A method and system for optimizing a communication protocol for a neutron generator

Abstract

The present application relates to the technical field of communication, in particular to a kind of neutron generator communication protocol optimization method and system.The method includes: obtaining neutron generator operating parameter and rate of change to construct composite state vector;Each parameter abnormal score is calculated and weighted sum is obtained to obtain comprehensive risk index;According to the index, dynamically divide core and non-core parameter, and adjust the update cycle of non-core parameter;For the core parameter of high correlation, only the complete data of a reference parameter and the difference value or prediction residual of the remaining parameters are transmitted, and the communication data frame is constructed.The scheme of the present application can dynamically evaluate risk, intelligently adjust data transmission strategy, improve early warning reliability while significantly optimizing communication efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of communication. More particularly, the present application relates to a neutron generator communication protocol optimization method and system. BACKGROUND

[0002] In order to ensure the long-term stable and safe operation of the neutron generator, a remote monitoring system is usually deployed to collect and analyze the key operating parameters of the neutron generator in real time, such as the voltage and current of the high-voltage power supply, the ion source state, the vacuum degree of the vacuum chamber, and the neutron yield. In the existing remote monitoring technology, the communication protocol generally adopts a fixed cycle polling or data reporting mechanism, that is, the host computer or data center requests or receives the complete data set containing all monitoring parameters sent by the neutron generator controller at a preset and fixed time interval. However, this fixed cycle communication mechanism has inherent limitations: when the neutron generator is in stable operation, the fluctuations of its operating parameters are usually relatively gentle, at this time, high-frequency data transmission will generate a large amount of redundant information with similar content, which not only occupies valuable communication bandwidth resources, but also significantly increases the communication cost and causes network congestion in bandwidth-limited scenarios such as wireless communication or satellite communication.

[0003] In order to alleviate the above problems, some improvement methods are proposed in the prior art, for example, by using data compression and other technical means to process the data before transmission, in order to reduce the overall data transmission volume to some extent, thereby saving communication bandwidth.

[0004] However, the above-mentioned improvement method still has obvious deficiencies. First, most of these data compression methods are not deeply related to the actual operating state of the neutron generator, and the communication protocol cannot intelligently distinguish which parameters are core indicators reflecting the system health status or secondary auxiliary information according to the specific working conditions, resulting in uniform processing of all parameters without distinction. Secondly, the prior art also pays little attention to the internal correlation of the operating parameters in the time sequence. For example, in some specific working modes, the power supply voltage and current may show a high positive correlation, but the existing communication protocol usually encodes and transmits these related parameters independently, without effectively utilizing this inherent statistical property to further eliminate the redundant components in the data, thereby limiting the efficiency of data representation and transmission. SUMMARY

[0005] The purpose of the present application is to provide a neutron generator communication protocol optimization method and system to solve the problems of low communication efficiency and poor operation reliability in the prior art. To this end, the present application provides solutions in the following two aspects.

[0006] In the first aspect, the present application provides a neutron generator communication protocol optimization method, comprising the following steps:

[0007] The real-time operation parameters of the neutron generator, including voltage, current, vacuum degree and neutron yield, are acquired, a composite state vector is constructed based on the real-time operation parameters and the rate of change of the parameters within a preset time window, for each parameter in the composite state vector, a normalized deviation between an observation value and an expected baseline generated based on historical data and operating conditions is calculated to obtain an anomaly score of each parameter, all anomaly scores are weighted and summed to obtain a comprehensive risk index quantifying the current operation risk, an anomaly score threshold is dynamically acquired according to the comprehensive risk index, parameters with anomaly scores higher than the anomaly score threshold are divided into core parameters, and the rest are divided into non-core parameters, the update cycle ratio is determined based on the comprehensive risk index, and the update cycle of the non-core parameters is set to a corresponding multiple of the core parameter update cycle, a communication data frame is constructed according to the update cycle of each parameter, for the core parameters contained in the communication data frame, the time series correlation is calculated, if the correlation coefficient is higher than a preset correlation threshold, only the complete data of a reference parameter is transmitted, and the data of the remaining related parameters is converted into a differential value or a model prediction residual.

[0008] Preferably, the step of constructing a composite state vector comprises: constructing a composite state vector by the formula The real-time operation parameters are calculated The first-order rate of change within a time window T , wherein is the current parameter value, is the parameter value one second ago, is the preset time window; the real-time values of the voltage, current, vacuum degree and neutron yield and the first-order rate of change of the real-time values are concatenated into an 8-dimensional composite state vector in a preset order.

