Signal processing method, learning model generation method, signal processing device, radiation detection device, and computer program

Abstract

Provided are a signal processing method, a learning model generation method, a signal processing device, a radiation detection device, and a computer program which are capable of improving the accuracy of elemental analysis.　In a signal processing method for counting, by wave height, a staircase wave corresponding to radiation detection or a pulse wave obtained by converting the staircase wave, when at least one among a string of signal values constituting a signal including the staircase wave and a string of signal values constituting a signal including the pulse wave are input, at least one among the string of signal values constituting a signal including the staircase wave and the string of signal values constituting a signal including the pulse wave is input to a learning model that outputs information regarding whether a plurality of the pulse waves overlap or not, and the staircase wave or the pulse wave when the plurality of pulse waves do not overlap is counted in accordance with the information output by the learning model.

Description

[Technical field]

[0001] The present invention relates to a signal processing method for processing a signal generated by detection of radiation, a learning model generating method, a signal processing device, a radiation detection device, and a computer program. [Background technology]

[0002] A radiation detection device that detects radiation such as X-rays includes a radiation detector and a signal processing device that processes a signal output by the radiation detector. The radiation detector is configured using a semiconductor radiation detection element or the like, and outputs a step wave every time radiation is detected. The signal processing device converts the step wave into a pulse wave and measures the wave height of the pulse wave. The wave height of the pulse wave corresponds to the energy of the radiation.

[0003] When the interval at which radiation is detected is short, a pile-up may occur in which multiple pulse waves overlap and the height of the pulse waves changes. At this time, the energy of the radiation is measured incorrectly, and a peak having an incorrect energy, a so-called sum peak, occurs in the spectrum of the radiation. Therefore, the signal processing device detects the overlap of the pulse waves and suppresses the occurrence of the sum peak by not measuring the height of the overlapped pulse waves. Conventionally, there are several methods for detecting the overlap of the pulse waves. For example, a pulse wave is detected using a threshold value of the wave height, and when the detection interval of the multiple pulse waves is short, it is determined that the overlap of the pulse waves has occurred. In addition, there is a method for determining whether the overlap of the pulse waves has occurred based on the feature amount of the signal waveform, such as the length of the base of the pulse wave or the rise speed of the pulse wave. Patent Document 1 discloses a technology for determining that the overlap of the pulse waves has occurred when the differential waveform of the pulse wave changes from a negative value to a positive value. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2009-229127 A Summary of the Invention [Problem to be solved by the invention]

[0005] In any of the conventional methods, the overlap of the pulse waves may not be detected. In the method based on the detection interval of the pulse waves, when the detection interval is very short, the signal processing device may not be able to detect the overlap of the multiple pulse waves. The method based on the feature amount of the signal waveform, such as the length of the base, focuses only on the feature amount, and therefore has a problem of being vulnerable to noise. In addition, when the threshold value for detecting the pulse waves is lowered, the overlap of multiple pulse waves with high wave heights cannot be detected, and when the threshold value is increased, the pulse waves with low wave heights cannot be detected. When performing elemental analysis based on the spectrum of radiation, if the overlap of multiple pulse waves cannot be detected, a sum peak occurs, and there is a risk of erroneously detecting a non-existent element. If a pulse wave with a low wave height cannot be detected, it becomes difficult to detect elements with low radiation energy, such as light elements.

[0006] The present invention has been made in consideration of the above circumstances, and an object of the present invention is to provide a signal processing method, a learning model generation method, a signal processing device, a radiation detection device, and a computer program that can improve the accuracy of elemental analysis based on radiation spectra. [Means for solving the problem]

[0007] The signal processing method of the present invention is a signal processing method for counting a step wave or a pulse wave converted from the step wave in response to the detection of radiation, by wave height, characterized in that at least one of a sequence of signal values ​​constituting a signal including the step wave and a sequence of signal values ​​constituting a signal including the pulse wave is input to a learning model that outputs information regarding whether or not multiple pulse waves overlap when at least one of a sequence of signal values ​​constituting a signal including the step wave and a sequence of signal values ​​constituting a signal including the pulse wave is input, and the step wave or the pulse wave when there is no overlap of multiple pulse waves is counted according to the information output by the learning model.

[0008] In one aspect of the present invention, overlapping of pulse waves is detected using a learning model. The learning model outputs information regarding the presence or absence of overlapping of multiple pulse waves when a sequence of signal values ​​constituting a signal including a pulse wave and / or a sequence of signal values ​​constituting a signal including a step wave is input. By using the learning model, overlapping of multiple pulse waves can be effectively detected.

[0009] The signal processing method according to the present invention is characterized in that, depending on the information output by the learning model, the step wave or the pulse wave is not counted when a plurality of the pulse waves overlap.

[0010] In one aspect of the present invention, when a plurality of pulse waves overlap, counting according to detection of radiation is not performed, thereby making it possible to prevent erroneous measurement of radiation energy due to overlapping of pulse waves.

[0011] The signal processing method of the present invention is characterized in that the learning model is a learning model that outputs the information when a sequence of signal values ​​constituting a signal including a wave obtained by shaping the pulse wave is further input, and the sequence of signal values ​​constituting the signal is further input to the learning model.

[0012] In one embodiment of the present invention, in addition to a signal including a pulse wave and / or a signal including a step wave, a sequence of signal values ​​constituting a signal including a wave obtained by shaping a pulse wave is input to a learning model. The shape of a wave obtained by shaping a pulse wave using a filter such as a differential filter differs depending on whether or not multiple pulse waves are present. Therefore, by using a wave obtained by shaping a pulse wave, it is possible to more effectively detect the overlap of multiple pulse waves.

[0013] The signal processing method of the present invention is characterized in that the learning model is a learning model that outputs the information when a feature of at least one of the step wave and the pulse wave is further input, and the feature is input to the learning model.

[0014] In one embodiment of the present invention, the signal waveform feature amount, such as the wave height or the time width, is further input to the learning model. Since the signal waveform feature amount differs depending on the presence or absence of multiple pulse waves, the overlap of multiple pulse waves can be detected more effectively by using the signal waveform feature amount.

[0015] The learning model generating method according to the present invention is characterized in that it generates a plurality of pulse wave signals each consisting of a single pulse wave, and at least one of the wave height, rise and ramp of the single pulse wave being random, generates a plurality of superimposed wave signals each consisting of a superimposed wave in which a plurality of pulse waves are superimposed, and at least one of the intervals, wave height, rise and ramp of the plurality of pulse waves being random, and generates a learning model that outputs information regarding the presence or absence of overlap of a plurality of pulse waves when a string of signal values ​​constituting a signal including any pulse wave is input, using the plurality of pulse wave signals and the plurality of superimposed wave signals as teacher data.

[0016] In one aspect of the present invention, a plurality of pulse wave signals each consisting of a single pulse wave and a plurality of superimposed wave signals each consisting of a superimposed wave in which a plurality of pulse waves are superimposed are generated by simulation so that at least one of the wave height, the rising edge, and the rounding is random. A learning model is trained using the generated plurality of pulse wave signals and the plurality of superimposed wave signals as teacher data. A learning model that outputs information regarding the presence or absence of overlap of pulse waves can be generated. Furthermore, by randomizing the rounding of the pulse waves in the simulation, a signal close to an actual signal can be generated.

[0017] The learning model generating method according to the present invention is characterized in that it generates a plurality of step wave signals, each of which is composed of a single step wave, and at least one of the wave height, rise and ramp of the single step wave being random, generates a plurality of adjacent step wave signals, each of which is composed of a plurality of step waves, and in which a plurality of pulse waves converted from the plurality of step waves overlap, and at least one of the intervals, wave height, rise and ramp of the plurality of step waves is random, and generates a learning model that outputs information regarding the presence or absence of overlap of a plurality of pulse waves converted from the plurality of step waves when a sequence of signal values ​​constituting a signal including an arbitrary step wave is input, using the plurality of step wave signals and the adjacent step wave signals as teaching data.

[0018] In one aspect of the present invention, a plurality of step wave signals each consisting of a single step wave and a plurality of adjacent step wave signals are generated by simulation so that at least one of the wave height, the rising edge, and the rounding edge is random. A learning model is trained using the generated plurality of step wave signals and the plurality of adjacent step wave signals as teacher data. A learning model can be generated that outputs information regarding the presence or absence of overlap of pulse waves converted from the step wave.

[0019] A signal processing method according to the present invention is a signal processing method for converting a step wave corresponding to the detection of radiation into a pulse wave and counting the pulse waves according to their heights, characterized in that a product-sum operation is performed between a sequence of signal values ​​constituting a signal including the pulse wave and a predetermined sequence of coefficients, and the predetermined sequence of coefficients is determined in advance such that a result of the product-sum operation between a sequence of signal values ​​constituting a signal including a plurality of overlapping pulse waves and the predetermined sequence of coefficients is different from a result of the product-sum operation between a sequence of signal values ​​constituting a signal including a single pulse wave, in terms of whether it is greater or smaller than a predetermined value, and whether or not there is overlap between the plurality of pulse waves is determined according to the result of the product-sum operation, and the non-overlapping pulse waves are counted.

