Nuclear instrumentation device

Abstract

A nuclear instrumentation device (100) comprises: a waveform discrimination circuit (41) that extracts a neutron-derived pulse signal from a pulse signal detected by a detector (1); a pulse count circuit (42) that periodically counts the neutron-derived pulse signal; a count rate conversion circuit (43) that converts the pulse count value into a count rate; a zero count circuit (45) that calculates a continuation period in which the pulse count value of the pulse count circuit (42) becomes zero; a time constant selection circuit (46) that selects a time constant according to the period calculated by the zero count circuit (45) when the pulse count value in the pulse count circuit (42) is zero, and that selects a larger time constant as the count rate decreases and a smaller time constant as the count rate increases when the pulse count value is not zero; and an integration circuit (47) that calculates the count rate of neutrons in each period by using the selected time constant, the abovementioned circuits being composed of digital circuits.

Description

nuclear instrumentation device

[0001] The present disclosure relates to nuclear instrumentation.

[0002] The neutron detector detects neutrons emitted from the reactor, and the nuclear instrumentation device measures the pulse signal output from the neutron detector, thereby obtaining information useful for understanding the reactor's power output and detecting abnormal increases in power. Based on this information, the monitoring and control device controls the reactor's power output and other aspects.

[0003] In conventional nuclear instrumentation devices, pulse signal measurement is realized using multiple analog circuits, which results in a complex and expensive device configuration, a lack of convenience in setting thresholds for pulse height discrimination, which is a function for distinguishing between neutron-derived signals and other signals, and a long inspection period due to the need to check the operation of multiple analog circuits during inspection. As a means to solve these problems, there is an example of a pulse measurement device that uses digital circuits (see, for example, Patent Document 1).

[0004] The device disclosed in Patent Document 1 includes a means for outputting a pulse signal in response to neutrons, a means for amplifying the pulse signal, a means for converting the amplified pulse signal into a digital signal and extracting only a specific signal based on the pulse height value of the pulse, and a means for calculating and outputting a pulse count value, and further discloses that the output signal is divided into a pulse domain and a Campbell domain. By converting the pulse signal output from the preamplifier into a digital signal, performing pulse height discrimination using a digital circuit, and calculating and outputting the count rate, it is possible to measure neutron pulses over a wide range with a simple configuration.

[0005] Japanese Patent Application Laid-Open No. 2016-206004

[0006] The range of pulse measurements in reactor monitoring spans a six-digit range. Furthermore, statistical fluctuations must be taken into consideration when observing neutrons. Therefore, highly reliable measurements that suppress statistical fluctuations are required at low count rates, while fast response is required at high count rates to detect reactor power fluctuations and power increases early. However, Patent Document 1 does not specify a specific method for achieving this.

[0007] The present disclosure discloses a technology for solving the above-mentioned problems, and aims to achieve both reliable measurements at low count rates and fast response at high count rates in a nuclear instrumentation device equipped with a digital circuit.

[0008] The nuclear instrumentation device of the present disclosure is a nuclear instrumentation device that calculates a neutron counting rate from a pulse signal detected by a detector, comprising: a waveform discrimination circuit that converts the pulse signal into a digital signal and inputs it, and extracts a pulse signal derived from a neutron; a pulse counting circuit that periodically counts the number of times that the pulse signal derived from a neutron extracted by the waveform discrimination circuit arrives within a preset time; a memory unit that stores a pulse count value of a period immediately preceding the pulse count value counted by the pulse counting circuit; a counting rate conversion circuit that calculates fluctuations in the pulse signal using the pulse count value of a current period counted by the pulse counting circuit and the pulse count value of a period immediately preceding the current period read from the memory unit, and converts the pulse count value of a current period into a counting rate using a coefficient according to the fluctuations of the pulse signal; a zero counting circuit that increments a counter by one if the pulse count value counted by the pulse counting circuit is zero, and resets the counter if it is not zero, and calculates a period during which the pulse count value is zero; The counting device is provided with a time constant selection circuit which, when the pulse count value counted by the pulse counting circuit is zero, selects a time constant according to the period during which the pulse count value calculated by the zero counting circuit is zero, and, when the pulse count value counted by the pulse counting circuit is not zero, selects a larger time constant the smaller the counting rate converted by the counting rate conversion circuit, and selects a smaller time constant the larger the counting rate, and an integration circuit which calculates the neutron counting rate in each period using the time constant selected by the time constant selection circuit, and these circuits are configured as digital circuits.

[0009] The present disclosure makes it possible to provide a nuclear instrumentation device that achieves both reliability at low count rates and fast response at high count rates.

