Intelligent monitoring method and system of high-temperature superconducting reactor and storage medium

Abstract

The application discloses an intelligent monitoring method and system of a high-temperature superconducting reactor and a storage medium, relates to the field of high-temperature superconducting applications, and solves the technical problems that it is difficult to set an adaptive sensor layout according to the characteristics of different reactors and it is difficult to quickly identify the characteristic data of temperature and the coupling correlation characteristics of the 'temperature-pressure-liquid level' multiple parameters in the operation of the high-temperature superconducting reactor. The application sets the installation points of temperature sensors adaptively, obtains reference data based on the temperature sensors, synchronously matches the reference data to obtain the characteristic values of the reference data, divides time periods, performs trend prediction to obtain the predicted reference data of the next time period of the current time, obtains health factor one and health factor two based on the characteristic values of the reference data and the predicted reference data, and issues early warning information. The application can improve the data acquisition capability of the high-temperature superconducting reactor monitoring and improve the data analysis efficiency.

Description

Technical Field

[0001] This invention relates to the field of high-temperature superconducting applications, specifically to an intelligent monitoring method, system, and storage medium for high-temperature superconducting reactors. Background Technology

[0002] A high-temperature superconducting reactor is a reactor made using high-temperature superconducting materials. Its core structure includes a bushing, a Dewar liner, a Dewar, a superconducting magnet, and a boom. It features high efficiency, low energy consumption, and miniaturization. It has wide applications in power transmission, magnetic resonance imaging, and particle accelerators, significantly improving system performance and energy utilization efficiency.

[0003] Currently, most intelligent monitoring methods for high-temperature superconducting reactors struggle to adapt sensor layouts to the characteristics of different reactors. These methods often rely on manual experience and fail to consider the internal flow and temperature fields, leading to inadequate monitoring of critical areas and insufficient representativeness of the data. Furthermore, most intelligent monitoring methods for high-temperature superconducting reactors cannot quickly identify the characteristic temperature data during operation. They cannot achieve both accuracy and robustness through a single organized analysis during data acquisition; multiple iterative rounds are required, increasing computational complexity and time consumption.

[0004] Therefore, this invention discloses an intelligent monitoring method, system, and storage medium for high-temperature superconducting reactors to solve the above-mentioned technical problems. Summary of the Invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes an intelligent monitoring method, system and storage medium for high-temperature superconducting reactors, which is used to solve the technical problems of difficulty in setting adaptive sensor layout according to the characteristics of different reactors, and difficulty in quickly identifying the characteristic data of temperature and the coupling correlation characteristics of multiple parameters of "temperature-pressure-liquid level" during the operation of high-temperature superconducting reactors.

[0006] To achieve the above objectives, a first aspect of the present invention provides an intelligent monitoring method for high-temperature superconducting reactors, comprising:

[0007] The installation point of the temperature sensor is adaptively set, and reference data for the operation of the high-temperature superconducting reactor is obtained based on the temperature sensor and the preset monitoring device; the preset monitoring device includes a pressure transmitter and a capacitive level gauge; the reference data includes temperature, pressure and liquid level.

[0008] The feature values ​​of the reference data are obtained by synchronous matching of the reference data, and the high-temperature superconducting reactor is divided into time periods.

[0009] Trend prediction is performed based on the feature values ​​of the reference data to obtain the predicted reference data for the next time period of the current time; the feature values ​​of the reference data include feature temperature, feature pressure, and feature liquid level height;

[0010] Based on the feature values ​​of the reference data, the health factor 1 of the high-temperature superconducting reactor at the current time is obtained, and the health factor 2 of the high-temperature superconducting reactor in the next time period is obtained based on the predicted reference data; an early warning message is issued based on the health factor 1 and the health factor 2.

[0011] Preferably, the adaptive setting of the installation point of the temperature sensor includes:

[0012] Select the physical properties of solid and liquid materials, set the boundary conditions for convective and radiative heat transfer in actual use, as well as the Dewar inlet velocity and outlet pressure;

[0013] The gas-liquid two-phase flow was simulated using the VOF multiphase flow model in finite element simulation software. The temperature field was calculated using the energy equation, and turbulence was simulated using the K-epsilon viscosity equation. The temperature and flow fields at the Dewar and reactor frame were simulated and calculated. The physical properties included density, thermal conductivity, and viscosity.

[0014] Based on the temperature and flow fields at the Dewar and reactor frame, fiber optic sensors are arranged in large-area flow dead zones, and single-point temperature sensors are arranged in small-area flow dead zones. The temperature sensors include fiber optic sensors and single-point temperature sensors. Flow dead zones with a diameter of more than 1m are considered large-area flow dead zones, and flow dead zones with a diameter of less than 1m are considered small-area flow dead zones.