[0020] In one aspect of the present invention, a multiply-and-accumulate operation is performed between a sequence of signal values ​​constituting a signal including a pulse wave and a predetermined sequence of coefficients, and overlapping of pulse waves is detected based on the result of the multiply-and-accumulate operation. The sequence of coefficients is determined so that the result of the multiply-and-accumulate operation between the sequence of signal values ​​constituting a signal including a single pulse wave and the sequence of coefficients is different from the result of the multiply-and-accumulate operation between the sequence of signal values ​​constituting a signal including multiple overlapping pulse waves and the sequence of coefficients. This method can also effectively detect overlapping of multiple pulse waves.

[0021] The signal processing device according to the present invention is characterized in that, in the signal processing device that processes a signal including a step wave in response to detection of radiation, it is provided with a learning model that outputs information regarding the presence or absence of overlap of a plurality of the pulse waves when at least one of a sequence of signal values ​​constituting the signal including the step wave and a sequence of signal values ​​constituting a signal including a pulse wave converted from the step wave is input.

[0022] The radiation detection device according to the present invention is characterized in that it comprises a radiation detector that outputs a step wave corresponding to the energy of radiation when detecting radiation; a conversion unit that converts the step wave into a pulse wave; a learning model that outputs information regarding the presence or absence of overlap of a plurality of the pulse waves when at least one of a sequence of signal values ​​constituting a signal including the step wave and a sequence of signal values ​​constituting a signal including a pulse wave converted from the step wave is input; a counting unit that counts the step waves or the pulse waves when there is no overlap of the plurality of the pulse waves according to the information output by the learning model; and a spectrum generation unit that generates a spectrum of radiation according to the wave height and count number of the step wave or the pulse wave.

[0023] The computer program according to the present invention is characterized in that, when at least one of a sequence of signal values ​​constituting a signal including a step wave in response to the detection of radiation and a sequence of signal values ​​constituting a signal including a pulse wave converted from the step wave is input, the computer is caused to execute a process of inputting at least one of a sequence of signal values ​​constituting the signal including the step wave and a sequence of signal values ​​constituting the signal including the pulse wave into a learning model that outputs information regarding whether or not a plurality of the pulse waves overlap, and outputting the information.

[0024] In one aspect of the present invention, a signal processing device detects overlapping of pulse waves by utilizing a learning model. When a sequence of signal values ​​constituting a signal including a pulse wave and / or a sequence of signal values ​​constituting a signal including a step wave is input, the learning model outputs information regarding the presence or absence of overlapping of multiple pulse waves. The radiation detection device counts step waves or pulse waves when there is no overlapping of multiple pulse waves, and generates a spectrum of radiation. The signal processing device can effectively detect overlapping of multiple pulse waves, and the radiation detection device can improve the accuracy of elemental analysis based on the spectrum. Effect of the Invention

[0025] The present invention has excellent effects such as being able to effectively detect overlapping of pulse waves and improving the accuracy of elemental analysis. [Brief description of the drawings]

[0026] [Figure 1] 1 is a block diagram showing a functional configuration of a radiation detection apparatus according to a first embodiment. [Figure 2A] 5A to 5C are schematic characteristic diagrams showing examples of a step wave and a pulse wave. [Figure 2B] 5A to 5C are schematic characteristic diagrams showing examples of a step wave and a pulse wave. [Figure 3A] 10A and 10B are schematic characteristic diagrams showing examples of a step wave and a pulse wave when the interval at which radiation is detected is short. [Figure 3B] 10A and 10B are schematic characteristic diagrams showing examples of a step wave and a pulse wave when the interval at which radiation is detected is short. [Figure 4] 11 is a graph showing a schematic example of a signal input to a processing unit. [Diagram 5] FIG. 2 is a conceptual diagram showing an example of the functional configuration of a learning model according to the first embodiment. [Figure 6] FIG. 2 is a block diagram showing an example of the configuration of a learning device that learns a learning model. [Figure 7] 13 is a flowchart showing the steps of a process for generating a learning model. [Figure 8] 4 is a flowchart showing a procedure of processing executed by the signal processing device according to the first embodiment. [Figure 9] FIG. 1 is a spectral diagram showing radiation spectra obtained by a number of methods. [Figure 10] FIG. 11 is a block diagram showing the functional configuration of a radiation detection apparatus according to a second embodiment. [Figure 11] FIG. 11 is a block diagram showing the functional configuration of a radiation detection apparatus according to a third embodiment. [Figure 12] FIG. 11 is a conceptual diagram showing an example of the functional configuration of a learning model according to the third embodiment. [Figure 13] 11 is a flowchart showing a procedure of processing executed by a signal processing device according to a third embodiment. [Figure 14] FIG. 11 is a block diagram showing the functional configuration of a radiation detection apparatus according to a fourth embodiment. [Figure 15] FIG. 13 is a conceptual diagram showing an example of the functional configuration of a learning model according to the fourth embodiment. [Figure 16] 13 is a flowchart showing a procedure of processing executed by a signal processing device according to a fourth embodiment. [Figure 17] FIG. 13 is a block diagram showing the functional configuration of a processing unit according to the fifth embodiment. [Figure 18] FIG. 13 is a block diagram showing the functional configuration of a radiation detection apparatus according to a sixth embodiment. [Figure 19] 1 is a graph showing an example of a coefficient sequence. [Figure 20] 1 is a graph showing an example of a coefficient sequence. [Figure 21]13 is a flowchart showing the procedure of processing executed by a signal processing device according to a sixth embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0027] The present invention will now be described in detail with reference to the drawings showing embodiments thereof. <Embodiment 1> FIG. 1 is a block diagram showing a functional configuration of a radiation detection device 10 according to a first embodiment. The radiation detection device 10 includes a radiation detector 1, a signal processing device 2, and an analysis unit 3. The radiation detector 1 includes a radiation detection element 11 and a preamplifier 12. The radiation detection element 11 is a semiconductor radiation detection element such as an SDD (Silicon Drift Detector), which generates charges according to the energy of incident radiation and outputs a current signal according to the generated charges. The preamplifier 12 converts the current signal output by the radiation detection element 11 into a voltage signal, and generates a step wave whose signal value rises in a step-like manner when radiation is detected. The radiation detector 1 outputs a signal including the step wave generated by the preamplifier 12.

[0028] The signal output by the radiation detector 1 is input to the signal processing device 2. The signal processing device 2 executes a signal processing method. The signal processing device 2 includes an A / D (analog / digital) conversion unit 21. The A / D conversion unit 21 receives a signal including a step wave from the radiation detector 1 and A / D converts the signal including the step wave. A waveform shaping unit 22 is connected to the A / D conversion unit 21. The waveform shaping unit 22 receives a signal including a step wave from the A / D conversion unit 21. The waveform shaping unit 22 passes the signal including the step wave through a predetermined filter to shape the waveform of the signal, thereby converting the signal including the step wave into a signal including a pulse wave. The filter used by the waveform shaping unit 22 is, for example, a differential filter or a trapezoidal shaping filter. Through processing in the waveform shaping unit 22, the step wave is converted into a pulse wave, noise contained in the signal is reduced, and predetermined amplification is performed. The waveform shaping unit 22 outputs a signal including a pulse wave. The waveform shaping unit 22 corresponds to a conversion unit.

[0029] 2A and 2B are schematic characteristic diagrams showing examples of a step wave and a pulse wave. In the diagram, the horizontal axis indicates time, and the vertical axis indicates a signal value. FIG. 2A shows a signal including a step wave output by the radiation detector 1. The radiation detector 1 outputs a step wave in which the signal value rises in a step-like manner each time radiation is detected. In response to one radiation detection, one step wave in which the signal value rises in a step-like manner is generated. When the radiation detector 1 detects radiation multiple times, a signal including multiple step waves is output. Each time radiation is detected, the signal value rises. The height of the step in which the signal value rises is the wave height of the step wave. The wave height of the step wave corresponds to the energy of the radiation. In reality, the step wave is not a perfect step shape, and the signal waveform includes a rise and a rounding. The rise is a distortion of the signal waveform when the signal value rises from a reference value, and the rounding is a distortion of the signal waveform when the step wave ends.

[0030] FIG. 2B shows a signal obtained by converting the signal shown in FIG. 2A by the waveform shaping unit 22. The step wave is converted into a pulse wave. A pulse wave is a signal in which the signal value rises from a predetermined signal reference where the signal value is zero to a peak value, and then falls back to the signal reference. The signal reference is, for example, zero. The height from the signal reference to the peak value is the pulse wave height. The pulse wave height corresponds to the energy of the radiation. The shape of the pulse wave includes a rise and a drop. A rise is a distortion of the signal waveform when the signal value rises from the reference value, and a drop is a distortion of the signal waveform when the pulse wave ends. FIG. 2B shows an example in which the radiation detector 1 detects radiation multiple times at long intervals, and multiple pulse waves are not superimposed.