[0010] 1 is a configuration diagram for explaining the configuration of a nuclear reactor monitoring and control system according to a first embodiment. FIG. 2 is a block diagram showing the configuration of a nuclear instrumentation device according to the first embodiment. FIG. 3 is a diagram for explaining a coefficient for converting a pulse count to a count rate in the nuclear instrumentation device according to the first embodiment. FIG. 4 is a diagram showing the relationship between the count rate and a time constant used to calculate the output of the nuclear instrumentation device in the first embodiment. FIG. 5 is a diagram showing the relationship between the time constant used to calculate the output of the nuclear instrumentation device in the first embodiment and a period during which the pulse count is zero in the nuclear instrumentation device. FIG. 6 is a diagram showing an example of a pulse signal waveform before waveform discrimination in the nuclear counting device according to the second embodiment. FIG. 7 is a diagram showing the relationship between the maximum value of the pulse height and the pulse width in the nuclear counting device according to the second embodiment. FIG. 8 is a diagram showing the configuration of a learning device provided in a waveform discrimination circuit in the nuclear counting device according to the second embodiment. FIG. 9 is a diagram showing the configuration of an inference device provided in a waveform discrimination circuit in the nuclear counting device according to the second embodiment. FIG. 10 is a diagram showing the configuration of a learning device provided in a count rate conversion circuit in the nuclear counting device according to the third embodiment. FIG. 11 is a diagram showing the configuration of an inference device provided in a count rate conversion circuit in the nuclear counting device according to the third embodiment. FIG. 1 is a diagram showing the configuration of a learning device provided with a time constant selection circuit in a nucleus counting device according to embodiment 3. FIG. 2 is a diagram showing the configuration of an inference device provided with a time constant selection circuit in a nucleus counting device according to embodiment 3. FIG. 3 is a diagram showing an example of the hardware configuration of the nucleus counting devices according to embodiments 1 to 3. FIG. 4 is a diagram showing another example of the hardware configuration of the nucleus counting devices according to embodiments 1 to 3.

[0011] The present embodiment will be described below with reference to the drawings, in which the same reference numerals indicate the same or corresponding parts.

[0012] Embodiment 1. A nuclear instrumentation system according to embodiment 1 will be described below with reference to the drawings. Fig. 1 is a diagram for explaining the configuration of a nuclear reactor monitoring and control system using the nuclear instrumentation system according to embodiment 1. Fig. 2 is a block diagram showing the configuration of the nuclear instrumentation system according to embodiment 1.

[0013] <Configuration of Nuclear Reactor Monitoring and Control System> In Fig. 1, the reactor monitoring and control system includes equipment located inside the containment vessel 10, equipment located outside the containment vessel 30, and an electrical penetration 20 that serves as a passageway connecting the two. Inside the containment vessel 10, a nuclear reactor 11 and various detector assemblies 12, 13 attached to the outer periphery of the reactor 11 are arranged, and a current output in response to neutron detection is transmitted as a pulse signal to the outside of the containment vessel 30 via the electrical penetration 20. In Fig. 1, the configuration inside the containment vessel 10 is shown by a top view and a cross-sectional view taken along line A-A in the top view, and four power range detector assemblies 12 and two source range / intermediate range detector assemblies 13 are arranged around the reactor 11. The source range / intermediate range detector assembly 13 has an intermediate range neutron detector 13a and a source range neutron detector 13b as detectors 1 to detect neutrons from the source range to the intermediate range. The neutron flux is detected over a wide range from the source region to the power region by providing these various detectors, including the power range detector of the power range detector assembly 12. Note that the detector 13b reacts to neutrons and outputs a pulse signal, and a proportional counter, an ionization chamber, or the like is used.

[0014] A nuclear instrumentation device 100 disposed outside the containment vessel 30 receives the pulse signal detected by the detector 1 via the electrical penetration 20. The nuclear instrumentation device 100 counts neutrons, converts the count rate, and transmits the result to a host device. The host device is, for example, a monitoring and control device 110 that monitors and controls the nuclear reactor.

[0015] <Configuration and Operation of Nuclear Instrumentation System 100> The configuration and operation of the nuclear instrumentation system 100 will be described in detail below with reference to Figure 2. The nuclear instrumentation system 100 includes a preamplifier 2 that receives and amplifies the pulse signal detected by the detector 1, an A / D converter 3 that converts the amplified analog signal into a digital signal, and a digital processor 4 that processes the digital signal to calculate a count rate. While the preamplifier 2 is often installed near the electrical penetration 20 outside the containment vessel 30, independent of the A / D converter 3 and subsequent components, it may also be installed near the digital processor 4 and subsequent components of the A / D converter 3. Conversely, the digital processor 4 may also be installed together with the preamplifier 2 near the electrical penetration 20. Because the pulse width of the pulse signal may be very short, for example, approximately 0.2 μsec, it is desirable to use an A / D converter 3 that is capable of high-speed operation, for example, one that operates at 100 Msps.

[0016] The digital calculation unit 4 includes a waveform discrimination circuit 41, a pulse count circuit 42, a counting rate conversion circuit 43, a memory unit 44, a zero count circuit 45, a time constant selection circuit 46, and an integration circuit 47, which are all configured as digital circuits.

[0017] The waveform discrimination circuit 41 extracts only neutron-derived pulses from the digital signal converted by the A / D converter 3 and outputs them to the pulse count circuit 42. Specifically, since the pulse signal contains not only neutron-derived signals but also other signals such as noise, the neutron-derived pulse signals are separated by a comparison operation between the peak value of the pulse signal and a preset threshold (discrete level). Here, the comparison operation may be any of an operation that determines a signal above the discreet level as a neutron-derived pulse, an operation that determines a signal below the discreet level as a neutron-derived pulse, or an operation that sets two thresholds and determines a signal within an arbitrary range as a neutron-derived pulse.