[0015] Preferably, the acquisition of reference data for the operation of the high-temperature superconducting reactor based on the temperature sensor and the preset monitoring device includes:

[0016] A1: Obtain the number of warnings YC during the current operation of the high-temperature superconducting reactor, and determine whether the number of warnings YC exceeds the warning number threshold YZ; if yes, set the current dynamic monitoring frequency DP of the high-temperature superconducting reactor to the minimum value of the standard frequency range; if no, proceed to A2; wherein, the standard frequency range is a frequency range manually set according to the safety monitoring accuracy of the high-temperature superconducting reactor; the warning number threshold YZ is obtained based on the service life of the high-temperature superconducting reactor, and can be set to 3 times the service life, and the upper limit of the warning number threshold YZ is 30;

[0017] A2: Based on the number of warnings YC, the current dynamic monitoring frequency DP of the high-temperature superconducting reactor is determined by formula (1);

[0018] The calculation formula (1) is:

[0019]

[0020] In the formula, BPZ is the maximum value of the standard frequency range, and BPX is the minimum value of the standard frequency range.

[0021] A3: Based on the dynamic monitoring frequency DP, the temperature inside the Dewar of the high-temperature superconducting reactor is obtained through a temperature sensor during operation; based on the dynamic monitoring frequency DP, the pressure inside the Dewar of the high-temperature superconducting reactor is obtained through a pressure transmitter during operation; based on the dynamic monitoring frequency DP, the liquid level inside the Dewar of the high-temperature superconducting reactor is obtained through a capacitive level gauge during operation.

[0022] Preferably, the step of synchronously matching the reference data to obtain the feature values ​​of the reference data includes:

[0023] The temperature sensors are sequentially labeled as target temperature sensors, and the temperature T monitored by the target temperature sensor in the last n minutes is obtained. i (t), with respect to temperature T i (t) Perform linear fitting to determine the temperature trend Use temperature T in sequence i (t) minus temperature trend The temperature residual r is obtained i (t), obtain several temperature residuals r i Standard deviation of (t) The confidence weight ω of the current target sensor is determined by calculating formula (2). i Where i is the target temperature sensor number, and the value of i is [1, d], where d is the maximum value of the target temperature sensor number; t is the time number monitored within the last n minutes;

[0024] Based on the temperature T most recently acquired by the target temperature sensor. i And credibility weight ω i The weighted median T is determined by formula (3). e Based on the weighted median T e Temperature T i And credibility weight ω i The weighted median absolute deviation (MAD) is determined by formula (4). ω Based on weighted median absolute deviation (MAD) ω The rejection threshold TY is determined by calculation formula (5);

[0025] Based on the rejection threshold TY for temperature T i Outlier removal yields the temperature retention set BT i The temperature retention set BT i Satisfy the calculation formula (6); retain the temperature set BT iT retained in i and the corresponding credibility weight ω i The characteristic temperature TW is obtained by weighted calculation;

[0026] Extract the reference pressure CL and reference liquid level height CG corresponding to the characteristic temperature TW from the pressure and liquid level reference library; calculate the characteristic pressure TP by weighting the reference pressure CL with the maximum and average values ​​of the most recently acquired pressure in the reference data; calculate the characteristic liquid level height TG by weighting the reference liquid level height CG with the minimum and average values ​​of the most recently acquired liquid level height in the reference data.

[0027] The calculation formula (2) is:

[0028]

[0029] In the formula, ∈ is a small constant to prevent division by zero, which is manually set;

[0030] The calculation formula (3) is:

[0031] T e =WeightedMedian(T1,T2,…,T) d ;ω1,ω2,…,ω d );

[0032] In the formula, T1 is the temperature most recently acquired by the target temperature sensor corresponding to i=1, and T2 is the temperature most recently acquired by the target temperature sensor corresponding to i=2. d Let i = d be the temperature most recently acquired by the target temperature sensor at the current time.

[0033] The calculation formula (4) is:

[0034] MAD ω =WeightedMedian(|T i -T e |;ω i );

[0035] The calculation formula (5) is:

[0036] TY = k·MAD ω ;

[0037] In the formula, k is determined based on the temperature fluctuation range within the time interval from 2n to n minutes from the current time, and the value range of k is [2,3].

[0038] The calculation formula (6) is:

[0039] BTi ={T i | |T i -T c |≤TY}.

[0040] Preferably, the time period division of the high-temperature superconducting reactor includes:

[0041] Extract the most recently acquired characteristic temperature TW of the high-temperature superconducting reactor at the first time point of the current time period. f And based on the calculation formula (7), the two adjacent characteristic temperatures TW are determined. f The rate of temperature change between WB f Obtain several temperature change rates WB f The average value of PWB;

[0042] Extract the standard duration BC of the time period, and determine the dynamic duration DC of the next time period based on the average value PWB and the standard duration BC using formula (8); add the last time point of the current time period to the dynamic duration DC to obtain the time range of the next time period;

[0043] The calculation formula (7) is:

[0044]

[0045] The calculation formula (8) is:

[0046]

[0047] In the formula, BWB is the standard rate of change set according to the usage time of the high-temperature superconducting reactor. The longer the usage time of the high-temperature superconducting reactor, the smaller the value of the standard rate of change, which can reduce the dynamic duration DC of the next time period and increase the analysis accuracy of the system.