[0031] 3A and 3B are schematic characteristic diagrams showing examples of a step wave and a pulse wave when the interval at which radiation is detected is short. In the diagram, the horizontal axis indicates time, and the vertical axis indicates the signal value. FIG. 3A shows a signal including a step wave output by the radiation detector 1. Compared to the example shown in FIG. 2A, the interval at which the radiation detector 1 detects radiation multiple times is short, and the interval between the multiple step waves is short. FIG. 3B shows a signal obtained by converting the signal shown in FIG. 3A by the waveform shaping unit 22. The interval between the multiple pulse waves is short, and a superimposed wave in which the multiple pulse waves are overlapped is formed. The wave height of the superimposed wave is different from the wave height of a single pulse wave, and an erroneous radiation energy is measured depending on the wave height of the superimposed wave.

[0032] The processing unit 23 and the pulse detection unit 24 are connected to the waveform shaping unit 22. The waveform shaping unit 22 inputs a signal including a pulse wave to the processing unit 23 and the pulse detection unit 24. The pulse detection unit 24 receives the signal from the waveform shaping unit 22 and detects the pulse wave included in the signal. For example, the pulse detection unit 24 determines that a pulse wave has been detected when the signal value exceeds a predetermined threshold value. The pulse detection unit 24 is connected to the processing unit 23. When the pulse detection unit 24 detects a pulse wave, it inputs information indicating that a pulse wave has been detected to the processing unit 23.

[0033] The processing unit 23 receives a signal including a pulse wave from the waveform shaping unit 22, and receives information indicating that a pulse wave has been detected from the pulse detection unit 24. The processing unit 23 is configured using elements that perform calculations. The processing unit 23 includes a buffer memory 231 and a learning model 232 for determining whether or not there is overlap of multiple pulse waves. For example, the learning model 232 is configured using an FPGA (field-programmable gate array). When a pulse wave is detected, the processing unit 23 uses the learning model 232 to determine whether or not the signal includes overlap of multiple pulse waves. A method for determining whether or not there is overlap of multiple pulse waves will be described later.

[0034] The waveform shaping unit 22 and the processing unit 23 are connected to the pulse-height measuring unit 25. The waveform shaping unit 22 inputs a signal including a pulse wave to the pulse-height measuring unit 25. The processing unit 23 inputs information indicating the presence or absence of overlapping of multiple pulse waves to the pulse-height measuring unit 25. When there is no overlapping of multiple pulse waves, the pulse-height measuring unit 25 measures the height of the pulse wave included in the signal input from the waveform shaping unit 22. When there is overlapping of multiple pulse waves, the pulse-height measuring unit 25 does not measure the height of the pulse wave included in the signal input from the waveform shaping unit 22.

[0035] The counting unit 26 is connected to the pulse-height measuring unit 25. The pulse-height measuring unit 25 inputs the measured pulse-height of the pulse wave to the counting unit 26. The counting unit 26 counts the pulse waves for each wave height. For example, the counting unit 26 is a multi-channel analyzer. The counting unit 26 may be configured to count pulse waves for all wave heights, or may be configured to count pulse waves only for a specific wave height. The signal processing device 2 outputs data indicating the relationship between the pulse-height of the pulse wave and the count number counted by the counting unit 26. The count number corresponds to the number of times that the radiation detector 1 has detected radiation having energy corresponding to the pulse-height of the pulse wave.

[0036] When multiple pulse waves overlap, the pulse-height measuring unit 25 does not measure the height of the pulse waves, and the signal processing device 2 does not count the overlapped pulse waves. Note that the pulse-height measuring unit 25 may be configured to measure the height of the overlapped pulse waves but not input the measured wave height to the counting unit 26. The pulse-height measuring unit 25 may be configured to input the height of the overlapped pulse waves to the counting unit 26, and the counting unit 26 may not count the overlapped pulse waves. The pulse-height measuring unit 25 may be configured to input the height of the overlapped pulse waves to the counting unit 26, and the counting unit 26 may count the overlapped pulse waves separately from the non-overlapping pulse waves.

[0037] The analysis unit 3 is composed of a computer such as a personal computer. The analysis unit 3 receives the data output by the signal processing device 2. The analysis unit 3 performs a process of generating a spectrum of the radiation detected by the radiation detector 1 from the relationship between the pulse height of the pulse wave and the count number. The analysis unit 3 corresponds to a spectrum generation unit. The analysis unit 3 may further perform further processes such as elemental analysis of the radiation source based on the generated spectrum of radiation. For example, the analysis unit 3 irradiates a sample with radiation, detects characteristic X-rays generated from the sample with the radiation detector 1, and performs qualitative or quantitative analysis of elements contained in the sample based on the spectrum of the characteristic X-rays. The signal processing device 2 may also have a function of generating a spectrum of radiation.

[0038] When overlapping pulse waves are erroneously counted as a single pulse wave and an incorrect radiation energy is measured, the radiation spectrum will contain a sum peak, which is a peak having an incorrect energy. When elemental analysis is performed based on a spectrum containing a sum peak, there is a risk of erroneously detecting a non-existent element corresponding to the sum peak. Or, there is a risk of the sum peak overlapping the peak of an existing element, resulting in an excessive detection of the amount of that element. For example, when multiple pulse waves generated by detecting the K line of Fe (iron) overlap, the sum peak overlaps the peak of the L line of Pb (lead), and the amount of Pb is detected excessively. When multiple pulse waves generated by detecting the K line of Cu (copper) overlap, the sum peak overlaps the peak of the K line of Cd (cadmium), and the amount of Cd is detected excessively.

[0039] A method for determining whether or not there is overlap of multiple pulse waves in the processing unit 23 will be described. Fig. 4 is a graph that shows a schematic example of a signal input to the processing unit 23. The horizontal axis in the figure indicates time, and the vertical axis indicates the signal value. The signal is composed of a time series of discrete signal values ​​obtained at a predetermined time interval. In other words, the signal is represented as one-dimensional data consisting of a sequence of signal values.

[0040] 2B and 3B, a signal including a single pulse wave and a signal including a superimposed wave in which multiple pulse waves are superimposed have different signal shapes and also differ in the change in signal value over time. Learning model 232 is pre-trained so that when a sequence of signal values ​​constituting a signal including a pulse wave is input, it outputs information indicating whether or not multiple pulse waves overlap.

[0041] 5 is a conceptual diagram showing an example of a functional configuration of the learning model 232 according to the first embodiment. The learning model 232 uses a fully connected neural network including an input layer, a plurality of intermediate layers, and an output layer, each of which has a plurality of nodes. The input layer has a plurality of nodes 41 to which a sequence of signal values ​​constituting a signal including a pulse wave is input. One signal value in the sequence of signal values ​​is input to one node 41, and each signal value is input to one of the nodes 41. For example, the input layer includes m nodes 41, and m signal values ​​constituting a signal including a pulse wave are input to the input layer.

[0042] The learning model 232 has n (n is a natural number) layers of intermediate layers. The first intermediate layer has a plurality of nodes 421. Each node 41 in the input layer outputs a signal value to the plurality of nodes 421. The plurality of nodes 421 accepts a signal value from the node 41 in the input layer, performs a calculation on the signal value using a parameter, and outputs the data of the calculation result to a plurality of nodes 422 included in the second intermediate layer. The nodes included in each intermediate layer accept data from a plurality of nodes in the previous intermediate layer, perform a calculation on the accepted data using a parameter, and output the data to the node in the subsequent intermediate layer. For example, the node performs a calculation of f(Σ(w*x)+b) with the value of the data accepted from each node in the previous layer as x, the weight corresponding to each node as w, the bias value as b, and the activation function as f(), and outputs the data of the calculation result to a plurality of nodes in the subsequent layer. The activation function is, for example, a relu function or a sigmoid function. The activation function may be any other function generally used in machine learning.

[0043] The output layer of the learning model 232 has a single node 43. A plurality of nodes 42n included in the n-th intermediate layer output data to the node 43 included in the output layer. The node 43 in the output layer receives data from the plurality of nodes 42n, performs a calculation on the received data using parameters, and outputs information indicating whether or not a plurality of pulse waves overlap. For example, the activation function in the node 43 is a function that outputs data indicating whether or not the calculation result of (Σ(w*x)+b) is a positive value. For example, the node 43 may output a value of 1 as data indicating a positive value, and may output a value of zero as data indicating a value equal to or less than zero. For example, data indicating a positive value is information indicating that a plurality of pulse waves do not overlap, and data indicating a value equal to or less than zero is information indicating that a plurality of pulse waves overlap. Note that the node 43 may output the probability that a plurality of pulse waves overlap as information indicating whether or not a plurality of pulse waves overlap. The learning model 232 may use a convolutional neural network (CNN) or a recurrent neural network (RNN) as the neural network.

[0044] Learning of the learning model 232 is performed using a computer. FIG. 6 is a block diagram showing an example of the configuration of a learning device 5 that learns the learning model 232. The learning device 5 executes a learning model generation method. The learning device 5 is a computer such as a server device. The learning device 5 includes a calculation unit 51, a memory 52, a storage unit 53, a display unit 54, and an operation unit 55. The calculation unit 51 is configured using, for example, a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), or a multi-core CPU. The calculation unit 51 may be configured using a quantum computer. The memory 52 stores temporary data generated in association with the calculation. The memory 52 is, for example, a RAM (Random Access Memory). The storage unit 53 is non-volatile, for example, a hard disk. The display unit 54 is, for example, a liquid crystal display or an EL display (Electroluminescent Display). The operation unit 55 accepts input of information such as text by accepting an operation from a user. The operation unit 55 is, for example, a keyboard or a touch panel. The storage unit 53 stores a computer program 531. The calculation unit 51 executes processing in accordance with the computer program 531.