[0018] For example, the peak value of a pulse signal derived from neutrons is greater than the peak value of a pulse signal derived from gamma rays, which becomes noise. Therefore, in order to exclude pulse signals derived from gamma rays, a first threshold is set that is greater than the peak value of a pulse signal derived from gamma rays but smaller than the peak value of a pulse signal derived from neutrons, and a pulse equal to or greater than this threshold is determined to be a pulse derived from neutrons. Furthermore, since sudden electromagnetic noise, for example, is greater than the peak value of a pulse signal derived from neutrons, a second threshold is set that is greater than the peak value of a pulse signal derived from neutrons but smaller than the electromagnetic noise, and a pulse equal to or less than this threshold can be determined to be a pulse derived from neutrons. By combining these first and second thresholds, pulses derived from neutrons can be extracted with high accuracy.

[0019] Furthermore, a third threshold value greater than the maximum value of the pulse height of the pulse signal derived from neutrons and a fourth threshold value less than the minimum value of the pulse height of the pulse signal derived from neutrons may be set so as to include the waveform of the pulse signal derived from neutrons, and a pulse signal having a pulse height within the range from the third threshold value to the fourth threshold value may be determined to be a pulse signal derived from neutrons. In an analog circuit, a first threshold value is usually set to discriminate only pulse signals derived from neutrons and exclude pulse signals derived from gamma rays, but in a digital circuit, any number of threshold values ​​can be easily set, improving the accuracy of extracting pulse signals derived from neutrons.

[0020] The pulse count circuit 42 receives the neutron-derived pulses extracted by the waveform discrimination circuit 41 and counts the number of times that neutron-derived pulses arrive during one calculation cycle, for example, 1 msec. The memory unit 44 stores the pulse count value from the previous calculation cycle. Note that the following description will be given using an example in which the number of times that neutron-derived pulses arrive during one calculation cycle, for example, 1 msec, is counted, but the present invention is not limited to this. It is sufficient to count the number of times that neutron-derived pulses arrive periodically over any predetermined period.

[0021] The count rate conversion circuit 43 converts the pulse count value into a count rate by multiplying the pulse count value by a preset coefficient in accordance with the current pulse count value counted by the pulse count circuit 42 in the current calculation cycle and the pulse count value from one calculation cycle before that read out from the memory unit 44. The operation of the count rate conversion circuit 43 will be described in detail later.

[0022] The zero count circuit 45 increments an internal counter by one unit if the pulse count value of the previous calculation cycle read from the memory unit 44 is zero, and resets the counter if the pulse count value is not zero. This makes it possible to calculate the period during which the pulse count value is zero, i.e., the period during which no neutron-derived pulses arrive.

[0023] The time constant selection circuit 46 selects a time constant when the integration circuit 47 calculates the output using the count rate converted by the count rate conversion circuit 43. The method for selecting the time constant differs depending on whether the pulse count value counted by the pulse count circuit 42 is zero or not. If the pulse count value in the current calculation cycle is not zero, a preset time constant is automatically selected in accordance with the count rate calculated by the count rate conversion circuit 43. Furthermore, if the pulse count is zero, a preset time constant is automatically selected in accordance with the period during which the pulse count is zero calculated by the zero count circuit 45. The operation of the time constant selection circuit 46 will be described in detail later.

[0024] The integration circuit 47 performs the integration calculation of equation (1) using the count rate converted by the count rate conversion circuit 43 and the time constant selected by the time constant selection circuit 46 . Here, Y ｉ : Output in the i-th calculation period Y ｉ－１ : Output X in the i-1th calculation cycle ｉ : Counting rate in the i-th calculation period (input from the counting rate conversion circuit 43) τ ｉ : time constant in the i-th calculation cycle T: calculation cycle

[0025] <Operation of Counting Rate Conversion Circuit 43> Next, the operation of the counting rate conversion circuit 43 will be described in detail. Because neutrons are emitted randomly from the reactor, there are cases where the neutron-derived pulse signal increases in a short period of time, but the reactor output does not increase significantly, i.e., there is fluctuation in the neutrons themselves. For this reason, a certain limit is placed on the fluctuation between the pulse in the current calculation cycle and the pulse in the previous calculation cycle, and the counting rate is calculated using a coefficient k corresponding to this limit. There are two examples of limitations: 1) When the fluctuation is small, as the calculation cycle T progresses, the coefficient k gradually increases until it reaches a constant value (1). 2) When a large fluctuation occurs, the coefficient k is reduced (k0) to suppress the influence of the fluctuation.

[0026] Here, an example of the allowable fluctuation range is the current pulse count value C ｎ is the pulse count value C of one calculation cycle before ｎ－１ The time during which the pulse signal originating from neutrons is counted within this allowable fluctuation range is defined as tF, and the relationship between the coefficient k and the time tF is as shown in FIG. 3. The initial value of the coefficient k, i.e., the pulse count value C １ When , the coefficient is k0.

