Neutron beam current real-time regulation and control method and system based on two parameters

Abstract

The invention discloses a neutron beam real-time regulation and control method and system based on two parameters, and the method comprises the steps: collecting the two parameters and a target beam current in real time, and obtaining a data set; an LSTM-Attention initial model is constructed; training, verifying and testing the LSTM-Attention initial model by using the training set, the verification set and the test set to obtain a final neutron beam current control model; the deviation between the target beam current and a set value is monitored in real time, a final neutron beam control model is used for inversion to obtain a two-parameter adjusting quantity, and real-time neutron beam regulation and control are achieved; according to the method, the coupling relation between the two parameters and the beam is considered, the current and the air inflow of the ion source can be adjusted in real time, stable operation of the neutron source beam is ensured, the neutron yield stability is improved, and the application effect in practical application can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of neutron generator control, in particular to a neutron beam current real-time regulation method and system based on double parameters. BACKGROUND

[0002] The compact neutron source is a neutron generating device realized through miniaturization technology design relative to the traditional large neutron source (such as reactor neutron source and large spallation neutron source), and the core is to generate high-efficiency neutron beam through nuclear reaction (such as deuterium-deuterium and deuterium-beryllium reaction) in a limited space, which has the dual characteristics of "miniaturization" and "functionality", and can meet the diversified scene of neutron application demand, such as material nondestructive testing, small neutron imaging, nuclear physics basic research, medical radioisotope preparation, etc. The overall device volume of the compact neutron source is much smaller than that of the large neutron source, and can meet the needs of medium and low flux demand scenes, without the need for large-scale infrastructure, and the manufacturing and operation and maintenance costs are significantly reduced, which can be flexibly deployed in laboratory, industrial site and other scenes.

[0003] At present, the operation of the compact neutron source depends on the coordinated control of multiple key parameters, and these parameters do not act independently, but are coupled and related in a complex way. However, the current control process is mainly single parameter adjustment, ignoring the coupling effect between multiple parameters, which is easy to cause unstable neutron beam, large fluctuation of beam current to target, and further cause defects such as large decline of neutron yield stability, low overall efficiency, short service life, high safety risk, etc.

[0004] Therefore, an effective method is needed to solve these problems and realize the stability control of the compact neutron source beam. SUMMARY

[0005] The present application aims to provide a neutron beam current real-time regulation method and system based on double parameters to solve the technical problems that the control method of the compact neutron source in the prior art does not consider the coupling effect of multiple parameters, which is easy to cause unstable neutron beam and large fluctuation of beam current to target during operation. The present application establishes a neutron source beam stability control training model, adjusts the ion source current and gas inlet amount in real time, ensures the stable operation of the neutron source beam, improves the neutron yield stability, improves the imaging and element analysis accuracy, and prolongs the service life of the neutron target.

[0006] In order to solve the above technical problems, the present application adopts the following technical scheme:

[0007] In a first aspect, the present application provides a neutron beam current real-time regulation method based on double parameters, which is used for regulating the compact neutron source, and includes the following steps:

[0008] S1, real-time acquisition of ion source current, gas inlet amount and target beam current to obtain a data set, sorting the data set in time sequence, and dividing into a training set, a validation set and a test set;

[0009] S2, constructing an LSTM network layer and adding an attention layer after the LSTM network layer to obtain an LSTM-Attention initial model;

[0010] S3, training, validation and testing of the LSTM-Attention initial model using the training set, the validation set and the test set respectively to obtain a final neutron beam current control model;

[0011] S4, real-time monitoring of the deviation of the target beam current from the set value, and using the final neutron beam current control model to perform inversion, outputting the corresponding ion source current adjustment amount and / or gas inlet amount adjustment amount according to the hierarchical strategy, or outputting a shutdown instruction;

[0012] S5, adjusting the ion source current and / or gas inlet amount according to the output ion source current adjustment amount and / or gas inlet amount adjustment amount, or executing the shutdown instruction; repeating steps S4 and S5 to realize real-time regulation of the neutron beam current.

[0013] Further, in step S3, training, validation and testing of the LSTM-Attention initial model using the training set, the validation set and the test set respectively include the following steps:

[0014] S3.1, training the LSTM-Attention initial model using the training set to obtain a neutron beam current control model;

[0015] S3.2, verifying the neutron beam current control model using the validation set to obtain a verification result;

[0016] S3.3, adjusting the model parameters of the neutron beam current control model according to the verification result to obtain an optimal neutron beam current control model;

[0017] S3.4, testing the optimal neutron beam current control model using the test set;

[0018] If the test is qualified, a final neutron beam current control model is obtained, and step S4 is continued;

[0019] If the test is not qualified, steps S3.1 to S3.4 are returned to be re-executed.