[0045] FIG. 7 is a flowchart showing the procedure of the process of generating the learning model 232. Hereinafter, step is abbreviated as S. The calculation unit 51 executes the following process according to the computer program 531. The calculation unit 51 generates a plurality of pulse wave signals each consisting of a single pulse wave (S11). In S11, the calculation unit 51 generates a plurality of pulse wave signals in which at least one of the pulse wave height, rise, and accent is random by simulation. All of the pulse wave height, rise, and accent may be random. The pulse wave signal is composed of a sequence of signal values. For example, the number of signal values ​​included in the pulse wave signal is the same as the number of nodes 41 in the input layer of the learning model 232.

[0046] The calculation unit 51 generates the pulse wave height, the rising time constant, and the rising start time by uniformly distributed random numbers. As the random numbers, exponentially distributed or Poisson distributed random numbers may be used. The calculation unit 51 also randomly generates the time constant of the rounding of the pulse wave. The calculation unit 51 generates a pulse wave signal by superimposing white noise on the pulse wave according to the generated parameters. The noise superimposed on the pulse wave may be 1 / f noise, or may not be superimposed. It is preferable that the noise superimposed on the pulse wave is noise according to the characteristics of the radiation detector 1. The pulse wave signal is a signal as shown in FIG. 2B. The calculation unit 51 stores pulse wave data 532 including the generated multiple pulse wave signals in the storage unit 53. When the size of the radiation detector 1 is large, the rounding of the actually measured signal varies greatly. By randomizing the rounding of the pulse wave in the simulation, a signal close to the actual signal can be generated.

[0047] The calculation unit 51 then generates a plurality of superimposed wave signals each of which is a superimposed wave in which a plurality of pulse waves are superimposed (S12). In S12, the calculation unit 51 generates a plurality of superimposed wave signals each of which is a superimposed wave of two pulse waves, each of which has a random wave height, a rising time constant, and a rising start time, by simulation. The superimposed wave signal is also composed of a sequence of signal values. The calculation unit 51 generates the wave height, the rising time constant, and the rising start time of the first pulse wave, and the wave height, the rising time constant, and the rising start time of the second pulse wave, by uniformly distributed random numbers. The calculation unit 51 also randomly generates the time constant of the rounding of the pulse wave. The time constant of the rounding may be the same for the two pulse waves. The calculation unit 51 superimposes noise on the superimposed wave according to the generated parameters to generate the superimposed wave signal. The superimposed wave signal is a signal as shown in FIG. 3B. The calculation unit 51 stores in the storage unit 53 the superimposed wave data 533 including the generated multiple superimposed wave signals.

[0048] The calculation unit 51 then performs processing for generating the learning model 232 using the multiple pulse wave signals included in the pulse wave data 532 and the multiple superimposed wave signals included in the superimposed wave data 533 as teacher data (S13). In S13, the calculation unit 51 inputs a sequence of signal values ​​constituting the multiple pulse wave signals and the multiple superimposed wave signals to the input layer of the learning model 232. One signal value is input to each of the nodes 41 of the input layer. The calculation unit 51 associates the pulse wave signal with information indicating that there is no overlap of multiple pulse waves, and associates the superimposed wave signal with information indicating that there is overlap of multiple pulse waves. The learning model 232 outputs information indicating whether or not there is overlap of multiple pulse waves from the node 43 of the output layer. The calculation unit 51 calculates the error of the information using an error function with the information associated with the input pulse wave signal or the superimposed wave signal and the information output from the node 43 as variables, and adjusts the parameters of the calculation of each node of the learning model 232 so that the error is minimized by the error backpropagation method. In other words, the parameters are adjusted so that when a pulse wave signal is input, information close to information indicating that there is no overlap of multiple pulse waves is output, and when a superimposed wave signal is input, information close to information indicating that there is overlap of multiple pulse waves is output.

[0049] The calculation unit 51 repeats the process using a plurality of pulse wave signals and a plurality of superimposed wave signals to adjust the parameters of each node of the learning model 232, thereby performing machine learning of the learning model 232. The calculation unit 51 stores the learned data 534 in which the adjusted final parameters are recorded in the storage unit 53. In this manner, the learned learning model 232 is generated. After S13 is completed, the calculation unit 51 ends the process. The learning model 232 included in the processing unit 23 is manufactured based on the learned data 534. For example, the learning model 232 is manufactured by writing the parameters recorded in the learned data 534 to the FPGA included in the processing unit 23. Note that the calculation unit 51 may perform machine learning of the learning model 232 using the pulse wave signal and the superimposed wave signal that are actually measured, rather than using the pulse wave signal and the superimposed wave signal created by simulation.

[0050] Next, the process executed by the signal processing device 2 will be described. FIG. 8 is a flowchart showing the procedure of the process executed by the signal processing device 2 according to the first embodiment. When radiation is incident on the radiation detection element 11, the radiation detector 1 generates a step wave according to the energy of the radiation, and outputs a signal including the step wave. The signal processing device 2 receives the signal including the step wave from the radiation detector 1 (S21). The A / D conversion unit 21 performs A / D conversion on the input signal (S22). The A / D conversion unit 21 inputs the A / D converted signal to the waveform shaping unit 22. The waveform shaping unit 22 shapes the waveform of the input signal (S23). By shaping the waveform, the waveform shaping unit 22 reduces noise contained in the signal, and converts the step wave contained in the signal into a pulse wave. The waveform shaping unit 22 inputs the signal including the pulse wave to the processing unit 23, the pulse detection unit 24, and the pulse height measurement unit 25.

[0051] The signal input to processing unit 23 is composed of a time series of signal values. Processing unit 23 sequentially stores the signal values ​​in buffer memory 231 (S24). The processes of S21 to S24 are individually and repeatedly executed, and the signal values ​​are sequentially stored in buffer memory 231. Buffer memory 231 is a first-in, first-out memory, and stores a plurality of signal values ​​that are sequentially input. If a new signal value is input when the amount of a plurality of signal values ​​stored in buffer memory 231 has reached an upper limit, buffer memory 231 erases the first signal value stored among the plurality of signal values ​​stored, and stores the new signal value.

[0052] Pulse detection unit 24 waits for detection of a pulse wave contained in the input signal (S25). In S25, for example, pulse detection unit 24 determines that a pulse wave contained in the signal has been detected when the signal value exceeds a predetermined threshold. The threshold is stored in advance in processing unit 23. If a pulse wave is not detected (S25: NO), pulse detection unit 24 repeats the process of S25.

[0053] When a pulse wave contained in the signal is detected (S25: YES), the pulse detection unit 24 inputs information indicating that a pulse wave has been detected to the processing unit 23, and the processing unit 23 inputs a sequence of signal values ​​constituting the signal containing the pulse wave to the learning model 232 (S26). In S26, the processing unit 23 inputs, for example, a plurality of signal values ​​stored in the buffer memory 231 at the time when the information indicating that a pulse wave has been detected is input to the learning model 232. The processing unit 23 may thin out the signal values ​​constituting the signal before inputting them to the learning model 232. As described above, the learning model 232 to which the signal containing the pulse wave has been input performs neural network calculations and outputs information indicating whether or not a plurality of pulse waves overlap.

[0054] The processing unit 23 inputs information indicating the presence or absence of overlapping of the multiple pulse waves output by the learning model 232 to the pulse-height measuring unit 25. The pulse-height measuring unit 25 determines whether or not there is overlapping of the multiple pulse waves based on the input information (S27). If information indicating that there is no overlapping of the multiple pulse waves is input and there is no overlapping of the multiple pulse waves (S27: NO), the pulse-height measuring unit 25 measures the height of the pulse wave included in the input signal (S28). The pulse-height measuring unit 25 inputs the measured pulse wave height to the counting unit 26. The counting unit 26 counts the pulse waves for each pulse height input from the pulse-height measuring unit 25 (S29), and ends the process. If information indicating that there is overlapping of the multiple pulse waves is input and there is overlapping of the multiple pulse waves (S27: YES), the pulse-height measuring unit 25 does not measure the height of the overlapping multiple pulse waves, and the signal processing device 2 ends the process. As a result, the counting section 26 does not count multiple overlapping pulse waves. The signal processing device 2 repeatedly executes the processes of S21 to S29 individually.

[0055] The signal processing device 2 outputs data indicating the relationship between the wave height of the pulse wave and the number counted by the counting unit 26. The analysis unit 3 receives the data output by the signal processing device 2 and generates a spectrum of the radiation detected by the radiation detector 1 based on the data.