[0048] Preferably, the step of obtaining the prediction reference data for the next time period based on the feature values ​​of the reference data includes:

[0049] Extract the time range of several time periods from the historical reference data, the characteristic temperature, characteristic pressure and characteristic liquid level height obtained each time within the time period, and the time range of the next time period, the characteristic temperature, characteristic pressure and characteristic liquid level height at each time point within the time period;

[0050] The system integrates several time ranges, characteristic temperatures, pressures, and liquid levels acquired at each time point within a given time period, the corresponding time range of the next time period, and the characteristic temperatures, pressures, and liquid levels at each time point within that time period into training and testing data. The training data is used to train the AI ​​model, and the testing data is used to test the trained AI model. The AI ​​model is then adjusted based on the testing results. The final result is a reference data prediction model that takes the current time range, the characteristic temperatures, pressures, and liquid levels acquired at each time point within that time period, and the time range of the next time period as input, and outputs the characteristic temperatures, pressures, and liquid levels at each time point within the next time period. The AI ​​model includes both a BP neural network model and an RBF neural network model.

[0051] By inputting the time range of the current time period, the characteristic temperature, characteristic pressure, and characteristic liquid level height acquired each time within the time period, and the time range of the next time period into the reference data prediction model, the characteristic temperature, characteristic pressure, and characteristic liquid level height at each time point in the next time period can be obtained.

[0052] Preferably, the process of obtaining a health factor one for the high-temperature superconducting reactor at the current time based on the feature values ​​of the reference data, and obtaining a health factor two for the high-temperature superconducting reactor in the next time period based on the predicted reference data, includes:

[0053] Real-time extraction of the characteristic temperature TW, characteristic pressure TP, and characteristic liquid level height TG of the high-temperature superconducting reactor at the current time; determination of whether there is one or more of the following: characteristic temperature TW greater than temperature threshold WY, or the absolute value of the difference between characteristic liquid level height TG and the standard value of liquid level height CTG greater than the standard deviation BTC, or characteristic pressure TP greater than the pressure threshold YY; if yes, mark the current time health factor KZ1 as 0; if no, determine the current time health factor KZ1 based on the calculation formula (9); where the temperature threshold, the standard value of liquid level height, the standard deviation, and the pressure threshold are all set based on experience;

[0054] Extract the characteristic temperature YTW of the high-temperature superconducting reactor at each time point in the next time period. h Characteristic pressure YTP h and characteristic liquid level height YTG h ; sequentially determine whether the characteristic temperature YTW exists at each time point in the next time period. h The temperature is greater than the threshold value WY, or the characteristic liquid level height YTG. h The absolute value of the difference between the liquid level height and the standard value (CTG) is greater than the standard deviation (BTC), or the characteristic pressure (YTP) is greater. h If one or more of the stress thresholds YY are greater than the threshold; then the corresponding health factor KZ2 at the specified time point will be considered. hMark as 0; otherwise, determine the health factor KZ2 at the corresponding time point based on formula (10). h ;

[0055] The calculation formula (9) is:

[0056]

[0057] In the formula, α1, α2, and α3 are proportional adjustment coefficients determined based on the number of times that characteristic temperature, characteristic pressure, and characteristic liquid level height in history caused the health factor to be marked as 0; β is an artificially set standard constant, generally taken as 100.

[0058] The calculation formula (10) is:

[0059]

[0060] Preferably, the specific interpretation of α1, α2, and α3, which are proportional adjustment coefficients determined based on the number of times characteristic temperatures, characteristic pressures, and characteristic liquid levels in history have resulted in a health factor being marked as 0, is as follows:

[0061] Obtain the following historical records of the current high-temperature superconducting reactor after it has been put into operation: C1, the number of times the characteristic temperature TW is greater than the temperature threshold WY; C2, the number of times the characteristic pressure TP is greater than the pressure threshold YY; and C3, the number of times the absolute value of the difference between the characteristic liquid level height TG and the standard value of the liquid level height CTG is greater than the standard deviation BTC.

[0062] Extract the initial values ​​CZ1 for α1, CZ2 for α2, and CZ3 for α3; where the initial value of α1 can be 0.45, the initial value of α2 can be 0.2, and the initial value of α3 can be 0.35.

[0063] The reference scaling factor CB is determined by formula (11) based on the number of iterations C1, C2, C3, initial values ​​CZ1, CZ2, and CZ3. y Where y takes the values ​​1, 2, and 3; when y = 1, CB y α1 represents the reference scaling factor, Cy represents the number of iterations C1, and αy represents the scaling factor α1; when y = 2, CB y α represents the reference scaling factor α2, Cy represents the number of iterations C2, and αy represents the scaling factor α2; when y = 3, CB y α represents the reference scaling factor α3, Cy represents the number of times C3, and αy represents the scaling factor α3.

[0064] Based on the reference scaling factor CB y The values ​​of the proportional adjustment coefficients α1, α2 and α3 are determined by calculation formula (12);

[0065] The calculation formula (11) is:

[0066]

[0067] The calculation formula (12) is:

[0068]

[0069] Preferably, the issuance of early warning information based on health factor one and health factor two includes:

[0070] Extract health factor KZ1. When health factor KZ1 is not greater than alarm threshold 1, perform an emergency shutdown and send an alarm message to the administrator requiring immediate maintenance. When health factor KZ1 is greater than alarm threshold 1 but less than alarm threshold 2, activate redundant cooling and send an alarm message to the administrator requiring maintenance. When health factor KZ1 is not less than alarm threshold 2, do not send an alarm message. Alarm threshold 1 and alarm threshold 2 are manually set, and alarm threshold 1 is less than alarm threshold 2.