[0027] Figure 3 shows a coefficient curve that takes into account the fluctuation of the neutrons themselves. In Figure 3, the horizontal axis represents time tF, and the vertical axis represents the coefficient k used to calculate the count rate. ｎ But 0.1C ｎ－１ ≦C ｎ <10C ｎ－１ If the above condition is satisfied, the time tF is calculated in the current calculation cycle, and the corresponding coefficient k is found. For example, if the allowable fluctuation range has continued for three times up to the current calculation cycle, tF = 3T, and the coefficient is k(3T). Therefore, the counting rate is k(3T)*C ｎ For example, the current pulse count value C ｎ = 10C ｎ－１ When this happens, tF = 0 and the coefficient is k0. Therefore, the counting rate is k0. ｎ It is calculated as follows.

[0028] The counting rate conversion circuit 43 has a look-up table of the relationship between the coefficient k and the time tF shown in FIG. 3, and calculates the current pulse count value C ｎ and the pulse count value C of one calculation cycle before ｎ－１ is used to calculate whether it is within the allowable fluctuation range, calculate the time tF, obtain the coefficient k corresponding to the time tF from the look-up table, and calculate the pulse count value C ｎ The pulse count value is converted into a counting rate by multiplying by . Note that while it is known in analog circuits to take into account the fluctuations of the neutrons themselves, the relationship between the time tF and the coefficient k is fixed. Changing the relationship between the time tF and the coefficient k requires replacing components. In this embodiment, a digital circuit is used, so it is possible to easily change the settings depending on the magnitude of the neutron fluctuations and the responsiveness required of the nuclear instrumentation device.

[0029] <Operation of the Time Constant Selection Circuit 46> Next, details of the operation of the time constant selection circuit 46 will be described. As described above, in the time constant selection circuit 46, the method of selecting the time constant differs depending on whether the pulse count value counted in the pulse count circuit 42 is zero or not. (1) When the pulse count value counted in the pulse count circuit 42 is zero, the period t during which the pulse count calculated by the zero count circuit 45 is zero is selected. ｐ０ (2) If the pulse count value in the current calculation cycle is not zero, the time constant corresponding to the count rate calculated by the count rate conversion circuit 43 is automatically selected using the lookup table.

[0030] Next, we will explain the lookup table used when the pulse count value in the current calculation cycle is not zero. Figure 4 is a diagram showing the relationship between the count rate and the time constant τ. In the region where the count rate is small, the time constant τ is increased to improve reliability, and in the region where the count rate is large, the time constant τ is decreased to improve responsiveness. The relationship between the count rate and the time constant τ in Figure 4 is created as a lookup table.

[0031] Next, the lookup table to be used when the pulse count value in the current calculation cycle is zero will be described. ｐ０ When the pulse count is zero, the time constant is equal to the preset time constant τ ｐ０ is automatically selected. If the period of zero pulse count continues, the time constant increases according to FIG. 5 and eventually converges to a constant value. ｐ０ may be provided in multiple numbers according to the count rate in the cycle immediately before the pulse count reaches zero.

[0032] In this way, by providing a lookup table that increases the time constant τ in the low count rate range and decreases the time constant τ in the high count rate range, as shown in Figure 4, the digital circuit can automatically select the time constant τ according to the count rate. Furthermore, similar to analog circuits, a nuclear instrumentation device can be realized that has improved reliability in the low count rate range and improved responsiveness in the high count rate range.

[0033] As described above, according to the first embodiment, there is provided a nuclear instrumentation device that calculates a neutron counting rate from a pulse signal detected by a detector, the nuclear instrumentation device comprising: a waveform discrimination circuit that converts a pulse signal into a digital signal and inputs it, and extracts a pulse signal derived from a neutron; a pulse counting circuit that periodically counts the number of times the extracted pulse signal derived from a neutron arrives within a preset time; a memory unit that stores a pulse count value of a period immediately preceding the pulse count value counted by the pulse counting circuit; and a calculation unit that calculates fluctuations in the pulse signal using the pulse count value of a current period counted by the pulse counting circuit and the pulse count value of a period immediately preceding the current period read from the memory unit, and converts the pulse count value of the current period into a counting rate using a coefficient corresponding to the fluctuations in the pulse signal. The nuclear instrumentation device includes a count rate conversion circuit that converts a count rate into a count rate of neutrons, a zero count circuit that increments the counter by one if the pulse count value counted by the pulse count circuit is zero and resets the counter if the pulse count value is not zero and calculates the period during which the pulse count value is zero, a time constant selection circuit that, if the pulse count value counted by the pulse count circuit is zero, selects a time constant in accordance with the period during which the pulse count value calculated by the zero count circuit is zero, and, if the pulse count value counted by the pulse count circuit is not zero, selects a larger time constant for a smaller count rate converted by the count rate conversion circuit and a smaller time constant for a larger count rate, and an integrating circuit that calculates the neutron count rate for each period using the time constant selected by the time constant selection circuit. The components from the waveform discrimination circuit to the integrating circuit are configured as digital circuits. This allows the digital circuit to automatically select a time constant in accordance with the count rate, thereby realizing a nuclear instrumentation device with improved reliability in the low count rate range and improved responsiveness in the high count rate range.