[0020] Further, in step S2, the constructed LSTM network layer comprises a plurality of LSTM layers stacked in series. The use of a series design of multi-layer LSTM significantly improves the prediction accuracy and control robustness of the model for neutron source beam fluctuations by hierarchical feature learning, enhancing generalization ability and supporting complex time series modeling, providing key technical support for achieving a target beam current stability within ±5%.

[0021] Further, in step S1, before sorting the data set in time sequence, the data in the data set is first preprocessed, including the following steps:

[0022] S1.1, data cleaning: calculate the standard deviation and mean value of the data set corresponding to the same parameter, and calculate the ratio of each difference value to the standard deviation;

[0023] If the absolute value of the ratio is greater than or equal to 3, it is considered as an outlier and is removed;

[0024] If the absolute value of the ratio is less than 3, continue to step S1.2;

[0025] S1.2, normalize the data cleaned data to obtain a preprocessed data set, sort the preprocessed data set in time sequence, and divide it into a training set, a validation set and a test set, and then continue to step S2.

[0026] Further, in step S3.2, the validation result refers to:

[0027] First, input the ion source current and gas flow in the validation set into the neutron beam current control model to obtain a target beam current array;

[0028] Then, the model is verified and evaluated using evaluation indicators to obtain a validation result;

[0029] Specifically, the evaluation indicators refer to one or a combination of several of the mean square error, mean absolute error, and determination coefficient.

[0030] In step S3.3, adjusting the model parameters of the neutron beam current control model according to the validation result refers to predefining a threshold for the evaluation indicator and comparing it with the validation result;

[0031] If the validation result is within the threshold, the model parameters do not need to be adjusted, and the current model is used as the optimal neutron beam current control model;

[0032] If the validation result is outside the threshold range, the model parameters are adjusted until the validation result is within the threshold.

[0033] Further, in step S3.4, when the optimal neutron beam current control model is tested by using the test set, one or a combination of several of mean square error, mean absolute error, and determination coefficient is used as an evaluation index.

[0034] Further, in step S4, the grading strategy refers to grading according to the deviation of the target beam current from the set value, and first calculating the absolute value percentage of the deviation of the target beam current from the set value;

[0035] If the absolute value percentage of the deviation is ≤5%, and the duration is >5s, output the ion source current adjustment amount;

[0036] If 5%≤the absolute value percentage of the deviation≤10%, and the duration is >5s, output the ion source current adjustment amount and the gas adjustment amount;

[0037] If the absolute value percentage of the deviation is >10%, and the duration is >15s, start safety protection and automatically stop machine verification.

[0038] In a second aspect, the present application further provides a neutron beam current real-time regulation system based on double parameters, which is used to realize the neutron beam current real-time regulation method based on double parameters as described above, and comprises,

[0039] A data acquisition module is configured to acquire sensor data;

[0040] A data processing module is configured to store and run the final neutron beam current control model, and output corresponding ion source current adjustment amount and / or gas amount adjustment amount, or output a stop command;

[0041] An execution module is configured to adjust the ion source current and / or the gas amount according to the output ion source current adjustment amount and / or the gas amount adjustment amount, or execute the output stop command.

[0042] In a third aspect, the present application further provides an electronic device, characterized in that it comprises a processor and a memory, and the processor is coupled with the memory;

[0043] The memory is configured to store a computer program, and the processor is configured to execute the computer program stored in the memory to realize the neutron beam current real-time regulation method based on double parameters as described above.

[0044] In a fourth aspect, the present application further provides a computer readable storage medium configured to store a computer program, and the computer program is executed by a processor to realize the neutron beam current real-time regulation method based on double parameters as described above.

[0045] The present application has the following beneficial effects: the present application establishes a neutron beam current control model based on double parameters, obtains the coupling relationship between the double parameters and the beam current, can realize real-time adjustment of the ion source current and the gas inlet amount by monitoring the deviation of the target beam current from the set value in real time, ensures the stable operation of the neutron source beam, improves the stability of the neutron yield, and further improves the application effect in actual application. BRIEF DESCRIPTION OF DRAWINGS

[0046] In order to make the purpose, technical scheme and advantages of the application clearer, the application will be further described in detail below with reference to the drawings, in which:

[0047] Figure 1 The flowchart for establishing the final neutron beam current control model of the embodiment of the present application.