[0009] Preferably, the step of calculating the normalized deviation between the observation value and the expected baseline generated based on the historical data and the operating conditions to obtain the anomaly score of each parameter comprises: calculating the anomaly score by the Z-score standardization method , wherein the anomaly score represents the magnitude of the deviation, and the formula is:

[0010] , wherein is the parameter observation value, is the expected baseline, is the historical data standard deviation; the expected baseline is the arithmetic mean of the past 100 sampling points of the parameter under a specific operating condition, and the historical data standard deviation is the standard deviation of the 100 sampling points.

[0011] Preferably, the abnormal score of all is weighted and summed to obtain a comprehensive risk index quantifying the current operation risk, comprising: assigning fixed weights to the abnormal score of the 8-dimensional parameter in the composite state vector, wherein the real-time value of voltage weight is 0.2, the voltage change rate weight is 0.1, the real-time value of current weight is 0.2, the current change rate weight is 0.1, the real-time value of vacuum degree weight is 0.15, the vacuum degree change rate weight is 0.05, the real-time value of neutron yield weight is 0.15, and the neutron yield change rate weight is 0.05; the calculation formula of the comprehensive risk index is: , wherein is the weight of the i-th parameter, is the abnormal score of the i-th parameter.

[0012] Preferably, the step of dynamically obtaining the abnormal score threshold is realized by using the following segmented function:

[0013] , wherein, is the abnormal score threshold, is the comprehensive risk index.

[0014] Preferably, the step of determining the update cycle ratio is realized by using the following rule:

[0015] , wherein, is the update cycle ratio, is the comprehensive risk index, and the reference update cycle of the core parameter is 100 milliseconds.

[0016] Preferably, in the step of calculating the time series correlation, the Pearson correlation coefficient is used to pre-evaluate the time series correlation between each real-time operation parameter, and the parameters with a correlation coefficient higher than 0.95 are divided into a correlation parameter group.

[0017] Preferably, when multiple parameters in the same correlation parameter group are divided into core parameters, the parameter with the smallest measurement noise in the parameter group is selected as the reference parameter.

[0018] Preferably, the step of converting the data of the remaining correlation parameters into difference values or model prediction residuals comprises: for each parameter strongly correlated with the reference parameter , a linear regression model between the two is established based on historical data: ; the transmitted value is the model prediction residual , and the calculation method is: , wherein, is the current measurement value of the correlation parameter, ​​​a current measurement value of a reference parameter, and a pre-calculated regression coefficient.

[0019] In a second aspect, a neutron generator communication protocol optimization system comprises:

[0020] a processor; a memory storing computer instructions for neutron generator communication protocol optimization, which, when executed by the processor, causes the system to perform the above-mentioned neutron generator communication protocol optimization method.

[0021] The present application has the following beneficial effects: by constructing a composite state vector containing real-time values and change rates and quantifying it into a comprehensive risk index, more accurate and dynamic risk assessment of the operating state of the neutron generator can be performed, overcoming the blindness of the traditional fixed cycle communication method. Based on the comprehensive risk index, the core parameters that can best reflect the health status and the secondary non-core parameters can be intelligently distinguished, and the update frequency of the non-core parameters during stable operation of the system can be adaptively and substantially reduced, thereby reducing unnecessary redundant data transmission under the premise of ensuring monitoring accuracy. At the same time, for the core parameters that need high-frequency transmission, the present application uses the inherent time series correlation to only transmit the complete data of a reference parameter and the prediction residuals of the remaining parameters, thereby achieving deep compression of data load while ensuring the accuracy of critical information, ultimately achieving the optimization of communication efficiency and reduction of operating cost while improving the sensitivity and reliability of fault early warning, and improving the communication efficiency and operating reliability. BRIEF DESCRIPTION OF DRAWINGS

[0022] Figure 1 A step flowchart of the neutron generator communication protocol optimization method in the present embodiment is schematically shown;

[0023] Figure 2 A structural block diagram of the neutron generator communication protocol optimization system in the present embodiment is schematically shown. DETAILED DESCRIPTION

[0024] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application.

[0025] As shown in the drawings, Figure 1 a neutron generator communication protocol optimization method in the present embodiment comprises the following steps:

[0026] Step S1, at least real-time operating parameters including voltage, current, vacuum degree and neutron yield of the neutron generator are obtained, and a composite state vector is constructed based on the real-time operating parameters and the change rates of the parameters within a preset time window.