[0071] Extract health factor 2 KZ2 at each time point in the next time period h When the health factor KZ1 at a given time point is not greater than alarm threshold 1, an alarm message is sent to the manager indicating that the high-temperature superconducting reactor has a fault at the corresponding time and needs to be repaired in advance; when the health factor KZ1 at a given time point is greater than alarm threshold 1 but less than alarm threshold 2, an alarm message is sent to the manager indicating that the high-temperature superconducting reactor may have a fault at the corresponding time and needs to reduce its operating frequency.

[0072] A second aspect of the present invention provides an intelligent monitoring system for a high-temperature superconducting reactor, comprising: an intelligent fusion module, and a data processing module and a status analysis module connected to the intelligent fusion module;

[0073] The data processing module adaptively sets the installation location of the temperature sensor and acquires reference data for the operation of the high-temperature superconducting reactor based on the temperature sensor and a preset monitoring device; wherein, the preset monitoring device includes a pressure transmitter and a capacitive level gauge; the reference data includes temperature, pressure and liquid level.

[0074] The intelligent fusion module: synchronously matches the reference data to obtain the feature values ​​of the reference data, divides the high-temperature superconducting reactor into time periods, and performs trend prediction based on the feature values ​​of the reference data to obtain the predicted reference data for the next time period of the current time; wherein, the feature values ​​of the reference data include feature temperature, feature pressure, and feature liquid level height;

[0075] The status analysis module: obtains the health factor one of the high-temperature superconducting reactor at the current time based on the feature value of the reference data, and obtains the health factor two of the high-temperature superconducting reactor in the next time period based on the predicted reference data; and issues early warning information based on health factor one and health factor two.

[0076] A third aspect of the present invention provides a storage medium, characterized in that it is used to store a computer program, which, when executed, implements an intelligent monitoring method for a high-temperature superconducting reactor.

[0077] Compared with the prior art, the beneficial effects of the present invention are:

[0078] 1. This invention adaptively sets the installation points of temperature sensors, acquires reference data on the operation of high-temperature superconducting reactors based on temperature sensors and preset monitoring devices; synchronously matches the reference data to obtain feature values, and divides the high-temperature superconducting reactor into time periods; predicts the trend based on the feature values ​​of the reference data to obtain predicted reference data for the next time period; obtains a health factor one for the high-temperature superconducting reactor at the current time based on the feature values ​​of the reference data, and obtains a health factor two for the high-temperature superconducting reactor in the next time period based on the predicted reference data; and issues early warning information based on health factor one and health factor two. This invention solves the technical problems in high-temperature superconducting reactor monitoring, such as the difficulty in setting adaptive sensor layouts according to the characteristics of different reactors, and the difficulty in quickly identifying the characteristic data of temperature and the coupling correlation characteristics of multiple parameters such as "temperature-pressure-liquid level" during the operation of high-temperature superconducting reactors. This invention improves the intelligence and timeliness of high-temperature superconducting reactor monitoring.

[0079] 2. In acquiring characteristic temperatures, this invention treats temperature data from different sensors as a multi-dimensional temperature vector. It extracts the characteristic temperature using a combination of robust statistics, local weight adjustment, and dynamic modeling of sensor reliability. This design achieves a good balance between data processing stability, accuracy, and computational efficiency. Specifically, traditional outlier removal methods typically rely on variance iteration testing, which involves repeatedly removing outliers after calculating the center value and recalculating until the variance meets a set threshold. This leads to multiple iterations, increasing computational load and latency. For high-temperature superconducting reactors under real-time monitoring, this can result in response lag or even affect the timeliness of anomaly detection. This invention, however, fully utilizes the advantage of robust statistical methods in maintaining stable center estimates when facing outliers. Combined with a weighting mechanism, the importance of different sensors is dynamically determined by their historical stability and current performance, thus simultaneously completing center estimation and outlier identification in a single calculation process. Attached Figure Description

[0080] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0081] Figure 1 This is a schematic diagram of the operation steps of the present invention;

[0082] Figure 2 This is a schematic diagram of the system modules of the present invention;

[0083] Figure 3 This is a simulation diagram of the Dewar inlet nozzle outlet of the present invention;

[0084] Figure 4 Simulation diagram of the Dewar inlet nozzle outlet extended by 15mm according to the present invention;

[0085] Figure 5 This is a model diagram of the internal skeleton structure of the reactor Dewar of the present invention;

[0086] Figure 6 This is a Dewar internal flow field diagram of the reactor frame of the present invention;

[0087] Figure 7 This is the temperature field of the Dewar reactor coil of the present invention;

[0088] Figure 8 This is a heat flux density distribution diagram of the coil surface of the present invention. Detailed Implementation

[0089] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0090] Please see Figure 1 The first aspect of this invention provides an intelligent monitoring method for high-temperature superconducting reactors, comprising:

[0091] The installation point of the temperature sensor is adaptively set, and reference data for the operation of the high-temperature superconducting reactor is obtained based on the temperature sensor and the preset monitoring device; the preset monitoring device includes a pressure transmitter and a capacitive level gauge; the reference data includes temperature, pressure and liquid level.

[0092] The feature values ​​of the reference data are obtained by synchronous matching of the reference data, and the high-temperature superconducting reactor is divided into time periods.