[0048] Figure 2 The flowchart for the real-time regulation method of the neutron beam based on double parameters of the embodiment of the present application. DETAILED DESCRIPTION

[0049] The technical schemes in some embodiments disclosed by the present application will be described clearly and completely below with reference to the drawings. Obviously, the described embodiments are only some of the embodiments disclosed by the present application, not all the embodiments. Based on the embodiments provided by the present application, all other embodiments obtained by those skilled in the art belong to the scope of protection of the present application. It should be noted that, in the drawings with reference numerals, the same reference numerals and letters represent similar parts. Once a part is defined in one drawing, it will not be repeatedly defined and explained in subsequent drawings.

[0050] The present application can be applied to the field of neutron generator beam stability control, especially to the field of beam stability control of compact neutron sources such as deuterium-deuterium neutron sources, and solves the technical problem that the control method of the compact neutron source in the prior art does not consider the coupling effect of multiple parameters, which easily leads to unstable neutron beam and large fluctuation of target beam current during operation.

[0051] The technical concept of the present application is that the existing compact neutron source has many defects in the operation process. First, most of them use open-loop control, so that the plasma density is easily disturbed by factors such as temperature and electrode aging, which further leads to fluctuations in neutron beam. Second, in the adjustment mode, it is mostly single parameter adjustment, which ignores the coupling effect between the two.

[0052] Especially for the two parameters of gas flow and microwave power, the gas flow affects the plasma density, the microwave power affects the electron temperature, and the microwave power is related to the size of the ion source current. If the gas flow or the ion source current is adjusted independently, when the gas flow increases, although the plasma density will rise, the electron temperature will decrease, thereby reducing the ionization rate of deuterium ions. When the microwave power is too high, the electron temperature rises, and the density fluctuation intensifies, causing beam divergence. In addition, the beam current to the target fluctuates greatly, usually more than ±10%, which not only leads to unstable neutron yield, but also adversely affects the imaging and element analysis accuracy. Finally, beam instability can cause local overheating of the target material, greatly shortening the service life of the neutron target.

[0053] The technical effects of the neutron beam real-time regulation method and system based on double parameters disclosed by the application are as follows:

[0054] 1. By establishing a neutron beam control model based on double parameters, the coupling relationship between the double parameters and the beam current is obtained. The deviation of the beam current to the target from the set value can be monitored in real time to realize real-time adjustment of the ion source current and the gas flow, ensure stable operation of the neutron source beam, improve the stability of the neutron yield, and thus improve the application effect in actual application.

[0055] 2. The application adopts double-parameter feedback regulation and combines the innovative LSTM-Attention model, fully considers the coupling effect of the gas flow and the ion source current, can effectively suppress beam fluctuation, and solves the problems caused by the open-loop control and single-parameter regulation of the existing compact neutron source.

[0056] 3. Through accurate beam stability control, the beam current fluctuation to the target is greatly reduced, the neutron yield stability is improved, the imaging and element analysis accuracy are improved, the demand for high-precision measurement in related fields is met, and the popularization and application of the compact neutron source are improved.

[0057] 4. The application can avoid the problem of local overheating of the target material caused by beam instability, significantly prolong the service life of the neutron target, reduce the equipment maintenance cost and replacement frequency, and improve the reliability and stability of the equipment.

[0058] In order to further illustrate the neutron beam real-time regulation method and system based on double parameters provided by the application, the following embodiments are disclosed.

[0059] In some embodiments, referring to Figure 1 The embodiment provides a neutron beam real-time regulation method based on double parameters, which is used for regulating a compact neutron source, and comprises the following steps:

[0060] S1, collecting data sets of ion source current, gas inlet amount and target beam current in real time, sorting the data sets in time sequence, and dividing the data sets into a training set, a validation set and a test set;

[0061] S2, constructing an LSTM network layer and adding an attention layer after the LSTM network layer to obtain an LSTM-Attention initial model;

[0062] S3, training, validating and testing the LSTM-Attention initial model by using the training set, the validation set and the test set respectively to obtain a final neutron beam current control model;

[0063] S4, monitoring the deviation of the target beam current from the set value in real time, and inverting the final neutron beam current control model to output corresponding ion source current adjustment amount and / or gas inlet amount adjustment amount or a shutdown instruction according to a hierarchical strategy;

[0064] S5, adjusting the ion source current and / or the gas inlet amount according to the output ion source current adjustment amount and / or gas inlet amount adjustment amount or executing the shutdown instruction; repeating steps S4 and S5 to realize real-time regulation of the neutron beam current.