[0027] The data acquisition module acquires raw data such as high-voltage power output voltage, filament current, target current, vacuum gauge reading, and neutron detector count from various sensors connected to the neutron generator controller at a fixed high-frequency sampling rate. At the same time, a sliding time window is obtained, for example, containing the last 50 sampling points. For each parameter, the first-order difference is calculated using the data in the window or a straight line is fitted by the least squares method to obtain the rate of change. All parameter observations at the current time and the calculated rate of change are concatenated to form a higher-dimensional composite state vector, for example, S = [voltage value, current value,..., voltage rate of change, current rate of change,...].

[0028] In an optional embodiment, the step of constructing the composite state vector comprises:

[0029] The real-time operating parameters are calculated by the formula The first-order rate of change in the time window T Wherein is the current parameter value, is the parameter value 5 seconds ago, is the preset time window; the real-time values of the voltage, current, vacuum, and neutron yield and the first-order rate of change of the real-time values are concatenated into an 8-dimensional composite state vector in a predetermined order. Preferably, the time window is fixed at 5 seconds for calculating the rate of change of each parameter. For example, the voltage value collected at the current time is 100 kV, and the voltage value recorded 5 seconds ago is 98 kV, so the first-order rate of change of the voltage is 0.4 kV / s. The above calculation method is also applied to the current, vacuum, and neutron yield, and their respective first-order rates of change are obtained.

[0030] Step S2, for each parameter in the composite state vector, calculate the normalized deviation between the observation value and the expected baseline generated based on historical data and operating conditions to obtain the abnormal score of each parameter, and perform weighted summation on all abnormal scores to obtain a comprehensive risk index quantifying the current operating risk.

[0031] Step S2, for each parameter in the composite state vector, calculate the normalized deviation between the observation value and the expected baseline generated based on historical data and operating conditions to obtain the abnormal score of each parameter, and perform weighted summation on all abnormal scores to obtain a comprehensive risk index quantifying the current operating risk.

[0032] ​Specifically, based on the historical database accumulated by the neutron generator in different stable working conditions for a long time, one or more prediction models, such as an autoregressive moving average model ARMA, are trained for each working condition. In runtime, the corresponding model is selected according to the current working condition, and the model generates the expected value of each parameter at the next time using the historical data and the current input. The expected value is the expected baseline. In the normalization process, the real-time observation value of the sensor is subtracted from the expected baseline, and then divided by the standard deviation of the parameter in the historical normal database, to complete the normalization, that is, to obtain the dimensionless deviation value. The absolute value of the deviation value is taken as the abnormal score of the parameter. According to the expert experience, a weight coefficient is preset for each parameter abnormal score, for example, the weight of the neutron yield is set to 0.4, and the weight of the high-voltage voltage is set to 0.3. After multiplying all the abnormal scores of the parameters by the corresponding weights and summing them up, the comprehensive risk index R can be obtained.

[0033] In an optional embodiment, the normalized deviation between the observation value and the expected baseline generated based on the historical data and the running working condition is calculated to obtain the abnormal score of each parameter, including:

[0034] The Z-score standardization method is used to calculate the abnormal score , which represents the magnitude of the deviation, and the formula is: , wherein, is the parameter observation value, is the expected baseline, is the historical data standard deviation;

[0035] The expected baseline is the arithmetic mean of the past 100 sampling points of the parameter under a specific running working condition, and the historical data standard deviation is the standard deviation of the 100 sampling points.

[0036] Exemplarily, for a specific stable running working condition, the device is stably running at 90kV voltage, and a historical data buffer containing the last 100 sampling points is continuously maintained. For the voltage parameter, the arithmetic mean of the 100 data points is calculated to obtain the expected baseline, and the standard deviation of the 100 points is calculated. The expected baseline and the standard deviation define the normal fluctuation range of the voltage under the working condition. When a new voltage observation value comes, the abnormal score of the voltage observation value is calculated. The above process of calculating the abnormal score is repeated for all 8 dimensions in the composite state vector, and an independent abnormal score is calculated for each dimension to measure the abnormality of each dimension.

[0037] In an optional embodiment, all abnormal scores are weighted and summed to obtain a comprehensive risk index that quantifies the current running risk, including:

[0038] Fixed weights are assigned to the anomaly scores of the parameters in the eight dimensions of the composite state vector, where the weight of the real-time voltage value is 0.2, the weight of the voltage change rate is 0.1, the weight of the real-time current value is 0.2, the weight of the current change rate is 0.1, the weight of the real-time vacuum degree value is 0.15, the weight of the vacuum degree change rate is 0.05, the weight of the real-time neutron yield value is 0.15, and the weight of the neutron yield change rate is 0.05.