[0056] FIG. 9 is a spectrogram showing the spectrum of radiation obtained by a plurality of methods. The horizontal axis in the figure shows the radiation energy in keV units, and the vertical axis shows the radiation count number. The spectrum shown in FIG. 9 includes a peak near 11.8 keV and a peak near 12.4 keV. However, both of these peaks are sum peaks obtained by measuring the wave heights of a plurality of overlapping pulse waves. In FIG. 9, the spectrum when the overlap of pulse waves is not detected is shown by a dashed line, and the spectrum when the overlap of pulse waves is detected by the conventional technology is shown by a thin solid line. Moreover, the spectrum when the overlap of pulse waves is detected by the present embodiment is shown by a thick solid line.

[0057] As shown in FIG. 9, when the overlap of the pulse waves is detected, the intensity of the sum peak is reduced. In the present embodiment, the intensity of the sum peak is further reduced compared to the conventional technology. Therefore, in the present embodiment, the overlap of a plurality of pulse waves, which has been difficult to detect in the past, can be effectively detected. By effectively detecting the overlap of a plurality of pulse waves, it is possible to suppress erroneous measurement of the energy of the radiation due to the overlap of the pulse waves. This makes it possible to suppress the occurrence of a sum peak in the spectrum of the radiation, and improve the accuracy of elemental analysis based on the spectrum. For example, it is possible to suppress erroneous detection of a non-existent element. In addition, even if the threshold for detecting the pulse waves is lowered, it is possible to detect the overlap of a plurality of pulse waves. Therefore, it is possible to lower the threshold for detecting the pulse waves, and it becomes easy to detect elements with low radiation energy, such as light elements.

[0058] As described above in detail, the signal processing device 2 detects the overlap of multiple pulse waves using the learning model 232. When a sequence of signal values ​​constituting a signal including pulse waves is input, the learning model 232 outputs information indicating the presence or absence of overlap of multiple pulse waves. By using the learning model 232, it is possible to detect the overlap of multiple pulse waves, which was previously difficult to detect. Since the learning model 232 determines the presence or absence of overlap of multiple pulse waves from the entire waveform of the signal, it is less susceptible to the influence of noise compared to a conventional determination method based on the feature amount of the signal waveform, and the presence or absence of overlap of multiple pulse waves can be obtained more reliably. This allows the radiation detection device 10 to improve the accuracy of elemental analysis based on the spectrum of radiation. Moreover, the signal processing device 2 can execute processing at a higher speed than when processing an image of a signal by processing a sequence of signal values, and can detect the overlap of pulse waves almost in real time.

[0059] <Embodiment 2> 10 is a block diagram showing a functional configuration of a radiation detection device 10 according to embodiment 2. The configurations and functions of the radiation detector 1 and the analysis unit 3 are similar to those of embodiment 1. The processing unit 23 is connected to the waveform shaping unit 22 and the A / D conversion unit 21. The processing unit 23 receives an input of a signal including a pulse wave from the waveform shaping unit 22 and an input of a signal including a step wave from the A / D conversion unit 21. The processing unit 23 stores, in a buffer memory 231, signal values ​​constituting the signal including a pulse wave and signal values ​​constituting the signal including a step wave.

[0060] The learning model 232 is pre-trained to output information indicating whether or not there is an overlap of multiple pulse waves when a signal including a pulse wave and a signal including a step wave are input. The learning model 232 uses a fully connected neural network, as in the first embodiment. The input layer of the learning model 232 includes a plurality of nodes 41 to which a sequence of signal values ​​constituting a signal including a pulse wave is input, as well as a plurality of nodes 41 to which a sequence of signal values ​​constituting a signal including a step wave is input. The node 43 in the output layer of the learning model 232 outputs information indicating whether or not there is an overlap of multiple pulse waves, as in the first embodiment. The configuration of the signal processing device 2 other than the A / D conversion unit 21 and the processing unit 23 is the same as in the first embodiment.

[0061] The learning model 232 is learned by the learning device 5 in the same manner as in the first embodiment. The calculation unit 51 of the learning device 5 generates a step wave signal corresponding to a signal including a step wave before the pulse wave is converted by the waveform shaping unit 22, in addition to generating a plurality of pulse wave signals in S11. Conversely, the calculation unit 51 may generate a plurality of step wave signals including a step wave, at least one of the wave height, the rise, and the drop of the step wave being random, and superimposed with white noise, by simulation, and generate a pulse wave signal by differentiating the step wave signal. The noise superimposed on the signal may be 1 / f noise, or noise may not be superimposed. Furthermore, the calculation unit 51 generates an adjacent step wave signal corresponding to a signal including a plurality of step waves before the superimposed wave is converted by the waveform shaping unit 22, in addition to generating a plurality of superimposed wave signals in S12. Conversely, the calculation unit 51 may generate a plurality of adjacent step wave signals including two adjacent step waves, at least one of the wave height, the rise, and the drop of the step wave being random, by simulation, and generate a superimposed wave signal by differentiating the adjacent step wave signal. The calculation section 51 may use a step wave signal and an adjacent step wave signal that are actually measured, instead of using the step wave signal and the adjacent step wave signal that are generated by simulation.

[0062] In S13, the calculation unit 51 performs processing for generating the learning model 232 using the multiple pulse wave signals and step wave signals, and the multiple superimposed wave signals and adjacent step wave signals as teacher data. In S13, the calculation unit 51 inputs a sequence of signal values ​​constituting the pulse wave signals and step wave signals to the input layer of the learning model 232. The calculation unit 51 also inputs a sequence of signal values ​​constituting the superimposed wave signals and adjacent step wave signals to the input layer of the learning model 232. The calculation unit 51 adjusts the parameters of the calculation of each node of the learning model 232 by the backpropagation method. The calculation unit 51 repeats processing using the multiple pulse wave signals and the multiple superimposed wave signals to perform machine learning of the learning model 232.

[0063] The signal processing device 2 executes the processes of S21 to S25 as in the first embodiment. In S26, the processing unit 23 inputs to the learning model 232 a sequence of signal values ​​constituting the signal including the pulse wave input from the waveform shaping unit 22 and a sequence of signal values ​​constituting the signal including the step wave input from the A / D conversion unit 21. The learning model 232 performs neural network calculations and outputs information indicating whether or not there is overlap of a plurality of pulse waves. The signal processing device 2 executes the processes of S27 to S29 as in the first embodiment. The signal processing device 2 outputs data indicating the relationship between the pulse wave height and the count number, and the analysis unit 3 receives the data output by the signal processing device 2 and generates a spectrum of the radiation detected by the radiation detector 1. The analysis unit 3 may perform elemental analysis of the radiation source based on the spectrum of the radiation.

[0064] As described above in detail, in the second embodiment, the signal processing device 2 inputs a signal including a step wave before being converted into a pulse wave, in addition to a signal including a pulse wave, to the learning model 232 to detect overlapping of the pulse waves. When multiple pulse waves overlap, the interval between the step waves is short, and the waveform of the signal including the step wave is different from that when the pulse waves are not overlapped. Therefore, by using a signal including a step wave in addition to a signal including a pulse wave, it becomes easier to determine whether or not there is overlapping of the pulse waves. The signal processing device 2 can detect overlapping of multiple pulse waves more effectively than in the first embodiment, and the radiation detection device 10 can further improve the accuracy of elemental analysis based on the spectrum.

[0065] The signal processing device 2 may be configured to detect the overlap of multiple pulse waves using a signal including a step wave, instead of using a signal including a pulse wave. In this configuration, the learning model 232 is pre-trained to output information indicating whether or not there is an overlap of multiple pulse waves when a signal including a step wave is input. The input layer of the learning model 232 includes multiple nodes 41 to which a sequence of signal values ​​constituting the signal including a step wave is input. In learning of the learning model 232, the learning device 5 generates a step wave signal and an adjacent step wave signal, and performs machine learning of the learning model 232 using the multiple step wave signals and the adjacent step wave signals as teacher data. In S26, the signal processing device 2 does not input the sequence of signal values ​​constituting the signal including a pulse wave to the learning model 232, but inputs the sequence of signal values ​​constituting the signal including a step wave to the learning model 232. The learning model 232 similarly outputs information indicating whether or not there is an overlap of multiple pulse waves. The signal processing device 2 can effectively detect the overlap of multiple pulse waves, similarly to the first embodiment.

[0066] Furthermore, the signal processing device 2 may be configured to further input a signal including a shaped wave obtained by shaping the pulse wave by the waveform shaping unit 22 to the learning model 232 in addition to the signal including a pulse wave and / or the signal including a step wave, and detect the overlap of a plurality of pulse waves. The shaped wave corresponds to, for example, a second-order differential of a step wave. The learning model 232 is pre-trained to output information indicating whether or not there is an overlap of a plurality of pulse waves when a sequence of signal values ​​constituting a signal including a pulse wave and / or a sequence of signal values ​​constituting a signal including a step wave and a sequence of signal values ​​constituting a signal including a shaped wave are input. The input layer of the learning model 232 includes a plurality of nodes 41 to which a sequence of signal values ​​constituting a signal including a shaped wave is input. In the learning of the learning model 232, the learning device 5 generates a signal including a shaped wave and a signal obtained by shaping a superimposed wave signal, and performs machine learning using the pulse wave signal and / or the step wave signal, the signal including the shaped wave, the superimposed wave signal and / or the adjacent step wave signal, and the signal obtained by shaping the superimposed wave signal as teacher data. In S26, the signal processing device 2 inputs a sequence of signal values ​​constituting a signal including a pulse wave and / or a sequence of signal values ​​constituting a signal including a step wave, and a sequence of signal values ​​constituting a signal including a shaped wave to the learning model 232. The learning model 232 outputs information indicating whether or not there is overlap of multiple pulse waves. Even in this form, the signal processing device 2 can more effectively detect overlap of multiple pulse waves.