[0065] In some embodiments, the gas inlet amount refers to deuterium gas inlet amount or tritium gas inlet amount.

[0066] In some embodiments, in step S1, the division ratio of the training set, the validation set and the test set is preferably 7:2:1.

[0067] Specifically, the training set, the validation set and the test set also need to be sorted in time sequence when being divided.

[0068] In some embodiments, a current sensor is connected in series in an ion source power supply circuit to collect ion source current data in real time; a gas mass flow meter is installed in a deuterium / tritium gas inlet pipeline to measure gas inlet amount; a neutron source high-voltage power supply current sensor is used to measure target beam current; and a data acquisition card is used to synchronously collect and transmit data.

[0069] In some embodiments, the accuracy of the current sensor is ±0.5%, and the sampling frequency is 1 kHz.

[0070] In some embodiments, the range of the gas mass flow meter is 0-50 sccm, the response time is <100 ms, and the gas mass flow meter can output a 4-20 mA standard signal corresponding to the gas inlet amount.

[0071] In some embodiments, the neutron source high-voltage power supply current sensor is integrated in the neutron source high-voltage power supply to monitor the target current in real time.

[0072] In some embodiments, the data acquisition card simultaneously acquires three signals: one from a current sensor, one from a gas mass flow meter, and one from a neutron source high-voltage power supply current sensor. The data acquisition card has 16-bit resolution, a sampling frequency of 1kHz, and transmits data via wired or wireless means. The acquisition frequency of the data acquisition card is set according to the operating characteristics of the neutron source to ensure that dynamic changes in key parameters can be captured.

[0073] In some embodiments, in step S2, the LSTM is a Long Short-Term Memory network, and the attention layer is an attention mechanism model. The number of neurons in the LSTM network layer is set according to the data complexity, preferably in the range of 128 to 256. An attention layer is added after the LSTM layer to calculate the attention weights of each input parameter (ion source current, air intake, and historical data of current to the target beam).

[0074] In step S3, before training the model, the data in the training set, validation set, and test set are first converted into a time-series input format [number of samples, time step length, feature dimension]. The number of samples is determined according to the actual situation, preferably 32 or 64. The time step length is 10 to 20 acquisition times before the current time. The feature dimension refers to the two features included in each time step: ion source current and gas intake.

[0075] Then, the training parameters were set: the learning rate was set between 0.001 and 0.01, the Adam optimizer was used, the mean squared error (MSE) was used as the loss function, and the number of iterations was set between 500 and 1000. The collected data on ion source current, gas inlet volume, and target beam current at the same time were organized into a time series dataset in chronological order. Each sample contained parameter data of a certain time length, which were used as the model input to obtain the output of the LSTM network layer.

[0076] Then, the output of the LSTM network layer is input into the attention layer. The output of the LSTM network layer is converted into query vector Q, key vector K, and sum vector V respectively. Then, the similarity scores of Q and K are calculated according to the attention formula. After Softmax normalization, the attention weights are obtained. The weights are used to sum V to obtain the attention output.

[0077] The dimensions of the query vector Q, key vector K, and value vector V are [B, L]. Q ,d K ]、[B,L K ,d K ]、

[0078] [B,L K ,d V ], where B is the batch size, LQ L is the length of the query sequence K d is the length of the key sequence K d is the dimension of each query vector V d is the dimension of each value vector, the dimension of the attention output is consistent with d V .

[0079] where the similarity score of Q and K is calculated according to the attention formula, and the greater the value, the stronger the correlation:

[0080]

[0081] In the formula, K T is the transpose matrix of the key vector, and after transposition, the dimension of K becomes [B, d K , L K ], d K is the dimension of each query vector, and Q is the query vector.

[0082] where the attention weight is obtained by Softmax normalization, and the following formula is used:

[0083]

[0084] In the formula, Softmax is used to normalize the length of the key sequence corresponding to each query dimension. The sum of the weights of each query is 1, and the dimension of the weight matrix is consistent with the score matrix, which is [B, L Q , d K ].

[0085] where the weighted sum of V using the weight is carried out using the following formula:

[0086]

[0087] In the formula, V is the value vector, K T is the transpose matrix of the key vector, and after transposition, the dimension of K becomes [B, d K , L K ], d K is the dimension of each query vector, and Q is the query vector. The dimension of the output after weighted sum is [B, L Q , d V ]. Each query corresponds to a d V dimensional fusion feature, and the length is consistent with the query sequence L Q .

