9A, the pulse height spectrum detects peaks corresponding to the detected radionuclide, and if there is a change in the external environment or an abnormality occurs in the radiation detector, the shape of the pulse height spectrum will change as shown in Figure 9B. Therefore, power maintenance personnel and manufacturer personnel can determine the soundness of the radiation monitoring device by observing the long-term change trend of the pulse height spectrum over time.

[0053] The inference device 40 can also reduce the data volume by using data that has been preprocessed from time-series data. For example, in the case of a pulse-height spectrum, only the peak position is used instead of the entire spectrum. Alternatively, the pulse height value, pulse width, etc. are used instead of the raw waveform of the detector signal. Alternatively, for temperature, pressure, flow rate, etc., the average value or moving average value over a given time period is used. By performing this type of preprocessing, it is possible to reduce the amount of data and shorten the learning time of the AI.

[0054] The hardware configuration of the control device of the inference device 40 will now be described. The control device 1 is configured to have a processor 1A and a storage device 1B, as shown in an example of hardware in Fig. 10. The storage device 1B includes a volatile storage device such as a random access memory and a non-volatile auxiliary storage device such as a flash memory, both of which are not shown.

[0055] Also, a hard disk auxiliary storage device may be provided instead of flash memory. Processor 1A executes a program input from storage device 1B. In this case, the program is input from the auxiliary storage device to processor 1A via a volatile storage device. Processor 1A may output data such as calculation results to a volatile storage device of storage device 101, or may store the data in the auxiliary storage device via the volatile storage device.

[0056] Although the present disclosure describes exemplary embodiments, the various features, aspects, and functions described in the embodiments are not limited to application to a particular embodiment, but may be applied to the embodiments alone or in various combinations. Therefore, countless variations not illustrated are contemplated within the scope of the technology disclosed in this specification, including, for example, modifying, adding, or omitting at least one component.

[0057] Various aspects of the present disclosure are summarized below as appendices. (Appendix 1) a data acquisition unit that acquires, as learning data, a status signal that is output from the radiation detection device and indicates an operating status of the radiation detection device; an inference unit that outputs the operating status of the radiation detection device based on the status signal acquired by the data acquisition unit, using a trained model that infers the operating status of the radiation detection device, the trained model being constructed by machine learning based on the feature amount of the status signal; Reasoning device. (Appendix 2) The data acquisition unit acquiring an operating state of each of a plurality of components constituting the radiation detection device as the status signal for each of the components; The inference unit outputting an operating status of each of the components as an operating status of the radiation detection device based on each of the status signals acquired by the data acquisition unit, using the trained model constructed by machine learning based on the feature amounts derived for each of the components constituting the radiation detection device; 10. The inference apparatus of claim 1. (Appendix 3) The data acquisition unit acquiring a plurality of status signals from the radiation detection device at set time intervals; The inference unit outputting the operating status of the radiation detection device using the trained model constructed by machine learning based on the feature amounts derived in accordance with each of the status signals; 10. The inference apparatus of claim 2. (Appendix 4) The inference unit outputting the abnormality factor of the radiation detection device based on the status signal acquired by the data acquisition unit, using the trained model constructed by supervised learning that associates abnormality factors with the feature amounts when the operating status of the radiation detection device is in an abnormal state; 10. The inference apparatus of claim 3. (Appendix 5) The inference unit outputting whether or not an abnormality has occurred in the environment in which the radiation detection device is disposed, using the trained model constructed by supervised learning in which abnormality information is assigned to the feature amount when environmental state information indicating the environment in which the radiation detection device is disposed as the state signal is abnormal; 10. The inference apparatus of claim 4. (Appendix 6) The inference unit outputting an operating status of the radiation detection device based on the status signal acquired by the data acquisition unit, using the trained model constructed by unsupervised learning that learns the feature amount of the status signal from the radiation detection device in a normal state; 10. The inference device according to claim 1, wherein: (Appendix 7) The inference unit When an abnormality or an abnormal symptom of the radiation detection device is inferred and output as the operating status of the radiation detection device, selecting and outputting, from among the plurality of components, a component for which an abnormality or an abnormal symptom has been inferred, using the trained model constructed by machine learning based on the feature amounts for each of the components constituting the radiation detection device; 10. The inference apparatus of claim 2. (Appendix 8) The trained model is The system is constructed by incorporating a tree analysis that analyzes the causal relationship between an abnormality in the operating state of the radiation detection device and the effect of the abnormality. 10. The inference device of claim 4 or 5. (Appendix 9) The inference unit using the trained model constructed by machine learning based on the feature amounts of each of the components constituting the radiation detection device, selecting and outputting the components that require inspection from among the plurality of components based on the environmental state information and maintenance data that records past inspection cycles of each of the components; 10. The inference device of claim 5. (Appendix 10) The inference unit outputting a normal state, an abnormal state, or an abnormality sign of the radiation detection device as the operating state of the radiation detection device; 10. The inference device of any one of Supplementary Note 1 to Supplementary Note 9. (Appendix 11) The status signal is the radiation detection device includes at least one of an energy distribution indicating a count for each energy value of radiation emitted from a radioactive substance, a radiation intensity derived based on the energy distribution, and a signal output as the status signal by each of a plurality of components constituting the radiation detection device, 11. The inference device of any one of Supplementary Note 1 to Supplementary Note 10. (Appendix 12) An inference device according to any one of Supplementary Note 1 to Supplementary Note 11; a display unit that acquires the operating status of the radiation detection device output by the inference device via a network and displays the acquired operating status, Inference system. (Appendix 13) a plurality of the radiation detection devices; a data collection unit that collects the status signals output from the plurality of radiation detection devices; the inference system of claim 12; Equipment maintenance system. [Explanation of symbols]