[0067] The signal processing device 2 may be configured to detect overlapping of pulse waves by inputting a feature of the signal waveform to the learning model 232 in addition to the signal including a pulse wave and / or a signal including a step wave, for example. The feature of the signal waveform is, for example, a wave height, a rise time constant, or a time width of the signal wave. The learning model 232 has been trained in advance to output information indicating whether or not there is overlapping of multiple pulse waves when a sequence of signal values ​​constituting a signal including a pulse wave and / or a sequence of signal values ​​constituting a signal including a step wave and a feature of the signal waveform are input. In S26, the signal processing device 2 inputs a sequence of signal values ​​constituting a signal including a pulse wave and / or a sequence of signal values ​​constituting a signal including a step wave and a feature of the signal waveform to the learning model 232. The learning model 2 outputs information indicating whether or not there is overlapping of multiple pulse waves. Furthermore, a signal including a shaped wave may also be used. Since the feature of the signal waveform differs between a single pulse wave and a superimposed wave, the overlapping of multiple pulse waves can be detected more effectively by using the feature of the signal waveform. Therefore, the radiation detection device 10 can further improve the accuracy of elemental analysis based on the spectrum.

[0068] <Embodiment 3> In the third embodiment, a learning model 232 determines the number of overlapping pulse waves. FIG. 11 is a block diagram showing a functional configuration of a radiation detection device 10 according to the third embodiment. The configurations and functions of the radiation detector 1 and the analysis unit 3 are similar to those of the first embodiment. The signal processing device 2 includes an A / D conversion unit 21, a waveform shaping unit 22, a processing unit 23, a pulse height measurement unit 25, and a counting unit 26. The processing unit 23 does not have a buffer memory. The processing unit 23 receives at least one of a signal including a step wave from the A / D conversion unit 21 and a signal including a pulse wave from the waveform shaping unit 22. The processing unit 23 inputs information indicating the number of overlapping pulse waves to the pulse height measurement unit 25.

[0069] FIG. 12 is a conceptual diagram showing an example of a functional configuration of the learning model 232 according to the third embodiment. The learning model 232 has been trained in advance so as to output the probability of the number of overlapping pulse waves when a sequence of signal values ​​constituting a signal including a pulse wave and / or a sequence of signal values ​​constituting a signal including a step wave is input. The output layer of the learning model 232 has a plurality of nodes 43. Each node 43 receives data from a plurality of nodes 42n, performs calculations on the received data using parameters, and outputs the probability that the number of overlapping pulse waves included in the signal is a specific number. For example, one node 43 outputs the probability that no pulse wave is included in the signal, another node 43 outputs the probability that multiple pulse waves are not overlapped and the number of pulse waves is one, and still another node 43 outputs the probability that the number of overlapping pulse waves is two. The node 43 may output the probability that the number of overlapping pulse waves is a specific number as a real number from 0 to 1, or as a binary value of 0 or 1.

[0070] In learning the learning model 232, the learning device 5 performs machine learning using the signal not including a pulse wave and a step wave, the pulse wave signal and / or the step wave signal, and the superimposed wave signal and / or the adjacent step wave signal as teacher data. The learning device 5 associates the signal not including a pulse wave and a step wave with information in which the probability of no pulse wave is 1 and the other probabilities are 0, associates the pulse wave signal and / or the step wave signal with information in which the probability of one pulse wave is 1 and the other probabilities are 0, and associates the superimposed wave signal and / or the adjacent step wave signal with information in which the probability of multiple pulse waves is 1 and the other probabilities are 0. The learning device 5 calculates the error of the information using an error function with the variables being the respective probabilities indicated by the information associated with the signal input to the input layer and the respective probabilities output from the output layer, and adjusts the parameters of the calculations of each node of the learning model 232 so that the error is minimized by the error backpropagation method.

[0071] 13 is a flowchart showing the procedure of processing executed by the signal processing device 2 according to the third embodiment. The signal processing device 2 receives a signal including a step wave from the radiation detector 1 (S31), and the A / D conversion unit 21 A / D converts the signal (S32). The A / D conversion unit 21 inputs the A / D converted signal to the waveform shaping unit 22. The waveform shaping unit 22 shapes the waveform of the input signal (S33). The step wave is converted into a pulse wave by the waveform shaping. The waveform shaping unit 22 inputs the signal including the pulse wave to the pulse height measurement unit 25.

[0072] The processing unit 23 receives the signal including the step wave from the A / D conversion unit 21 and / or the signal including the pulse wave from the waveform shaping unit 22, and inputs a sequence of signal values ​​constituting the input signal to the learning model 232 (S34). As described above, the learning model 232 performs neural network calculations and outputs the probability of the number of overlapping pulse waves. The processing unit 23 inputs information indicating the number of overlapping pulse waves to the wave-height measurement unit 25. At this time, the processing unit 23 may input the probability output by the learning model 232 to the wave-height measurement unit 25 as the information indicating the number of overlapping pulse waves. The processing unit 23 may also determine the number of overlapping pulse waves based on the probability, and input information indicating the determined number to the wave-height measurement unit 25 as the information indicating the number of overlapping pulse waves. For example, the processing unit 23 determines that the number with the maximum probability or the number with the probability equal to or greater than a predetermined value is the number of overlapping pulse waves.

[0073] The pulse-height measuring unit 25 determines whether or not the number of pulse waves is one based on the input information (S35). If the number of pulse waves is one (S35: YES), the pulse-height measuring unit 25 measures the height of the pulse wave contained in the input signal (S36). The pulse-height measuring unit 25 inputs the measured pulse-height to the counting unit 26. The counting unit 26 counts the pulse waves by wave height (S37) and ends the process. If the number of pulse waves is other than one (S35: NO), the pulse-height measuring unit 25 does not measure the pulse height, and the signal processing device 2 ends the process. As a result, the counting unit 26 does not count multiple overlapping pulse waves. The signal processing device 2 repeatedly executes the processes of S31 to S37.

[0074] The signal processing device 2 outputs data indicating the relationship between the pulse wave height and the number of counts counted by the counting unit 26. The analysis unit 3 receives the data output by the signal processing device 2 and generates a spectrum of the radiation detected by the radiation detector 1 based on the data. The analysis unit 3 may perform elemental analysis of the radiation source based on the radiation spectrum. In this embodiment as well, the signal processing device 2 can effectively detect the overlap of multiple pulse waves, and the radiation detection device 10 can further improve the accuracy of elemental analysis based on the spectrum.

[0075] The signal processing device 2 may be configured to detect the overlap of a plurality of pulse waves by also using a signal including a shaped wave obtained by further shaping a pulse wave by the waveform shaping unit 22. The signal processing device 2 may be configured to detect the overlap of a plurality of pulse waves by also using a feature amount of the signal waveform.

[0076] <Embodiment 4> In the fourth embodiment, a learning model 232 determines the number and height of step waves. FIG. 14 is a block diagram showing a functional configuration of a radiation detection device 10 according to the fourth embodiment. The configurations and functions of the radiation detector 1 and the analysis unit 3 are similar to those of the first embodiment. The signal processing device 2 includes an A / D conversion unit 21, a processing unit 23, and a counting unit 26. The A / D conversion unit 21 inputs a signal including a step wave to the processing unit 23.

[0077] FIG. 15 is a conceptual diagram showing an example of a functional configuration of the learning model 232 according to the fourth embodiment. The learning model 232 has been trained in advance to output the probability of the number of staircase waves included in the signal and the wave height of the staircase waves when a sequence of signal values ​​constituting a signal including a staircase wave is input. The output layer of the learning model 232 has a plurality of nodes 43. Each node 43 receives data from a plurality of nodes 42n and performs calculations using parameters on the received data. One node 43 outputs the wave height of the staircase wave included in the signal. The other nodes 43 output the probability of the number of staircase waves included in the signal. For example, there is a node 43 that outputs the probability that a staircase wave is not included in the signal, a node 43 that outputs the probability that the number of staircase waves is one, and a node 43 that outputs the probability that the number of staircase waves is two. The node 43 may output the probability that the number of staircase waves is a specific number as a real number from 0 to 1, or may output it as a binary value of 0 or 1. The probability of the number of staircase waves included in the signal is information on the presence or absence of overlap of a plurality of pulse waves converted from a plurality of staircase waves.

[0078] In learning the learning model 232, the learning device 5 performs machine learning using a signal not including a step wave, a step wave signal, and an adjacent step wave signal as teacher data. The learning device 5 associates with the signal not including a step wave information indicating that the wave height is zero, that the probability of there being no step wave is 1, and the other probabilities are 0. The learning device 5 associates with the step wave signal information indicating that the wave height of the step wave and the probability that there is one step wave are 1, and the other probabilities are 0. The learning device 5 associates with the adjacent step wave signal information indicating that the sum of the wave heights of multiple step waves and the probability that there are multiple step waves are 1, and the other probabilities are 0. The learning device 5 calculates the error of the information by an error function with the wave height and each probability indicated by the information associated with the signal input to the input layer and each probability output from the output layer as variables, and adjusts the parameters of the calculation of each node of the learning model 232 so that the error is minimized by the error backpropagation method.