[0088] In some embodiments, in step S3, the training set, the validation set and the test set are used to train, validate and test the LSTM-Attention initial model, respectively, including the following steps:

[0089] S3.1, training the LSTM-Attention initial model with the training set to obtain a neutron beam current control model;

[0090] S3.2, verifying the neutron beam current control model with the verification set to obtain a verification result;

[0091] S3.3, adjusting the model parameters of the neutron beam current control model according to the verification result to obtain an optimal neutron beam current control model;

[0092] S3.4, testing the optimal neutron beam current control model with the test set;

[0093] If the test is qualified, a final neutron beam current control model is obtained, and step S4 is continued to be executed;

[0094] If the test is unqualified, steps S3.1 to S3.4 are re-executed.

[0095] In some embodiments, in step S2, the LSTM network layer includes a plurality of LSTM layers stacked in series. The use of a series design of multiple-layer LSTM significantly improves the prediction accuracy and control robustness of the model for neutron source beam current fluctuations through hierarchical feature learning, enhanced generalization ability and support for complex time series modeling, providing key technical support for achieving a target beam current stability within ±5%.

[0096] In some embodiments, before the data set is sorted by time sequence in step S1, the data in the data set is first preprocessed, including the following steps:

[0097] S1.1, data cleaning: calculating the standard deviation and mean value of the data set corresponding to the same parameter, and calculating the ratio of each data value in the data set to the mean value and the standard deviation;

[0098] If the absolute value of the ratio is greater than or equal to 3, it is considered as an abnormal value and is removed;

[0099] If the absolute value of the ratio is less than 3, step S1.2 is continued to be executed;

[0100] S1.2, normalizing the data cleaned data to obtain a preprocessed data set, sorting the preprocessed data set by time sequence, dividing the preprocessed data set into a training set, a verification set and a test set, and then continuing to execute step S2.

[0101] In some embodiments, in step S1.2, the normalization processing adopts a min-max normalization method to normalize each parameter data to the interval [0, 1], and the formula is where x is the current data to be processed, xmin and x max are minimum and maximum values of the respective data values in the data set, respectively.

[0102] In some embodiments, in step S3.2, the verification result refers to:

[0103] First, the ion source current and the amount of gas input in the verification set are input into the neutron beam current control model to obtain a target beam current array;

[0104] Then, the evaluation index is used to evaluate the model to obtain a verification result;

[0105] Specifically, the evaluation index refers to one or a combination of the mean square error, the mean absolute error, and the determination coefficient.

[0106] In step S3.3, according to the verification result, adjusting the model parameters of the neutron beam current control model refers to presetting a threshold of the evaluation index and comparing it with the verification result;

[0107] If the verification result is within the threshold, the model parameters do not need to be adjusted, and the current model is taken as the optimal neutron beam current control model;

[0108] If the verification result is outside the threshold range, the model parameters are adjusted until the verification result is within the threshold.

[0109] Specifically, when the evaluation index is one or a combination of the mean square error, the mean absolute error, and the determination coefficient, the corresponding threshold is: the mean square error is less than or equal to 0.01, the mean absolute error is less than or equal to 0.01, and the determination coefficient is greater than or equal to 0.99.

[0110] As a preferred, in step S3.3, if the verification result is within the threshold, a plurality of neutron beam current control models corresponding to different parameter settings in different training stages are selected respectively, and the evaluation index is used to evaluate each neutron beam current control model to obtain a plurality of verification results, and the model with the smallest mean square error (MSE), the smallest mean absolute error (MAE), and the determination coefficient (R2) closest to 1 is selected as the optimal neutron beam current control model.

[0111] In some embodiments, in step S3.4, when the optimal neutron beam current control model is tested by using the test set, one or a combination of the mean square error (MSE), the mean absolute error (MAE), and the determination coefficient (R2) is used as the evaluation index. Specifically, the corresponding thresholds of the mean square error, the mean absolute error, and the determination coefficient are: the mean square error is less than or equal to 0.01, the mean absolute error is less than or equal to 0.01, and the determination coefficient is greater than or equal to 0.99.