[0034] Furthermore, since the waveform discrimination circuit is configured as a digital circuit, it is possible to easily set multiple thresholds in the waveform discrimination circuit, thereby enabling the neutron-derived pulse signal to be extracted with high accuracy. In this case, in addition to the first threshold set to be greater than the peak value of the gamma-ray-derived pulse signal and smaller than the peak value of the neutron-derived pulse signal, by appropriately setting a threshold for removing electromagnetic noise or a threshold according to the waveform of the neutron-derived pulse signal, it is possible to extract the neutron-derived pulse signal with high accuracy.

[0035] Furthermore, in the counting rate conversion circuit, if the pulse count value in the current cycle counted by the pulse counting circuit is within a preset fluctuation range relative to the pulse count value in the previous cycle read out from the memory unit, the coefficient is increased according to the duration of this fluctuation, and if the pulse count value in the current cycle counted by the pulse counting circuit exceeds the preset fluctuation range relative to the pulse count value in the previous cycle read out from the memory unit, the coefficient is set to a preset initial value.This makes it possible to suppress the effects of sudden increases in the count number that are not due to actual neutron fluctuations, and ensures the reliability of neutron instrumentation at low counting rates.

[0036] Second Embodiment A nuclear instrumentation device according to a second embodiment will be described below with reference to the drawings. In the second embodiment, pulse signal waveforms are discriminated by machine learning in the waveform discrimination circuit 41 of the first embodiment. The configuration of the nuclear instrumentation device is the same as that shown in FIG. 2 of the first embodiment, and therefore a description thereof will be omitted.

[0037] First, the waveform of a pulse signal to be subjected to waveform discrimination will be described. FIG. 6A shows an example of a pulse signal waveform digitally converted by the A / D converter 3, and FIG. 6B shows the relationship between the maximum pulse height Ph and the pulse width Pd of the pulse signal. In FIG. 6A , the pulse signal waveform includes a neutron-derived pulse signal waveform N, a gamma-ray-derived pulse signal waveform γ, a maximum pulse signal waveform EM1 derived from electromagnetic noise, and a periodic pulse signal waveform EM2 derived from electromagnetic noise. As described in the first embodiment, by setting an appropriate threshold, the neutron-derived pulse signal waveform N and the gamma-ray-derived pulse signal waveform γ can be distinguished from each other, and the neutron-derived pulse signal waveform N and the maximum pulse signal waveform EM1 derived from electromagnetic noise can be distinguished from each other. However, the electromagnetic noise-derived periodic pulse signal waveform EM2 cannot be distinguished from each other by setting a threshold. Here, as shown in the figure, the maximum pulse height Ph and the pulse width Pd are calculated using the neutron-derived pulse signal waveform N as an example. Note that the maximum pulse height Ph and the pulse width Pd may also be calculated for other pulse signal waveforms and used for pulse waveform discrimination.

[0038] 6B, each pulse signal waveform is surrounded by a dashed line, and it can be seen that the maximum wave height Ph and pulse width Pd of the pulse signal are concentrated in a certain range depending on the origin of the pulse signal waveform. The waveform discrimination circuit 41 discriminates pulse signal waveforms using this relationship.

[0039] <Machine Learning in Waveform Discrimination Circuit 41> FIG. 7A is a diagram showing the configuration of a learning device provided in the waveform discrimination circuit 41, and FIG. 7B is a diagram showing the configuration of an inference device provided in the waveform discrimination circuit 41.

[0040] First, the learning device 411 will be described. In FIG. 7A , when a series of pulse signal waveforms such as those shown in FIG. 6A are input from the A / D converter 3, the waveform discrimination circuit 41 calculates the maximum pulse height Ph and pulse width Pd of the pulse signal for each pulse signal waveform. The calculated maximum pulse height Ph and pulse width Pd of the pulse signal are input to the data acquisition unit 412 of the learning device 411. At this time, the origin of the pulse signal waveform and its type are also input. The data acquisition unit 412 transmits the acquired data as learning data to the model generation unit 413. The model generation unit 413 learns based on the two inputs, the maximum pulse height Ph and pulse width Pd of the pulse signal and the origin of the pulse signal waveform, and generates a learned model. Based on the acquired maximum pulse height Ph, pulse width Pd, and signal type of the pulse signal, the model generation unit 413 updates the learned model and stores it in the learned model storage unit 415.

[0041] The model generation unit 413 can use, as a learning algorithm, a known algorithm using a neural network, etc. The generated trained model is a combination of the maximum value Ph of the pulse signal wave height and the pulse width Pd, and the type of pulse signal waveform from which the pulse signal waveform originates, as shown in FIG.

[0042] Next, the inference device 416 will be described. In Fig. 7B, when a series of pulse signal waveforms as shown in Fig. 6A is input from the A / D converter 3, the waveform discrimination circuit 41 calculates the maximum value Ph of the pulse signal wave height and the pulse width Pd for each pulse signal waveform.

[0043] The calculated maximum value Ph of the pulse height and pulse width Pd of the pulse signal are input to a data acquisition unit 417 of the inference device 416. The data acquisition unit 417 transmits the acquired data to the inference unit 418 as discrimination data of the pulse signal waveform. The inference unit 418 infers the origin of the pulse signal waveform from the maximum value Ph of the pulse height and pulse width Pd of the input pulse signal based on the learned model in the learned model storage unit 415, and outputs the result. If it is inferred that the signal waveform is derived from neutrons, it outputs 1, and otherwise outputs 0.