[0058] 10,10A,10B,10C,10D Radiation detection device, 30 Network, 35 display unit, 40 inference device, 42B inference unit, 50 inference system, 100 Monitoring device (equipment maintenance system).

Claims

1. a data acquisition unit that acquires, as learning data, a status signal that is output from the radiation detection device and indicates an operating status of the radiation detection device; an inference unit that outputs the operating status of the radiation detection device based on the status signal acquired by the data acquisition unit, using a trained model that infers the operating status of the radiation detection device, the trained model being constructed by machine learning based on the feature amount of the status signal; Reasoning device.

2. The data acquisition unit acquiring an operating state of each of a plurality of components constituting the radiation detection device as the status signal for each of the components; The inference unit outputting an operating status of each of the components as an operating status of the radiation detection device based on each of the status signals acquired by the data acquisition unit, using the trained model constructed by machine learning based on the feature amounts derived for each of the components constituting the radiation detection device; The inference device according to claim 1 .

3. The data acquisition unit acquiring a plurality of status signals from the radiation detection device at set time intervals; The inference unit outputting the operating status of the radiation detection device using the trained model constructed by machine learning based on the feature amounts derived in accordance with each of the status signals; The inference device according to claim 2 .

4. The inference unit outputting the abnormality factor of the radiation detection device based on the status signal acquired by the data acquisition unit, using the trained model constructed by supervised learning that associates abnormality factors with the feature amounts when the operating status of the radiation detection device is in an abnormal state; The inference device according to claim 3 .

5. The inference unit outputting whether or not an abnormality has occurred in the environment in which the radiation detection device is disposed, using the trained model constructed by supervised learning in which abnormality information is assigned to the feature amount when environmental state information indicating the environment in which the radiation detection device is disposed as the state signal indicates an abnormal environment; The inference device according to claim 4.

6. The inference unit outputting an operating status of the radiation detection device based on the status signal acquired by the data acquisition unit, using the trained model constructed by unsupervised learning that learns the feature amount of the status signal from the radiation detection device in a normal state; The inference device according to claim 1 .

7. The inference unit When an abnormality or an abnormal symptom of the radiation detection device is inferred and output as the operating status of the radiation detection device, selecting and outputting, from among the plurality of components, a component for which an abnormality or an abnormal symptom has been inferred, using the trained model constructed by machine learning based on the feature amounts for each of the components constituting the radiation detection device; The inference device according to claim 2 .

8. The trained model is The system is constructed by incorporating a tree analysis that analyzes the causal relationship between an abnormality in the operating state of the radiation detection device and the effect of the abnormality. The inference device according to claim 4.

9. The inference unit using the trained model constructed by machine learning based on the feature amounts of each of the components constituting the radiation detection device, selecting and outputting the components that require inspection from among the plurality of components based on the environmental state information and maintenance data that records past inspection cycles of each of the components; The inference device according to claim 5 .

10. The inference unit outputting a normal state, an abnormal state, or an abnormality sign of the radiation detection device as the operating state of the radiation detection device; The inference device according to claim 1 .

11. The status signal is the radiation detection device includes at least one of an energy distribution indicating a count for each energy value of radiation emitted from a radioactive substance, a radiation intensity derived based on the energy distribution, and a signal output as the status signal by each of a plurality of components constituting the radiation detection device, The inference device according to claim 1 .

12. An inference device according to any one of claims 1 to 11; a display unit that acquires the operating status of the radiation detection device output by the inference device via a network and displays the acquired operating status, Inference system.

13. a plurality of the radiation detection devices; a data collection unit that collects the status signals output from the plurality of radiation detection devices; The inference system of claim 12. Equipment maintenance system.

Images