[0112] In some embodiments, please refer toFigure 2 In step S4, the grading strategy refers to grading according to the deviation of the target beam current from the set value, first calculating the absolute value percentage of the deviation of the target beam current I from the set value I0

[0113]

[0114] Primary strategy: if the absolute value percentage of the deviation is ≤5% and the duration is >5s, output the ion source current adjustment amount;

[0115] Secondary strategy: if 5%≤the absolute value percentage of the deviation≤10% and the duration is >5s, output the ion source current adjustment amount and the gas inlet adjustment amount;

[0116] Tertiary strategy: if the absolute value percentage of the deviation is >10% and the duration is >15s, start safety protection and automatically shut down for verification.

[0117] Specifically, if the absolute value percentage of the deviation is ≤5% and the duration is >5s, the ion source current adjustment amount is output according to the following formula:

[0118]

[0119] In formula (1), ΔI src1 is the ion source current adjustment amount output in the primary strategy case, K1 is the gain coefficient, obtained through the final neutron beam current control model, U is the microwave power supply voltage, U is a fixed value, I0 is the target beam current set value, and I is the target beam current measured value. If the value of I0-I is less than 0, the ion source current is reduced, if the value of I0-I is greater than 0, the ion source current is increased, and if the value of I0-I is equal to 0, the ion source current is not adjusted.

[0120] Specifically, if 5%≤the absolute value percentage of the deviation≤10% and the duration is >5s, the ion source current adjustment amount ΔQ and the gas inlet adjustment amount ΔI src2 are output according to the following formula:

[0121] ΔQ=K2(I0-I)(2)

[0122]

[0123] In formula (2) and formula (4), ΔQ is the gas inlet adjustment amount, U is the microwave power supply voltage, U is a fixed value, K2, K4 are gain coefficients, K3 is a coupling compensation coefficient, and K2, K3 and K4 are obtained through the final neutron beam current control model. I0 is the target beam current set value, and I is the target beam current measured value. If the value of ΔQ is less than 0, the gas inlet amount is reduced, if the value of ΔQ is greater than 0, the gas inlet amount is increased, and if the value of ΔQ is equal to 0, the gas inlet amount is not adjusted. If the value of ΔI src2If the value of ΔI is less than 0, the ion source current is adjusted to be reduced, if the value of ΔI is greater than 0, the ion source current is adjusted to be increased, and if the value of ΔI is equal to 0, the ion source current is not adjusted. src2 If the value of ΔI is less than 0, the ion source current is adjusted to be reduced, if the value of ΔI is greater than 0, the ion source current is adjusted to be increased, and if the value of ΔI is equal to 0, the ion source current is not adjusted. src2 If the value of ΔI is less than 0, the ion source current is adjusted to be reduced, if the value of ΔI is greater than 0, the ion source current is adjusted to be increased, and if the value of ΔI is equal to 0, the ion source current is not adjusted.

[0124] Specifically, the microwave power supply voltage is selected according to actual conditions.

[0125] In some embodiments, a dual-parameter-based neutron beam current real-time regulation system is also provided for implementing the dual-parameter-based neutron beam current real-time regulation method as described above, comprising,

[0126] A data acquisition module is configured to acquire sensor data.

[0127] A data processing module is configured to store and run a final neutron beam current control model and output corresponding ion source current adjustment amount and / or gas inlet amount adjustment amount, or output a shutdown instruction.

[0128] An execution module is configured to adjust the ion source current and / or the gas inlet amount according to the output ion source current adjustment amount and / or gas inlet amount adjustment amount, or execute the output shutdown instruction.

[0129] Specifically, the data acquisition module comprises an ion source current sensor, a mass flow meter, and a neutron source high-voltage power supply current sensor. The gas mass flow meter has a range of 0-6 sccm, a response time of <100 ms, and can output a 4-20 mA standard signal corresponding to the gas inlet amount. The neutron source high-voltage power supply current sensor is integrated in the neutron source high-voltage power supply and can monitor the target current in real time. The data acquisition card synchronously acquires three signals of the ion source current sensor, the gas mass flow meter, and the neutron source high-voltage power supply current sensor. The data acquisition card has a 16-bit resolution and a sampling frequency of 1 kHz, and can realize data transmission in a wired or wireless manner. The acquisition frequency of the data acquisition card is set according to the running characteristics of the neutron source to ensure that the dynamic changes of the key parameters can be captured.

[0130] Specifically, the data acquisition module comprises a data acquisition card configured to collect sensing data of each sensor.

[0131] Specifically, the data processing module can also store real-time data and historical logs.

[0132] Specifically, the execution module adopts an FPGA auxiliary control unit configured to process the adjustment instruction output by the data processing module and generate a PWM signal to control the execution mechanism, so as to ensure that the adjustment response time is <10 ms.