[0044] If it is only necessary to discriminate signal waveforms originating from neutrons, the learning device 411 may be configured as a trained model that has learned only signal waveforms originating from neutrons, and the inference device 416 may output whether the maximum value Ph of the wave height and the pulse width Pd of the input pulse signal are signal waveforms originating from neutrons. In this case as well, if it is inferred that the pulse is originating from neutrons, it may output 1, and if it is determined that the pulse is not originating from neutrons, it may output 0, and the pulse signal waveform for which 1 is output may be extracted as originating from neutrons and transmitted to the next pulse count circuit 42.

[0045] The configuration from the pulse count circuit 42 onwards is the same as in embodiment 1, and the pulse count circuit 42 increments its internal counter by one unit when the output from the waveform discrimination circuit 41 is 1, and counts the number of 1s output from the waveform discrimination circuit 41 during one calculation period.

[0046] As described above, according to the second embodiment, the waveform discrimination circuit 41 discriminates pulse signal waveforms by machine learning, eliminating the need to set a threshold (discrimination level), and thus providing even greater convenience than the first embodiment. Furthermore, while conventional waveform discrimination uses peak values, the present embodiment uses the maximum peak value and pulse width for learning and generates a trained model, thereby improving the accuracy of waveform discrimination. Furthermore, by adding data other than neutron-derived pulse signals, such as electromagnetic noise data, to the training data, it becomes possible to discriminate signals that could not be discriminated by conventional waveform discrimination circuits or the first embodiment.

[0047] Third Embodiment A nuclear instrumentation device according to a third embodiment will now be described with reference to the accompanying drawings. In the third embodiment, the counting rate conversion circuit 43 and the time constant selection circuit 46 of the first embodiment select coefficients and time constants by machine learning. The configuration of the nuclear instrumentation device is the same as that shown in FIG. 2 of the first embodiment, and therefore a description thereof will be omitted.

[0048] <Machine Learning in Counting Rate Conversion Circuit 43> FIG. 8A is a diagram showing the configuration of a learning device provided in the counting rate conversion circuit 43, and FIG. 8B is a diagram showing the configuration of an inference device provided in the counting rate conversion circuit 43.

[0049] First, the learning device 431 will be described. In FIG. 8A, the pulse count value C ｎ However, the pulse count value C of the previous calculation cycle is stored in the memory unit 44. ｎ－１ When input, the time tF is input to the data acquisition unit 432 of the learning device 431. The coefficient k at that time is also input. The data acquisition unit 432 transmits the acquired data to the model generation unit 433 as learning data.

[0050] As explained in the first embodiment, the coefficient k is determined so that when a large fluctuation occurs, the coefficient k is made small (k0) to suppress the influence of the fluctuation. Specifically, the same allowable fluctuation range as in the first embodiment is set. That is, the current pulse count value C ｎ is the pulse count value C of one calculation cycle before ｎ－１ The pulse count value C this time is in the range of 1 / 10 or more and less than 10 times. ｎ But 0.1C ｎ－１ ≦C ｎ <10C ｎ－１ If the above expression is satisfied, the time tF is calculated for the current calculation cycle, and the corresponding coefficient k is determined. As in the first embodiment, the time tF is the time during which the neutron-derived pulse signal is counted within the allowable fluctuation range; for example, if this is satisfied three times in the calculation cycle, tF = 3T. If the allowable fluctuation range is satisfied and the fluctuation is small, the coefficient k gradually increases as the calculation cycle T progresses, and the coefficient k is determined so that it reaches a constant value (1). The coefficient k is also teacher data, and the same data as in the first embodiment can be used.

[0051] The model generation unit 433 learns the relationship between the time tF and the coefficient k and generates a trained model. The trained model is updated by repeating the learning process and stored in the trained model storage unit 435. Note that the model generation unit 433 can use a known algorithm using a neural network or the like as the learning algorithm. The trained model can calculate the relationship between the time tF and the coefficient k as a regression curve, allowing the coefficient k to be selected more finely than the discrete value applied in the first embodiment, thereby enabling optimal calculation.

[0052] Next, the inference device 436 will be described. In Fig. 8B, when the time tF is input to the counting rate conversion circuit 43, it is input to the data acquisition unit 437 of the inference device 436. The data acquisition unit 437 transmits the acquired data, that is, the time tF, to the inference unit 438.

[0053] The inference unit 438 outputs a coefficient k corresponding to the time tF based on the learned model in the learned model storage unit 435.

[0054] <Machine Learning in Time Constant Selection Circuit 46> FIG. 9A is a diagram showing the configuration of a learning device provided in the time constant selection circuit 46, and FIG. 9B is a diagram showing the configuration of an inference device provided in the time constant selection circuit 46.