[0093] Trend prediction is performed based on the feature values ​​of the reference data to obtain the predicted reference data for the next time period of the current time; the feature values ​​of the reference data include feature temperature, feature pressure, and feature liquid level height;

[0094] Based on the feature values ​​of the reference data, the health factor 1 of the high-temperature superconducting reactor at the current time is obtained, and the health factor 2 of the high-temperature superconducting reactor in the next time period is obtained based on the predicted reference data; an early warning message is issued based on the health factor 1 and the health factor 2.

[0095] In this application, the installation point of the temperature sensor is adaptively set, including:

[0096] Select the physical properties of solid and liquid materials, set the boundary conditions for convective and radiative heat transfer in actual use, as well as the Dewar inlet velocity and outlet pressure;

[0097] The gas-liquid two-phase flow was simulated using the VOF multiphase flow model in finite element simulation software. The temperature field was calculated using the energy equation, and turbulence was simulated using the K-epsilon viscosity equation. The temperature and flow fields at the Dewar and reactor frame were simulated and calculated. The physical properties included density, thermal conductivity, and viscosity.

[0098] Based on the temperature and flow fields at the Dewar and reactor frame, fiber optic sensors are arranged in large-area flow dead zones, and single-point temperature sensors are arranged in small-area flow dead zones. The temperature sensors include fiber optic sensors and single-point temperature sensors. Flow dead zones with a diameter of more than 1m are considered large-area flow dead zones, and flow dead zones with a diameter of less than 1m are considered small-area flow dead zones.

[0099] It is worth noting that this invention constructs a flow-thermal coupling analysis model of the superconducting reactor using finite element simulation software. Based on the simulation results of the flow field and temperature field distribution inside the reactor's Dewar, fiber optic sensors are arranged in the large-area flow dead zone, reducing space occupancy. Single-point temperature sensors are arranged in the small-range flow dead zone. Fiber optic temperature measurement through fiber optic sensors ensures the stability of the temperature response. When used in conjunction with single-point temperature sensors, it not only ensures the adaptability of the testing system but also improves the system's monitoring accuracy. Furthermore, it effectively reduces the blind spots in system monitoring without affecting the flow of the cooling medium inside the Dewar.

[0100] It should be noted that the physical properties of the corresponding solid and liquid materials are obtained based on experimental measurements.

[0101] The boundary conditions for convective heat transfer and radiation heat transfer, the Dewar inlet velocity, and the outlet pressure can be as follows: The convective heat transfer boundary condition uses a convective heat transfer coefficient of 100 W / m. 2K, the radiation heat transfer boundary condition adopts a blackbody radiation model with an emissivity of 0.8, the inlet velocity is set to 0.5 m / s, and the outlet pressure is set to atmospheric pressure.

[0102] It should be noted that, because the outlet length of the Dewar inlet nozzle affects the stability of the jet, in this embodiment, the outlet of the Dewar inlet nozzle is extended by 15mm. Figure 3 and Figure 4 It can be seen that when the outlet of the Dewar inlet nozzle is extended by 15mm, the jet becomes smoother and splash-free, with the liquid nitrogen extending parallel forward along the wall, exhibiting the characteristics of "smooth jet, strong directionality, and long coverage"; among them, Figure 3 This is a simulation diagram of the nozzle outlet of the Dewar inlet. Figure 4 Simulation diagram showing the extension of the nozzle outlet of the Dewar inlet by 15mm.

[0103] It should be noted that, Figure 5 This is a model diagram of the internal skeleton structure of the reactor Dewar.

[0104] It should be noted that, through Figure 6 It can be seen that the streamline distribution inside the Dewar is uniform and orderly, with few velocity stagnation zones and flow dead zones, forming a wall-attached spiral flow. Liquid nitrogen circulation is good in the wall-attached spiral region, and the temperature can be maintained within the superconducting temperature range. Liquid nitrogen in the flow dead zone undergoes a phase change, leading to a temperature rise; therefore, placing a temperature sensor in this location yields significant benefits. Figure 6 This is a flow field diagram inside the Dewar shaped reactor frame.

[0105] It should be noted that, Figure 7 For the temperature field of the Dewar reactor coil, Figure 8 The heat flux density distribution on the coil surface is shown in the diagram. The temperature field of the Dewar reactor coil reveals good overall temperature uniformity, with no obvious localized high-temperature areas. Further observation of the heat flux density distribution on the coil surface reveals that the heat transfer capacity is stronger in the edge regions of each coil, while the heat transfer capacity is relatively weaker in the flat areas. This may be because the edge regions of the coil generate greater disturbance to the flow of liquid nitrogen, thus making the heat transfer effect in these areas more significant.

[0106] In this application, reference data for the operation of a high-temperature superconducting reactor is obtained based on a temperature sensor and a preset monitoring device, including:

[0107] A1: Obtain the number of warnings YC during the current operation of the high-temperature superconducting reactor, and determine whether the number of warnings YC exceeds the warning number threshold YZ; if yes, set the current dynamic monitoring frequency DP of the high-temperature superconducting reactor to the minimum value of the standard frequency range; if no, jump to A2; where the standard frequency range is the frequency range manually set according to the safety monitoring accuracy of the high-temperature superconducting reactor; the warning number threshold YZ is obtained based on the service life of the high-temperature superconducting reactor, and can be set to 3 times the service life, and the upper limit of the warning number threshold YZ is 30;

[0108] A2: Based on the number of warnings YC, the current dynamic monitoring frequency DP of the high-temperature superconducting reactor is determined by formula (1);

[0109] The calculation formula (1) is:

[0110]

[0111] In the formula, BPZ is the maximum value of the standard frequency range, and BPX is the minimum value of the standard frequency range.