[0079] 16 is a flowchart showing the procedure of processing executed by the signal processing device 2 according to the fourth embodiment. The signal processing device 2 receives a signal including a step wave from the radiation detector 1 (S41), and the A / D conversion unit 21 A / D converts the signal (S42). The A / D conversion unit 21 inputs the A / D converted signal to the processing unit 23.

[0080] The processing unit 23 inputs a sequence of signal values ​​constituting the input signal to the learning model 232 (S43). As described above, the learning model 232 performs neural network calculations and outputs the probability of the wave height and number of step waves of the step waves included in the signal. The processing unit 23 inputs information indicating the wave height and the number of step waves to the counting unit 26. At this time, the processing unit 23 may input the probability output by the learning model 232 to the counting unit 26 as information indicating the number of step waves. The processing unit 23 may also determine the number of step waves based on the probability, and input information indicating the determined number to the counting unit 26 as information indicating the number of step waves. For example, the processing unit 23 determines the number with the maximum probability or the number with the probability equal to or greater than a predetermined value to be the number of step waves.

[0081] The counting unit 26 determines whether the number of staircase waves is one based on the input information (S44). If the number of staircase waves is one (S44: YES), the counting unit 26 counts the staircase waves by wave height (S45) and ends the process. As a result, the counting unit 26 counts staircase waves in the case where there is no overlap of multiple pulse waves converted from multiple staircase waves. If the number of staircase waves is a number other than one (S44: NO), the counting unit 26 does not count, and the signal processing device 2 ends the process. As a result, the counting unit 26 does not count staircase waves in the case where multiple pulse waves converted from multiple staircase waves overlap. The signal processing device 2 repeatedly executes the processes of S41 to S45.

[0082] The signal processing device 2 outputs data indicating the relationship between the wave height of the step wave and the number of counts counted by the counting unit 26. The analysis unit 3 receives the data output by the signal processing device 2 and generates a spectrum of the radiation detected by the radiation detector 1 based on the data. The analysis unit 3 may perform elemental analysis of the radiation source based on the spectrum of the radiation. In this embodiment as well, the signal processing device 2 can efficiently detect the overlap of multiple pulse waves converted from multiple step waves, and the radiation detection device 10 can further improve the accuracy of elemental analysis based on the spectrum. The signal processing device 2 may be configured to detect the overlap of multiple pulse waves by also using the feature amount of the signal waveform.

[0083] <Embodiment 5> 17 is a block diagram showing a functional configuration of the processing unit 23 according to the fifth embodiment. The processing unit 23 has a calculation unit 233 and a memory 234. The calculation unit 233 is configured using, for example, a CPU, a GPU, or a multi-core CPU. The calculation unit 233 may be configured using a quantum computer. The memory 234 is a non-volatile memory. The memory 234 stores a computer program 235. The calculation unit 233 executes processing required for the processing unit 23 according to the computer program 235. The learning model 232 is realized by the calculation unit 233 executing information processing according to the computer program 235.

[0084] The calculation unit 233 executes information processing according to the computer program 235, thereby executing processing required for the processing unit 23 in the first to fourth embodiments. In this manner, the processing unit 23 in the first to fourth embodiments is realized. The configurations and functions of the radiation detector 1 and the analysis unit 3 are the same as those in the first to fourth embodiments. The configurations and functions of the parts of the signal processing device 2 other than the processing unit 23 are the same as those in the first to fourth embodiments. The signal processing device 2 and the radiation detection device 10 execute the same processing as those in the first to fourth embodiments. In this embodiment as well, the signal processing device 2 can efficiently detect the overlap of a plurality of pulse waves converted from a plurality of step waves, and the radiation detection device 10 can further improve the accuracy of the elemental analysis based on the spectrum. Note that a part or all of the parts of the signal processing device 2 other than the processing unit 23 may also be realized using a computer program.

[0085] <Embodiment 6> In the sixth embodiment, the signal processing device 2 effectively detects the overlap of the pulse waves without using a learning model. FIG. 18 is a block diagram showing the functional configuration of the radiation detection device 10 according to the sixth embodiment. The configuration and function of the radiation detector 1 and the analysis unit 3 are the same as those of the first embodiment. The signal processing device 2 includes an A / D conversion unit 21, a waveform shaping unit 22, a processing unit 23, a pulse detection unit 24, a pulse height measurement unit 25, a counting unit 26, and a storage unit 27. The processing unit 23 does not include a learning model. The storage unit 27 is connected to the processing unit 23. The storage unit 27 is non-volatile. For example, the storage unit 27 is composed of a non-volatile semiconductor memory. The storage unit 27 stores a predetermined coefficient sequence. The configuration of the signal processing device 2 other than the processing unit 23 and the storage unit 27 is the same as that of the first embodiment.

[0086] 19 and 20 are graphs showing examples of coefficient sequences. In the diagrams, the horizontal axis indicates time in units of clocks, and the vertical axis indicates coefficient values ​​or signal values. A coefficient sequence is a one-dimensional array of multiple coefficients. In FIG. 19, a signal including a single pulse wave is shown by a dashed line, and the coefficient sequence is shown by a solid line. The sequence of signal values ​​that make up the signal including a pulse wave is shown arranged at clock intervals. In addition, the multiple coefficients included in the coefficient sequence are shown arranged in order at clock intervals. The multiple coefficients included in the coefficient sequence are for performing product-sum operations with the sequence of signal values ​​that make up the signal including a pulse wave. The number of coefficients included in the coefficient sequence is m. Let the coefficient sequence be a1, a2, ..., a m , and the sequence of signal values ​​that make up the signal containing the pulse wave is x1, x2, .... The multiply-and-accumulate operation is (a1*x1+a2*x2+...+a m *x m +C), where C is a constant value. The constant value C is stored in the storage unit 27.

[0087] As shown in FIG. 19, the coefficient sequence rises almost simultaneously with the pulse wave, increases and decreases in a period shorter than that of the pulse wave, becomes almost zero near the peak of the pulse wave, becomes a negative value thereafter, and slowly approaches the value of zero. The coefficient sequence is set so that the result of a multiplication and accumulation operation with a sequence of signal values ​​constituting a signal including a single pulse wave is almost zero. As shown in FIG. 19, the value of the signal including a pulse wave becomes almost zero at the time of clock 150, and the multiplication and accumulation operation thereafter is almost zero. Before the time of clock 150, the result of a multiplication and accumulation operation with a signal value before the peak of the pulse wave and a positive coefficient is a positive value, and the result of a multiplication and accumulation operation with a signal value after the peak of the pulse wave and a negative coefficient is a negative value, and the change in the value of the coefficient is determined so that the total is almost zero. Therefore, the result of a multiplication and accumulation operation with a sequence of signal values ​​constituting a signal including a single pulse wave and the coefficient sequence is almost zero. In this way, the coefficient sequence is determined so that the result of a multiplication and accumulation operation with a signal including a single pulse wave is almost zero.

[0088] In FIG. 20, the signal including the superimposed wave is shown by a dashed line, and the coefficient sequence is shown by a solid line. The coefficient sequence shown in FIG. 19 and FIG. 20 is the same. As shown in FIG. 20, since the time width of the superimposed wave is larger than that of a single pulse wave, the signal value of the signal including the superimposed wave maintains a positive value even after the time of clock 150. In addition, the coefficient sequence has a negative value after the time of clock 150. Therefore, the result of the product-sum operation between the sequence of signal values ​​constituting the signal including the superimposed wave and the coefficient sequence becomes a negative value after the time of clock 150. Before the time of clock 150, the result of the product-sum operation is almost zero, as with a single pulse wave. Therefore, the result of the product-sum operation between the sequence of signal values ​​constituting the signal including the superimposed wave and the coefficient sequence becomes a negative value. In this way, the coefficient sequence is determined so that the result of the product-sum operation with the signal including the superimposed wave becomes a negative value. In other words, a predetermined value less than zero is used as a reference value, and the coefficient sequence is determined so that the result of a product-sum operation with a signal including a single pulse wave is greater than the reference value, and the result of a product-sum operation with a signal including a superimposed wave is less than the reference value.

[0089] 21 is a flowchart showing the procedure of processing executed by the signal processing device 2 according to embodiment 6. The signal processing device 2 executes the processes of S51 to S55 similar to the processes of S21 to S25 in embodiment 1. When a pulse wave contained in a signal is detected (S25: YES), the pulse detection unit 24 inputs information indicating that a pulse wave has been detected to the processing unit 23, and the processing unit 23 performs a product-sum operation on the sequence of signal values ​​constituting the input signal and the sequence of coefficients stored in the storage unit 27 (S56).