[0055] First, the learning device 461 will be described. In FIG. 9A, the count rate output from the count rate conversion circuit 43 and the pulse count value output from the zero count circuit 45 are 0 during a period t ｐ０ When the time constants τ corresponding to the respective parameters are input, they are input to a data acquisition unit 462 of the learning device 461. The data acquisition unit 462 transmits the acquired data as learning data to a model generation unit 463. The time constant τ is teacher data, and the one used in the first embodiment can be used.

[0056] The model generation unit 463 calculates the counting rate, the time constant τ, and the period t ｐ０ and the time constant τ to generate a trained model. The trained model is updated by repeated learning and stored in the trained model storage unit 465. Note that the model generation unit 463 can use a known algorithm using a neural network or the like as a learning algorithm. The trained model is generated by learning the relationship between the counting rate, the time constant τ, and the period t as a regression curve. ｐ０ and the time constant τ can be calculated, the time constant τ can be selected more precisely than the discrete values ​​used in the first embodiment, and therefore optimal calculation is possible.

[0057] Next, the inference device 466 will be described. In FIG. 9B, the count rate output from the count rate conversion circuit 43 or the pulse count value output from the zero count circuit 45 becomes 0 during a period t ｐ０When input, it is input to the data acquisition unit 467 of the inference device 466. The data acquisition unit 467 acquires the count rate output from the count rate conversion circuit 43 or the pulse count value output from the zero count circuit 45 during the period t ｐ０ The inference unit 468 determines, based on the learned model in the learned model storage unit 465, the period t during which the count rate output from the count rate conversion circuit 43 or the pulse count value output from the zero count circuit 45 becomes 0. ｐ０ From the time constant τ or period t corresponding to the current counting rate ｐ０ The time constant τ corresponding to the above is calculated and output to the integrating circuit 47.

[0058] As described above, according to the third embodiment, the counting rate conversion circuit 43 converts the pulse count value into a counting rate through machine learning, and the time constant selection circuit 46 sets a time constant according to the counting rate through machine learning. This eliminates the need for a lookup table as in the first embodiment, and also makes it possible to interpolate between values ​​in the lookup table, enabling more precise calculations to be performed, thereby achieving more appropriate response time and measurement accuracy.

[0059] In the third embodiment, an example has been described in which machine learning is applied to both the counting rate conversion circuit 43 and the time constant selection circuit 46, but it is possible to achieve the respective effects by applying only one of them. It goes without saying that the second and third embodiments can be combined.

[0060] In the first to third embodiments, the allowable fluctuation range of the pulse count value in the counting rate conversion circuit 43 is set to the current pulse count value C ｎ is the pulse count value C of one calculation cycle before ｎ－１ In contrast, the range is 1 / 10 or more and less than 10 times, i.e., 0.1C ｎ－１ ≦C ｎ <10C ｎ－１ However, the present invention is not limited to this example.

[0061] The nuclear instrumentation system 100 according to the first to third embodiments includes a processor 200 and a storage device 300, as shown in FIG. 10 . Although not shown, the storage device may include a volatile storage device such as a random access memory (RAM) and a nonvolatile auxiliary storage device such as a flash memory. A hard disk auxiliary storage device may be used instead of the flash memory. The processor 200 may be a processor such as a central processing unit (CPU), a digital signal processor (DSP), or a field programmable gate array (FPGA). FPGAs are programmable and offer high computing performance, making them suitable for this embodiment. The processor 200 executes a program input from the storage device 300. In this case, the program is input from the auxiliary storage device to the processor 200 via the volatile storage device. Furthermore, the processor 200 may output data such as calculation results to a volatile storage device of the storage device 300, or may store the data in an auxiliary storage device via the volatile storage device.

[0062] Furthermore, the nuclear instrumentation system 100 according to the first to third embodiments may be configured as shown in Fig. 11, which includes a communication device 400 in addition to the configuration shown in Fig. 10. For example, by transmitting the output of the nuclear instrumentation system 100 to a supervisory control device 110, which is a higher-level device, by wire or wirelessly, it becomes possible to remotely monitor the nuclear reactor.

[0063] Although various exemplary embodiments and examples are described in this disclosure, the various features, aspects, and functions described in one or more embodiments are not limited to the application of a particular embodiment, but may be applied to the embodiments alone or in various combinations. Therefore, countless variations not illustrated are anticipated within the scope of the technology disclosed in this specification. For example, this includes cases where at least one component is modified, added, or omitted, or where at least one component is extracted and combined with components of another embodiment.

[0064] 1: Detector, 2: Preamplifier, 3: A / D Converter, 4: Digital Processing Unit, 10: Inside the Reactor Containment Vessel, 11: Reactor, 12: Power Range Detector Assembly, 13: Source Range / Intermediate Range Detector Assembly, 13a: Intermediate Range Neutron Detector, 13b: Source Range Neutron Detector, 20: Electrical Penetration, 30: Outside the Reactor Containment Vessel, 41: Waveform Discrimination Circuit, 42: Pulse Counting Circuit, 43: Count Rate Conversion Circuit, 44: Memory Unit, 45: Zero Count Circuit, 46: Time Constant Selection Circuit, 47: Integration Circuit, 100: Nuclear Instrumentation Device, 110: Monitoring and Control Device, 200: Processor, 300: Storage Device, 400: Communication Device, 411, 431, 461: Learning Device, 412, 417, 432, 437, 462, 467: Data acquisition unit, 413, 433, 463: Model generation unit, 415, 435, 465: Learned model storage unit, 416, 436, 466: Inference device, 418, 438, 468: Inference unit.