[0112] A3: Based on the dynamic monitoring frequency DP, the temperature inside the Dewar of the high-temperature superconducting reactor is obtained through a temperature sensor during operation; based on the dynamic monitoring frequency DP, the pressure inside the Dewar of the high-temperature superconducting reactor is obtained through a pressure transmitter during operation; based on the dynamic monitoring frequency DP, the liquid level inside the Dewar of the high-temperature superconducting reactor is obtained through a capacitive level gauge during operation.

[0113] In this application, the feature values ​​of the reference data are obtained by synchronous matching of the reference data, including:

[0114] The temperature sensors are sequentially labeled as target temperature sensors, and the temperature T monitored by the target temperature sensor in the last n minutes is obtained. i (t), with respect to temperature T i (t) Perform linear fitting to determine the temperature trend Use temperature T in sequence i (t) minus temperature trend The temperature residual r is obtained i (t), obtain several temperature residuals r i Standard deviation of (t) The confidence weight ω of the current target sensor is determined by calculating formula (2). i Where i is the target temperature sensor number, and the value of i is [1, d], where d is the maximum value of the target temperature sensor number; t is the time number monitored within the last n minutes;

[0115] Based on the temperature T most recently acquired by the target temperature sensor. i And credibility weight ω iThe weighted median T is determined by formula (3). e Based on the weighted median T e Temperature T i And credibility weight ω i The weighted median absolute deviation (MAD) is determined by formula (4). ω Based on weighted median absolute deviation (MAD) ω The rejection threshold TY is determined by calculation formula (5);

[0116] Based on the rejection threshold TY for temperature T i Outlier removal yields the temperature retention set BT i The temperature retention set BT i Satisfy the calculation formula (6); retain the temperature set BT i T retained in i and the corresponding credibility weight ω i The characteristic temperature TW is obtained by weighted calculation;

[0117] Extract the reference pressure CL and reference liquid level height CG corresponding to the characteristic temperature TW from the pressure and liquid level reference library; calculate the characteristic pressure TP by weighting the reference pressure CL with the maximum and average values ​​of the most recently acquired pressure in the reference data; calculate the characteristic liquid level height TG by weighting the reference liquid level height CG with the minimum and average values ​​of the most recently acquired liquid level height in the reference data.

[0118] The calculation formula (2) is:

[0119]

[0120] In the formula, ∈ is a small constant to prevent division by zero, which is manually set;

[0121] The calculation formula (3) is:

[0122] T e =WeightedMedian(T1,T2,…,T) d ;ω1,ω2,…,ω d );

[0123] In the formula, T1 is the temperature most recently acquired by the target temperature sensor corresponding to i=1, and T2 is the temperature most recently acquired by the target temperature sensor corresponding to i=2. d Let i = d be the temperature most recently acquired by the target temperature sensor at the current time.

[0124] The calculation formula (4) is:

[0125] MADω =WeightedMedian(|T i -T e |;ω i );

[0126] The calculation formula (5) is:

[0127] TY = k·MAD ω ;

[0128] In the formula, k is determined based on the temperature fluctuation range within the time interval from 2n to n minutes from the current time, and the value range of k is [2,3].

[0129] The calculation formula (6) is:

[0130] BT i ={T i | |T i -T c |≤TY}.

[0131] It is worth noting that in this step, when acquiring the characteristic temperature, the temperature data from different sensors are treated as a multi-dimensional temperature vector. A combination of robust statistics, local weight adjustment, and dynamic modeling of sensor reliability is used to extract the characteristic temperature. This design achieves a good balance between data processing stability, accuracy, and computational efficiency. Specifically, traditional outlier removal methods typically rely on variance iterative testing, which involves repeatedly removing outliers after calculating the center value and recalculating until the variance meets a set threshold. This leads to multiple iterations, increasing computational load and latency. For high-temperature superconducting reactors under real-time monitoring, this can result in response lag or even affect the timeliness of anomaly detection. This step, however, fully utilizes the advantage of robust statistical methods in maintaining stable center estimates when facing outliers. Combined with a weighting mechanism, the importance of different sensors is dynamically determined by their historical stability and current performance, thus completing center estimation and outlier identification simultaneously in a single calculation process.

[0132] The specific process is as follows: First, a weighted center is established. This center value reflects the current temperature readings of each sensor and incorporates dynamic adjustments to sensor reliability, avoiding excessive influence of a single value or sensor on the overall result. Then, the weighted deviation is calculated, which combines the absolute deviation of each sensor from the center with its weight to obtain a targeted measure of dispersion. Based on the distribution of the weighted deviation, an adaptive rejection range is determined, removing outlier data that exceeds reasonable fluctuations. Finally, the weighted center is recalculated on the retained valid data to directly obtain the final characteristic temperature. In this way, only one organized analysis is needed in the characteristic temperature acquisition process to obtain a result that balances accuracy and robustness, eliminating the need for multiple rounds of iterative judgment and effectively reducing computational complexity and time consumption. The innovation of this method lies in replacing the traditional iterative variance test with a one-time adaptive processing, which makes the algorithm stand out in scenarios with high real-time requirements. At the same time, the dynamic weight mechanism ensures the dominant role of high-quality sensors while weakening the influence of low-quality or temporarily disturbed sensors. The introduction of robust statistics ensures that the results remain stable and reliable when outliers exist, thereby significantly improving the temperature feature extraction capability and continuous computing efficiency of the high-temperature superconducting reactor monitoring system under dynamic operating conditions.