[0133] Specifically, the intake amount adjustment in the execution module adopts a mass flow meter, preferably with a range of 0-6sccm and an adjustment accuracy of 0.01sccm.

[0134] In some embodiments, an electronic device is also provided, characterized in that comprising a processor, a memory, the processor being coupled with the memory;

[0135] The memory is configured to store a computer program, and the processor is configured to execute the computer program stored in the memory to implement the dual-parameter-based neutron beam current real-time regulation method as described above.

[0136] In some embodiments, a computer readable storage medium is also provided, configured to store a computer program, the computer program being executed by a processor to implement the dual-parameter-based neutron beam current real-time regulation method as described above.

[0137] In the description of the present disclosure, it should be understood that the terms "center", "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the purpose of facilitating the description of the present disclosure and simplifying the description, and do not indicate or imply that the device or element referred to has a particular orientation, is constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present disclosure.

[0138] Unless otherwise required by context, the term "comprises" in the specification and claims is to be construed as open, inclusive, meaning "including, but not limited to". In the description of the specification, the terms "one embodiment", "some embodiments", "exemplary embodiment", "exemplarily" or "some examples" and the like are intended to mean that a particular feature, structure, material or characteristic included in at least one embodiment or example of the present disclosure. The illustrative representation of the above terms does not necessarily refer to the same embodiment or example. In addition, the specific features, structures, materials or characteristics described can be included in any one or more embodiments or examples in any appropriate manner.

[0139] The terms "first", "second" are only used to distinguish different description objects, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features, i.e. the position, order, priority, number or content of the described objects. Therefore, the features limited by "first", "second" can explicitly or implicitly include one or more features. In the description of the embodiments of the present disclosure, unless otherwise stated, the meaning of "multiple" is two or more.

[0140] In describing some embodiments, the use of "connection" and / or "coupled to" can be used. For example, the term "connected" can be used to mean that two or more elements are in direct physical or electrical contact with each other. The embodiments disclosed herein are not necessarily limited in this regard.

[0141] "A, B, and C at least one of" is synonymous with "at least one of A, B, or C," and includes the following

[0142] combinations of A, B, and C: A alone, B alone, C alone, A and B together, A and C together, B and C together, and A, B, and C together. In describing embodiments of the present application, the use of "or" means "and / or" unless stated otherwise, e.g., a condition A or B is satisfied by any one of A or B. The use of "and" is not intended to mean "and / or" unless specifically stated otherwise.

[0143] The use of "configured to" herein is meant "open and inclusive," in that a device or pieces of equipment that are configured to perform additional tasks or steps are still within the scope of the embodiments.

[0144] Additionally, the use of "based on" is meant to be open and inclusive, in that a process, step, calculation, or other action "based on" one or more recited conditions or values may, in practice, be based on additional conditions or values beyond those that are recited.

[0145] As used herein, "about," "approximately" include the recited value and the average value within an acceptable range of deviation from the particular value, as determined by one of ordinary skill in the art considering the measurement in question and the error in measuring the particular quantity (i.e., the limitations of the measurement system).

[0146] Exemplary embodiments are described herein with reference to cross-sectional and / or plan view illustrations that are schematic illustrations of idealized embodiments. In this regard, the thicknesses of the layers and regions can be exaggerated in the drawings for clarity. Thus, the exemplary embodiments should not be construed as limited to the shapes of regions as illustrated in the drawings, which are schematic representations. For example, the etched regions illustrated in the drawings typically have a curved or rounded shape. In addition, the exemplary embodiments are not limited to linear or planar shapes for the regions illustrated in the drawings, which are schematic representations. The exemplary embodiments can have a variety of shapes and are therefore not limited to the shapes illustrated in the drawings.

[0147] The above merely describes specific embodiments of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can think of changes or replacements within the technical range disclosed by the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A real-time neutron beam control method based on two parameters for controlling a compact neutron source, characterized in that, Includes the following steps: S1. Real-time acquisition of ion source current, gas inlet volume and target beam current to obtain a dataset, sort the dataset according to time series and divide it into training set, validation set and test set; S2. Construct an LSTM network layer and add an attention layer after the LSTM network layer to obtain the initial LSTM-Attention model; S3. The initial LSTM-Attention model is trained, validated, and tested using the training set, validation set, and test set, respectively, to obtain the final neutron beam control model. S4. Real-time monitoring of the deviation between the target beam current and the set value, and inversion using the final neutron beam control model, outputting the corresponding ion source current adjustment amount and / or gas inlet adjustment amount according to the hierarchical strategy, or outputting a shutdown command. S5. Adjust the ion source current and / or gas intake according to the output ion source current adjustment amount and / or gas intake adjustment amount, or execute the shutdown command; repeat steps S4 and S5 to achieve real-time control of the neutron beam.