[0039] Overall Risk Index The calculation formula is:

[0040] .

[0041] in For the first The weights of each parameter, For the first The abnormal score of each parameter.

[0042] An overall risk score is calculated by assigning different importance weights to the anomaly scores of different parameters. The weights reflect the degree of influence of different parameters on stability; for example, real-time values ​​of voltage and current are considered the most critical and therefore have the highest weight of 0.2, while their rates of change are second, with a weight of 0.1. In contrast, the rates of change of vacuum degree and neutron yield contribute relatively little to the risk, with a weight of 0.05.

[0043] Step S3: Dynamically obtain the abnormal score threshold based on the comprehensive risk index, classify parameters with abnormal scores higher than the abnormal score threshold as core parameters, and classify the rest as non-core parameters; determine the update cycle multiple based on the comprehensive risk index, and set the update cycle of non-core parameters as the corresponding multiple of the update cycle of core parameters.

[0044] Specifically, the step of dynamically obtaining the anomaly score threshold is implemented using the following piecewise function:

[0045] .

[0046] in, The threshold for abnormal scores. This is a comprehensive risk index.

[0047] The parameters to be considered core parameters are determined based on the current overall risk level. The thresholds are obtained in stages; the higher the risk, the lower the threshold, which means that the parameters will be more sensitive to minor anomalies.

[0048] For example, when the operation is very stable and the calculated comprehensive risk index is 0.4, which is less than 0.5, a higher threshold of 1.5 is selected, and only those parameters with abnormal scores exceeding 1.5 are identified as core parameters that need to be given priority attention. If some fluctuations occur, the comprehensive risk index rises to 0.7, between 0.5 and 0.8, and the threshold is lowered to 1.0. If the risk increases and the comprehensive risk index reaches 1.715, which is greater than 0.8, the most stringent threshold of 0.7 is adopted, and even parameters with low abnormality will be marked as core parameters, thereby ensuring the highest degree of monitoring.

[0049] The step of determining the update cycle rate is implemented using the following rules:

[0050] .

[0051] wherein, is the update cycle rate, is the comprehensive risk index, and the reference update cycle of the core parameters is 100 milliseconds.

[0052] When determining the update cycle rate, the data update frequency of non-core parameters is dynamically adjusted according to the risk level, which can balance the communication bandwidth and monitoring needs. The core parameters are always updated at the reference cycle of 100 milliseconds, while the update cycle of other non-core parameters is a multiple of this reference cycle, and the multiple is determined by the comprehensive risk index.

[0053] For example, when the operation is stable and the comprehensive risk index is 0.4, it is determined that the risk is low, and the update cycle rate is set to 10, so the actual update cycle of non-core parameters is 1 second, saving bandwidth. When the risk rises to 0.7, the rate is adjusted to 5, and the update cycle of non-core parameters is shortened to 500 milliseconds, at which time the monitoring frequency is increased. In the high-risk state where the comprehensive risk index is as high as 1.715, the rate will be reduced to the minimum value of 2, at which time the update cycle of non-core parameters will be 200 milliseconds, ensuring that even minor parameters can be updated more frequently in dangerous situations.

[0054] In step S4, a communication data frame is constructed according to the update cycle of each parameter, wherein for the core parameters contained in the communication data frame, the time series correlation is calculated, and if the correlation coefficient is higher than a preset correlation threshold, only the complete data of a reference parameter is transmitted, and the data of the remaining related parameters is converted into a differential value or a model prediction residual.

[0055] Specifically, when preparing to send a data frame in each cycle, all parameters divided into cores and non-core parameters that need to be updated in the current cycle are packaged. For all core parameters in the frame, the data sequence of the last time window, such as the past 1 minute, is extracted, the Pearson correlation coefficient between any two parameter sequences is calculated, and a high correlation threshold, such as 0.95, is obtained.

[0056] For example, if the correlation coefficient between the high-voltage power supply voltage and the target current is found to be 0.98, the voltage is selected as the reference parameter, and the complete floating-point value is written into the data frame. For the target current, instead of writing the complete value, a linear regression model is established based on the historical data of the two, the predicted current value is calculated using the current voltage value, and the residual error between the true value and the predicted value is calculated. The data frame only contains the complete value of the voltage and the residual error value of the current, and the receiving end can restore the true current value using the same regression model and the received voltage value. More specifically, a self-describing frame header is set in the communication data frame structure, the frame header includes a parameter existence mask, such as a bitmap, each bit corresponds to a parameter, and is used to indicate whether it is a core parameter and is included in the current frame, and an update cycle ratio field. Each frame of data encapsulates its own context information, and the receiving end can analyze which parameters are being transmitted and the expected update frequency of non-core parameters without any out-of-band channel, thereby analyzing the data.