[0133] It should be noted that obtaining the characteristic temperature in this step can be understood as:

[0134] A stability coefficient is calculated within a historical window using data from each sensor, serving as the initial weight for the sensor. In this step, the weights of different sensors can be dynamically updated over time to avoid assigning excessive influence to sensors that have previously exhibited unstable performance.

[0135] Then, by adjusting the weighted median based on the current sensor reliability and real-time deviation, the impact of outliers can be significantly reduced.

[0136] Finally, instead of relying on a fixed variance threshold, we adaptively define a removal threshold based on the deviation distribution of the current weighting center, and then calculate the characteristic temperature after determining the temperature retention set based on the removal threshold.

[0137] It should be noted that for temperature T i (t) Perform linear fitting to determine the temperature trend Use temperature T in sequence i (t) minus temperature trend The temperature residual r is obtained i The specific steps for (t) are as follows:

[0138] For example, the temperature T monitored by the temperature sensor numbered i in the last n minutes i (t) is:

[0139] (t1,T i (t1)),(t2,Ti (t2)),…,(t c ,T i (t c ));

[0140] Where c is the maximum value of the temperature numbers monitored within the last n minutes; (t1,…,t c The values ​​of each data item in the table are obtained by subtracting the time corresponding to t1 from the corresponding time.

[0141] Based on formula Determinant a i Where j takes values ​​in the range [1, c]; The monitoring time (t1,…,t) within the most recent n minutes c The average value of ) This is the average temperature measured over the last n minutes;

[0142] Based on formula Coefficient of determination b i ;

[0143] Get any time t within the last n minutes j Temperature trend for:

[0144]

[0145] Operating temperature T i (t) minus temperature trend The temperature residual r is obtained i (t) is:

[0146]

[0147] It should be noted that the calculation formula (3) is the weighted median T of the temperature most recently acquired by each target temperature sensor, calculated using the weighted median. e This allows us to find the most robust center value among several data points. In the calculation formula (3), the WeightedMedian() function sorts and accumulates the values ​​by weight until half of the total weight is reached, and then takes the value at that position.

[0148] It should be noted that the weighted median absolute deviation (MAD) ω Based on the weighted median T of temperature data e Its dispersion is more robust than the standard deviation, and it is almost unaffected by extreme outliers;

[0149] Weighted Median Absolute Deviation (MAD) ω The weight ω is introduced iThis allows historically stable sensors to have a greater impact in the discreteness calculation, while the impact of sensors with low weights is weakened in the deviation statistics, thus reducing their contamination of the deviation distribution.

[0150] It should be noted that the exclusion threshold TY is set to the weighted median absolute deviation (MAD). ω The value of k is increased to ensure that data exceeding the normal fluctuation range is removed. When the temperature fluctuation range increases within the time interval of 2n to n minutes from the current time, the value of k also increases, making the removal criteria more lenient and retaining more data. When the temperature fluctuation range decreases within the time interval of 2n to n minutes from the current time, the value of k also decreases, making the removal criteria more stringent and anomalies easier to remove.

[0151] It should be noted that the function of calculation formula (6) is to convert |T i -T c |≤TY corresponds to the temperature T i Keep it.

[0152] It should be noted that the temperature retention collection BT... i T retained in i and the corresponding credibility weight ω i The formula for obtaining the characteristic temperature TW through weighted calculation is as follows:

[0153]

[0154] It should be noted that the characteristic temperature can also be obtained through the following steps:

[0155] Each acquired temperature is integrated into a temperature group. The variance of the temperature group is obtained, and it is determined whether the variance is less than the variance threshold. If yes, the average value of the temperatures in the temperature group is calculated to obtain the characteristic temperature. If no, the temperature with the largest absolute difference from the average temperature is removed from the temperature group, and the variance is re-evaluated until the variance of the temperature group is less than the variance threshold. Then, the average value of the remaining data in the temperature group is calculated to obtain the characteristic temperature. The variance threshold is determined empirically. When removing the temperature with the largest absolute difference from the average temperature in the temperature group includes both the highest and lowest temperatures, the lowest temperature is removed first.

[0156] It should be noted that the reference pressure CL corresponding to the characteristic temperature TW in the pressure level reference library is calculated based on the formula CL = δ1·TW; the reference liquid level CG corresponding to the characteristic temperature TW in the pressure level reference library is calculated based on the formula CG = δ2·TW; where δ1 is the pressure adjustment factor obtained empirically for the pressure in the Dewar of the high-temperature superconducting reactor as a function of temperature; and δ2 is the liquid level adjustment factor obtained empirically for the liquid level height in the Dewar of the high-temperature superconducting reactor as a function of temperature.