2. The real-time neutron beam control method based on two parameters according to claim 1, characterized in that, Step S3 involves training, validating, and testing the initial LSTM-Attention model using the training set, validation set, and test set, respectively, including the following steps: S3.

1. The initial LSTM-Attention model is trained using the training set to obtain the neutron beam control model; S3.

2. The neutron beam control model is validated using a validation set to obtain validation results; S3.3 Adjust the model parameters of the neutron beam control model according to the verification results to obtain the optimal neutron beam control model; S3.

4. Test the optimal neutron beam control model using a test set; If the test is successful, the final neutron beam control model is obtained, and step S4 is executed. If the test fails, return to and re-execute steps S3.1 to S3.

4.

3. The real-time neutron beam control method based on two parameters according to claim 1, characterized in that, In step S2, the construction of the LSTM network layer includes multiple LSTM layers stacked in series.

4. The real-time neutron beam control method based on two parameters according to claim 1, characterized in that, In step S1, before sorting the dataset by time series, the data in the dataset is first preprocessed, including the following steps: S1.1 Data cleaning: Calculate the standard deviation and mean of the dataset corresponding to the same parameter, subtract each data value in the dataset from the mean, and calculate the ratio of each difference to the standard deviation; If the absolute value of the ratio is greater than or equal to 3, it is considered an outlier and is removed. If the absolute value of the ratio is less than 3, then continue to step S1.2; S1.

2. Normalize the cleaned data to obtain a preprocessed dataset. Sort the preprocessed dataset by time series and divide it into training set, validation set and test set. Then continue to execute step S2.

5. The real-time neutron beam control method based on two parameters according to claim 2, characterized in that, In step S3.2, the verification result refers to: First, the ion source current and gas inlet volume from the validation set are input into the neutron beam control model to obtain the target beam current array; Then, the model is validated and evaluated using evaluation metrics to obtain validation results; In step S3.3, adjusting the model parameters of the neutron beam control model according to the verification results means pre-setting the threshold of the evaluation index and comparing it with the verification results. If the verification results are within the threshold, there is no need to adjust the model parameters, and the current model will be used as the optimal neutron beam control model. If the verification result is outside the threshold range, adjust the model parameters until the verification result is within the threshold range.

6. The real-time neutron beam control method based on two parameters according to claim 2, characterized in that, In step S3.4, when testing the optimal neutron beam control model using the test set, one or more of the following are used as evaluation indicators: mean square error, mean absolute error, and coefficient of determination.

7. The real-time neutron beam control method based on two parameters according to claim 1, characterized in that, In step S4, the grading strategy refers to grading according to the deviation between the target beam current and the set value. First, the absolute percentage of the deviation between the target beam current and the set value is calculated. If the absolute value of the deviation is ≤5% and the duration is >5s, adjust the output ion source current. If 5% ≤ absolute deviation percentage ≤ 10%, and the duration > 5s, adjust the output ion source current and intake air volume. If the absolute percentage of the deviation is greater than 10% and the duration is greater than 15 seconds, the safety protection will be activated, and the machine will automatically stop for verification.

8. A real-time neutron beam control system based on two parameters, used to implement the real-time neutron beam control method based on two parameters as described in any one of claims 1 to 7, characterized in that, include, The data acquisition module is used to collect sensor data; The data processing module is used to store and run the final neutron beam control model, and output the corresponding ion source current adjustment and / or gas inlet adjustment, or output a shutdown command. The execution module is used to adjust the ion source current and / or the air intake according to the output ion source current adjustment amount and / or air intake adjustment amount, or to execute the output shutdown command.

9. An electronic device, characterized in that, Includes a processor and a memory, wherein the processor is coupled to the memory; The memory is used to store computer programs, and the processor is used to execute the computer programs stored in the memory to implement the real-time neutron beam control method based on two parameters as described in any one of claims 1 to 7.

10. A computer-readable storage medium for storing a computer program, characterized in that, The computer program is executed by a processor to implement the real-time neutron beam control method based on two parameters as described in any one of claims 1 to 7.

Images