[0057] In an optional embodiment, in the step of calculating the time series correlation, the Pearson correlation coefficient is used to evaluate the time series correlation between each real-time running parameter in advance, and the parameters with a correlation coefficient higher than 0.95 are divided into a correlation parameter group.

[0058] Preferably, when multiple parameters in the same correlation parameter group are divided into core parameters, the parameter with the smallest measurement noise in the parameter group is selected as the reference parameter.

[0059] Specifically, an offline data analysis is performed before formal operation to find the internal correlation between parameters to reduce data redundancy. Subsequently, a large amount of historical running data is used to calculate the Pearson correlation coefficient of any two parameters in the time series, such as the Pearson correlation coefficient of voltage and current in the time series. If the correlation coefficient of voltage and current is calculated to be 0.97, which is higher than the preset threshold of 0.95, then the voltage and current are classified into the same correlation parameter group.

[0060] During real-time monitoring, assuming that voltage and current are identified as core parameters based on anomaly scores and require high-frequency transmission, their complete data will not be transmitted simultaneously. Then, pre-evaluated parameter characteristics are queried, such as which parameter's sensor measurement noise is lower or historical data fluctuations are more stable. If historical data shows that the standard deviation of the voltage sensor is less than that of the current sensor, then voltage is selected as the benchmark parameter for this group, and only the complete real-time data of voltage will be sent in subsequent data transmissions.

[0061] In an optional embodiment, converting the data for the remaining relevant parameters into differencing values ​​or model prediction residuals includes:

[0062] For each reference parameter Strongly correlated parameters A linear regression model between the two is established based on historical data: ;

[0063] The transmitted values ​​are the model prediction residuals. The calculation method is as follows:

[0064] .

[0065] in, The current measured values ​​of the relevant parameters. The current measured value of the reference parameter. and These are the pre-calculated regression coefficients.

[0066] By utilizing a pre-established linear regression model between parameters, only the minute residuals between actual and model predictions are transmitted. This significantly reduces data load, improves communication efficiency in bandwidth-constrained environments, and lowers costs. The receiving end can also accurately and losslessly recover the original data, ensuring monitoring precision. Furthermore, the prediction residuals themselves serve as a sensitive system health indicator; abnormal fluctuations can act as early warning signals for faults. Thus, while optimizing communication, this also enhances system security and reliability.

[0067] This invention also provides a neutron generator communication protocol optimization system. For example... Figure 2 As shown, the system includes a processor and a memory, the memory storing computer program instructions, which, when executed by the processor, implement the neutron generator communication protocol optimization method described above according to the present invention.

[0068] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and therefore will not be described in detail here.

[0069] In this description, the term "application" also means any computer program product storing such a program for use with or in connection with a computer system, apparatus or device. The program can be stored on any apparatus-readable medium, for example, but not limited to, any volatile memory or non-volatile memory. In this description, the term "memory" also means any computer program product storing such a program for use with or in connection with a computer system, apparatus or device. The program can be stored on any apparatus-readable medium, for example, but not limited to, any volatile memory or non-volatile memory. In this description, the term "computer-readable medium" means any tangible medium that stores, communicates, or otherwise provides data that can be used by an instruction execution system, apparatus or device. The computer-readable medium can be any suitable magnetic storage medium or magneto-optical storage medium, such as, for example, resistive random access memory (RRAM), dynamic random access memory (DRAM), static random access memory (SRAM), enhanced dynamic random access memory (EDRAM), high-bandwidth memory (HBM), hybrid memory cube (HMC), and the like, or any other medium that can be used to store the desired information and that can be accessed by an application, module, or both. Any such computer storage media can be part of the device or accessible or connectable thereto. Any application or module described in this description can be implemented by computer-readable / executable instructions stored or otherwise held by such computer-readable media.

[0070] In the description of the present description, the meaning of "a plurality of" is at least two, for example, two, three or more, and the like, unless otherwise explicitly specified.

[0071] Although the present description has shown and described several embodiments of the present application, it will be apparent to those skilled in the art that many modifications, changes and substitutions can be made thereto without departing from the spirit and scope of the present application.