[0090] The processing unit 23 judges whether or not the input signal includes overlapping pulse waves according to the result of the product-sum calculation (S57). In S57, the processing unit 23 judges that there is overlapping of multiple pulse waves when the result of the product-sum calculation is less than a reference value less than zero, and judges that there is no overlapping of multiple pulse waves when the result of the product-sum calculation is equal to or greater than the reference value. The reference value is stored in advance in the processing unit 23 or the storage unit 27. The processing unit 23 inputs information indicating whether or not there is overlapping of multiple pulse waves to the pulse-height measurement unit 25. If there is no overlapping of multiple pulse waves (S57: NO), the pulse-height measurement unit 25 measures the pulse height of the pulse wave included in the input signal (S58) and inputs the measured pulse height to the counting unit 26. The counting unit 26 counts the pulse waves for each pulse height input from the pulse-height measurement unit 25 (S59), and ends the process.

[0091] If there is overlap of multiple pulse waves (S57: YES), the pulse-height measuring unit 25 does not measure the pulse height of the overlapping pulse waves, and the signal processing device 2 ends the process. As a result, the counting unit 26 does not count the overlapping pulse waves. The signal processing device 2 repeatedly executes the processes of S51 to S59 individually. The signal processing device 2 outputs data indicating the relationship between the pulse wave height and the count number. The data output by the signal processing device 2 is input to the analysis unit 3, which generates a spectrum of the radiation detected by the radiation detector 1 based on the data. The analysis unit 3 may perform elemental analysis of the radiation source based on the spectrum of the radiation.

[0092] The coefficient sequence may be determined so that the result of the product-sum operation with the sequence of signal values ​​constituting the signal including a single pulse wave is smaller than a reference value, and the result of the product-sum operation with the sequence of signal values ​​constituting the signal including a superimposed wave is larger than the reference value. For example, the coefficient sequence may be determined so that the result of the product-sum operation with the sequence of signal values ​​constituting the signal including a single pulse wave and the coefficient sequence is approximately zero, and the result of the product-sum operation with the signal including a superimposed wave is a positive value. In this embodiment, in S57, the processing unit 23 determines that there is overlap of multiple pulse waves when the result of the product-sum operation exceeds the reference value, using a predetermined value exceeding zero as the reference value, and determines that there is no overlap of multiple pulse waves when the result of the product-sum operation is equal to or smaller than the reference value.

[0093] As described above in detail, in the sixth embodiment, the signal processing device 2 performs a product-sum operation between a sequence of signal values ​​constituting a signal including a pulse wave and a predetermined sequence of coefficients, and detects overlapping of pulse waves based on the result of the product-sum operation. In the sixth embodiment as well, the signal processing device 2 can effectively detect overlapping of pulse waves, and the radiation detection device 10 can further improve the accuracy of elemental analysis based on a spectrum. Since the determination is performed by a product-sum operation without using a learning model, the signal processing device 2 can execute processing at a higher speed.

[0094] The signal processing device 2 may be in a form that executes both the method using the learning model 232 shown in the first or second embodiment and the method using the result of the product-sum calculation shown in the sixth embodiment. For example, the signal processing device 2 determines whether or not there is an overlap of pulse waves by both the method using the learning model 232 and the method using the result of the product-sum calculation, and when it is determined that there is an overlap of pulse waves by either one of the methods, it determines that an overlap of pulse waves has occurred. In the first to sixth embodiments, the signal processing device 2 includes the counting unit 26, but the counting unit 26 may be provided outside the signal processing device 2. In addition, in the first to sixth embodiments, the radiation detection element 11 is a semiconductor radiation detection element, but the radiation detector 1 may be in a form that uses a radiation detection element 11 other than a semiconductor radiation detection element. For example, the radiation detector 1 may be a scintillation detector.

[0095] The present invention is not limited to the contents of the above-described embodiment, and various modifications are possible within the scope of the claims. In other words, embodiments obtained by combining technical means appropriately modified within the scope of the claims are also included in the technical scope of the present invention. [Explanation of symbols]

[0096] 1. Radiation detector 10 Radiation detection equipment 11 Radiation detection element 2. Signal Processing Device 21 A / D conversion section 22 Waveform shaping section 23 Processing section 232 Learning Model 235 Computer Programs 25 Wave height measurement section 26 Counting Section 27 Memory section 3 Analysis Department 5 Learning Device

Claims

1. 1. A signal processing method for counting a step wave or a pulse wave converted from the step wave in response to detection of radiation, by wave height, comprising: inputting at least one of the sequence of signal values ​​constituting the signal including the step wave and the sequence of signal values ​​constituting the signal including the pulse wave into a learning model which outputs information regarding the presence or absence of overlap of a plurality of the pulse waves when at least one of the sequence of signal values ​​constituting the signal including the step wave and the sequence of signal values ​​constituting the signal including the pulse wave is input; Counting the step wave or the pulse wave when there is no overlap of the plurality of pulse waves according to the information output by the learning model. A signal processing method comprising:

2. According to the information output by the learning model, the step wave or the pulse wave is not counted when a plurality of the pulse waves overlap.

2. The signal processing method according to claim 1, further comprising:

3. the learning model is a learning model that outputs the information when a sequence of signal values ​​constituting a signal including a wave obtained by shaping the pulse wave is further input, A sequence of signal values ​​constituting the signal is further input to the learning model.

3. The signal processing method according to claim 1 or 2, characterized in that:

4. the learning model is a learning model that outputs the information when a feature amount of at least one of the step wave and the pulse wave is further input, Inputting the feature amount into the learning model.

4. A signal processing method according to claim 1, further comprising:

5. The information includes information indicating whether or not a plurality of the pulse waves overlap, a probability of the number of overlapping pulse waves, or a probability of the number of step waves included in the signal.

5. A signal processing method according to claim 1, further comprising:

6. generating a plurality of pulse wave signals each of which is composed of a single pulse wave, and at least one of a wave height, a rise, and a drop of the single pulse wave being random; generating a plurality of superimposed wave signals each of which is composed of a plurality of superimposed pulse waves, the plurality of pulse waves having at least one of an interval, a wave height, a rise, and a decay time being random; generating a learning model that outputs information regarding the presence or absence of overlap of a plurality of pulse waves when a sequence of signal values ​​constituting a signal including any pulse wave is input using the plurality of pulse wave signals and the plurality of superimposed wave signals as teacher data; A learning model generation method characterized by:

7. generating a plurality of step wave signals each of which is composed of a single step wave, and at least one of a wave height, a rising edge, and a decay rate of the single step wave being random; A plurality of adjacent step wave signals are generated, each of which is composed of a plurality of step waves, and a plurality of pulse waves obtained by converting the plurality of step waves are overlapped, and at least one of the intervals, wave heights, rising edges, and rounding edges of the plurality of step waves is random. A learning model is generated that outputs information regarding the presence or absence of overlap of a plurality of pulse waves obtained by converting the plurality of step waves when a sequence of signal values ​​constituting a signal including any step wave is input using the plurality of step wave signals and the adjacent step wave signals as teacher data. A learning model generation method characterized by:

8. 1. A signal processing method for converting a step wave corresponding to detection of radiation into a pulse wave and counting the pulse wave according to the wave height of the pulse wave, comprising: performing a product-sum operation on a sequence of signal values ​​constituting the signal including the pulse wave and a predetermined sequence of coefficients; the predetermined coefficient sequence is determined in advance such that a result of a multiplication and accumulation operation between a sequence of signal values ​​constituting a signal including a plurality of overlapping pulse waves and the predetermined coefficient sequence is different from a result of a multiplication and accumulation operation between a sequence of signal values ​​constituting a signal including a single pulse wave and the predetermined coefficient sequence in terms of whether it is larger or smaller than a predetermined value; determining whether or not there is overlap of the plurality of pulse waves according to a result of the product-sum calculation; Counting the non-overlapping pulse waves. A signal processing method comprising:

9. 1. A signal processing device for processing a signal including a step wave in response to detection of radiation, comprising: a learning model that outputs information regarding the presence or absence of overlap of a plurality of pulse waves when at least one of a sequence of signal values ​​constituting the signal including the step wave and a sequence of signal values ​​constituting the signal including a pulse wave obtained by converting the step wave is input. A signal processing device comprising:

10. a radiation detector that outputs a step wave corresponding to the energy of radiation when radiation is detected; A conversion unit that converts the step wave into a pulse wave; a learning model that outputs information regarding the presence or absence of overlap of a plurality of the pulse waves when at least one of a sequence of signal values ​​constituting the signal including the step wave and a sequence of signal values ​​constituting a signal including a pulse wave obtained by converting the step wave is input; A counting unit that counts the step wave or the pulse wave when there is no overlap of the plurality of pulse waves according to the information output by the learning model; a spectrum generating unit that generates a spectrum of radiation according to the wave height and count number of the step wave or the pulse wave; A radiation detection device comprising:

11. At least one of a sequence of signal values ​​constituting a signal including a step wave corresponding to the detection of radiation and a sequence of signal values ​​constituting a signal including a pulse wave obtained by converting the step wave is input to a learning model which outputs information regarding the presence or absence of overlap of a plurality of pulse waves when the sequence of signal values ​​constituting the signal including the step wave and the sequence of signal values ​​constituting the signal including the pulse wave are input, and the information is output. A computer program that causes a computer to execute a process.

Images