Claims

1. A nuclear instrumentation device which calculates a neutron counting rate from a pulse signal detected by a detector, comprising: a waveform discrimination circuit which converts the pulse signal into a digital signal and inputs it, and extracts a pulse signal derived from a neutron; a pulse counting circuit which periodically counts the number of times that the pulse signal derived from a neutron extracted by the waveform discrimination circuit arrives within a preset time; a memory unit which stores a pulse count value of a cycle immediately preceding the pulse count value counted by the pulse counting circuit; a counting rate conversion circuit which calculates fluctuations in the pulse signal using the pulse count value of a current cycle counted by the pulse counting circuit and the pulse count value of a cycle immediately preceding the current cycle read from the memory unit, and converts the pulse count value of a current cycle into a counting rate using a coefficient according to the fluctuations of the pulse signal; a zero counting circuit which increments a counter by one if the pulse count value counted by the pulse counting circuit is zero, and resets the counter if it is not zero, and calculates a period during which the pulse count value is zero; A nuclear instrumentation device comprising: a time constant selection circuit which, when the pulse count value counted by the pulse count circuit is zero, selects a time constant in accordance with the period during which the pulse count value calculated by the zero count circuit is zero, and, when the pulse count value counted by the pulse count circuit is not zero, selects a larger time constant as the count rate converted by the count rate conversion circuit is smaller, and selects a smaller time constant as the count rate is larger; and an integration circuit which calculates the neutron count rate in each period using the time constant selected by the time constant selection circuit, and wherein these are configured as digital circuits.

2. The nuclear instrumentation device of claim 1, wherein said time constant selection circuit automatically selects a time constant depending on whether the pulse count value counted by said pulse count circuit is zero or not, and when the pulse count value counted by said pulse count circuit is not zero, a larger time constant is selected as the count rate is smaller, and a smaller time constant is selected as the count rate is higher, and when the pulse count value counted by said pulse count circuit is zero, a larger time constant is selected as the period during which the pulse count value calculated by said zero count circuit is zero is longer.

3. A nuclear instrumentation device as claimed in claim 1 or 2, wherein said count rate conversion circuit, when the pulse count value in the current period counted by said pulse counting circuit is within a preset range of variation for the pulse count value in the immediately preceding period read out from said memory unit, increases said coefficient in accordance with the duration of this, and, when the pulse count value in the current period counted by said pulse counting circuit exceeds the preset range of variation for the pulse count value in the immediately preceding period read out from said memory unit, converts the pulse count value in the current period into a count rate with said coefficient as an initial value.

4. A nuclear instrumentation device as claimed in any one of claims 1 to 3, wherein the waveform discrimination circuit extracts a pulse signal derived from a neutron by setting at least a plurality of threshold values ​​including a threshold value set to be greater than the peak value of a pulse signal derived from a gamma ray and less than the peak value of a pulse signal derived from a neutron.

5. The nuclear instrumentation device according to any one of claims 1 to 3, wherein the waveform discrimination circuit comprises: a learning device having a learning data acquisition unit that acquires learning data including the maximum value of wave height and pulse width of a pulse signal and information specifying the origin of the pulse signal and its type, and a model generation unit that generates a trained model using the learning data for inferring the type of the pulse signal; and an inference device having a data acquisition unit that acquires the maximum value of wave height and pulse width of a pulse signal waveform contained in a pulse signal detected by the detector and converted into a digital signal, and an inference unit that uses the trained model to infer the origin of the pulse signal and its type based on the data acquired by the data acquisition unit, and extracts a pulse signal derived from a neutron from the pulse signal detected by the detector and converted into a digital signal.

6. The count rate conversion circuit comprises: a learning device having a learning data acquisition unit that acquires learning data including a duration of time during which the pulse count value in the current cycle counted by the pulse count circuit falls within a preset variation range relative to the pulse count value in the previous cycle read out from the memory unit, and a model generation unit that generates a trained model for inferring the coefficient using the learning data; and an inference device having a data acquisition unit that acquires the time, and an inference unit that infers the coefficient corresponding to the time acquired using the trained model, and converts the pulse count value in the current cycle to a count rate using the coefficient inferred by the inference device.

7. A nuclear instrumentation device as claimed in any one of claims 1 to 6, wherein the time constant selection circuit comprises: a learning device having a learning data acquisition unit that acquires learning data including the count rate converted by the count rate conversion circuit and a time constant corresponding to the count rate, a period during which the pulse count is zero calculated by the zero count circuit and a time constant corresponding to the period, and a model generation unit that generates a trained model for inferring the time constant using the learning data; and an inference device having a data acquisition unit that acquires the count rate converted by the count rate conversion circuit or the period during which the pulse count is zero calculated by the zero count circuit, and an inference unit that uses the trained model to infer the time constant corresponding to the count rate converted by the count rate conversion circuit or the time constant corresponding to the period during which the pulse count is zero calculated by the zero count circuit, and selects the time constant inferred by the inference device.

Images