Claims

1. A method for optimizing a neutron generator communication protocol, characterized in that, The method comprises the following steps: acquiring real-time operation parameters of a neutron generator, including voltage, current, vacuum degree and neutron yield, constructing a composite state vector based on the real-time operation parameters and the change rate of the parameters within a preset time window; calculating the normalized deviation between the observation value and the expected baseline generated based on historical data and operation conditions for each parameter in the composite state vector to obtain an anomaly score for each parameter, and performing weighted summation on all anomaly scores to obtain a comprehensive risk index quantifying the current operation risk; dynamically acquiring an anomaly score threshold based on the comprehensive risk index, dividing parameters with anomaly scores higher than the anomaly score threshold into core parameters, and dividing the remaining parameters into non-core parameters; determining an update cycle multiplier based on the comprehensive risk index, and setting the update cycle of non-core parameters to a corresponding multiple of the update cycle of core parameters; constructing a communication data frame according to the update cycle of each parameter, wherein for core parameters contained in the communication data frame, the time series correlation is calculated, and if the correlation coefficient is higher than a preset correlation threshold, only the complete data of a reference parameter is transmitted, and the data of the remaining related parameters is converted into a differential value or a model prediction residual.

2. The method of claim 1, wherein, The step of constructing the composite state vector comprises: Through formula Calculate real-time operating parameters First-order rate of change within time window T ,in The current parameter value. for The parameter value from seconds ago, Preset time window; The real-time values of the voltage, current, vacuum degree and neutron yield and the first-order change rate corresponding to the real-time values are concatenated in a preset order to form an 8-dimensional composite state vector.

3. The method of claim 1, wherein, The step of calculating the normalized deviation between the observation value and the expected baseline generated based on historical data and operation conditions to obtain an anomaly score for each parameter comprises: An abnormality score is calculated using the Z-score standardization method , which represents the magnitude of deviation, and is given by the formula: ， wherein, is the parameter observation, is the expected baseline, is the historical data standard deviation; The desired baseline is the arithmetic mean of the past 100 sampling points of the parameter under the current operating condition, historical data standard deviation is the standard deviation of the 100 sampling points.

4. The method of claim 1, wherein, The step of performing weighted summation on all anomaly scores to obtain a comprehensive risk index quantifying the current operation risk comprises: The anomaly scores of the 8-dimensional parameters in the composite state vector are assigned fixed weights, wherein the weight of the real-time value of the voltage is 0.2, the weight of the change rate of the voltage is 0.1, the weight of the real-time value of the current is 0.2, the weight of the change rate of the current is 0.1, the weight of the real-time value of the vacuum degree is 0.15, the weight of the change rate of the vacuum degree is 0.05, the weight of the real-time value of the neutron yield is 0.15, and the weight of the change rate of the neutron yield is 0.

05. The formula for calculating the comprehensive risk index is: The formula for calculating the comprehensive risk index is: ， wherein is the weight of the th parameter, is the anomaly score of the th parameter.

5. The method of claim 1, wherein, The step of dynamically acquiring an anomaly score threshold is implemented using the following piecewise function: ， wherein, is an abnormal score threshold, is a composite risk index.

6. The method of claim 1, wherein, The step of determining an update cycle multiplier is implemented using the following rules: ， wherein, is the update cycle ratio, is the integrated risk index, and the reference update cycle of the core parameter is 100 milliseconds.

7. The method of claim 1, wherein, In the step of calculating the time series correlation, the Pearson correlation coefficient is used to pre-evaluate the time series correlation between each real-time operation parameter, and parameters with a correlation coefficient higher than 0.95 are divided into a related parameter group.

8. The method of claim 7, wherein, When multiple parameters in the same related parameter group are divided into core parameters, the parameter with the smallest measurement noise in the parameter group is selected as the reference parameter.

9. The method of claim 1, wherein, The step of converting the data of the remaining related parameters into a differential value or a model prediction residual comprises: For each parameter that is strongly correlated with the baseline parameter , a linear regression model is established between the two based on historical data: ;​ The transmitted values are model prediction residuals The calculation method is: ， wherein is a current measurement of a relevant parameter, is a current measurement of a reference parameter, and is a pre-computed regression coefficient.

10. A neutron generator communication protocol optimization system, comprising: The system comprises: a processor; a memory storing computer instructions for optimizing the communication protocol of a neutron generator, which, when executed by the processor, causes the system to perform the method for optimizing the communication protocol of a neutron generator according to any one of claims 1-9.

Images