[0157] It should be noted that in the characteristic pressure TP, which is calculated by weighting the reference pressure CL with the maximum and average values ​​of the most recently acquired pressure from the reference data, the weights are calculated manually based on the differences between the reference pressure CL and the corresponding maximum and average values ​​of the acquired pressure over several historical periods. For example, the weight of the reference pressure CL is 0.2, the weight of the maximum value of the most recently acquired pressure is 0.4, and the weight of the average value of the most recently acquired pressure is 0.4.

[0158] The feature liquid level height TG is obtained by weighting the reference liquid level height CG with the minimum and average liquid level heights most recently acquired from the reference data. The weights used in the weighting calculation are manually calculated based on the differences between the reference liquid level height CG and the corresponding minimum and average liquid level heights acquired over several historical periods. For example, the weight of the reference liquid level height CG is 0.3, the weight of the minimum liquid level height most recently acquired from the current time is 0.3, and the weight of the average liquid level height most recently acquired from the current time is 0.4.

[0159] This application divides the high-temperature superconducting reactor into time periods, including:

[0160] Extract the most recently acquired characteristic temperature TW of the high-temperature superconducting reactor at the first time point of the current time period. f And based on the calculation formula (7), the two adjacent characteristic temperatures TW are determined. f The rate of temperature change between WB f Obtain several temperature change rates WB f The average value of PWB;

[0161] Extract the standard duration BC of the time period, and determine the dynamic duration DC of the next time period based on the average value PWB and the standard duration BC using formula (8); add the last time point of the current time period to the dynamic duration DC to obtain the time range of the next time period;

[0162] The calculation formula (7) is:

[0163]

[0164] The calculation formula (8) is:

[0165]

[0166] In the formula, BWB is the standard rate of change set according to the usage time of the high-temperature superconducting reactor. The longer the usage time of the high-temperature superconducting reactor, the smaller the value of the standard rate of change, which can reduce the dynamic duration DC of the next time period and increase the analysis accuracy of the system.

[0167] It should be noted that when f = 1, the value of WB1 is 0.

[0168] In this application, trend prediction is performed based on the feature values ​​of reference data to obtain predictive reference data for the next time period from the current time, including:

[0169] Extract the time range of several time periods from the historical reference data, the characteristic temperature, characteristic pressure and characteristic liquid level height obtained each time within the time period, and the time range of the next time period, the characteristic temperature, characteristic pressure and characteristic liquid level height at each time point within the time period;

[0170] The system integrates several time ranges, characteristic temperatures, pressures, and liquid levels acquired at each time point within a given time period, the corresponding time range of the next time period, and the characteristic temperatures, pressures, and liquid levels at each time point within that time period into training and testing data. The training data is used to train the AI ​​model, and the testing data is used to test the trained AI model. The AI ​​model is then adjusted based on the testing results. The final result is a reference data prediction model that takes the current time range, the characteristic temperatures, pressures, and liquid levels acquired at each time point within that time period, and the time range of the next time period as input, and outputs the characteristic temperatures, pressures, and liquid levels at each time point within the next time period. The AI ​​model includes both a BP neural network model and an RBF neural network model.

[0171] By inputting the time range of the current time period, the characteristic temperature, characteristic pressure, and characteristic liquid level height acquired each time within the time period, and the time range of the next time period into the reference data prediction model, the characteristic temperature, characteristic pressure, and characteristic liquid level height at each time point in the next time period can be obtained.

[0172] It should be noted that the prediction of characteristic temperature, characteristic pressure, and characteristic liquid level height at each time point in the next time period in the reference data prediction model designed in this invention can be predicted by trends; for example: by analyzing the magnitude and variation of the characteristic temperature in the current time period, the characteristic temperature at each time point in the next time period can be obtained; by analyzing the magnitude and variation of the characteristic pressure in the current time period, the characteristic pressure at each time point in the next time period can be obtained; by analyzing the magnitude and variation of the characteristic liquid level height in the current time period, the characteristic liquid level height at each time point in the next time period can be obtained.

[0173] Specifically, the steps for testing the trained AI model using validation data and adjusting the AI ​​model based on the validation results are as follows:

[0174] The time range of the test data, the characteristic temperature, characteristic pressure, and characteristic liquid level height acquired each time within the time range, and the corresponding time range of the next time range are input into the trained artificial intelligence model to obtain the characteristic temperature, characteristic pressure, and characteristic liquid level height at each time point in the next time range. The characteristic temperature, characteristic pressure, and characteristic liquid level height at each time point in the next time range are compared with the characteristic temperature, characteristic pressure, and characteristic liquid level height at each time point in the test data. If the differences in the characteristic temperature, characteristic pressure, and characteristic liquid level height at each time point are all within the corresponding difference threshold (the threshold is obtained based on experience), then no parameter adjustment is required, and the next set of test data is tested. If the values ​​are not within the threshold, the corresponding parameters are adjusted until the differences in characteristic temperature, characteristic pressure, and characteristic liquid level at each time point of the corresponding test data are all within the corresponding difference threshold. Then, the next set of test data is tested. When the number of test data in which the differences in characteristic temperature, characteristic pressure, and characteristic liquid level at each time point of all test data are within the corresponding difference threshold accounts for 90% or more of the total test data, the following is obtained: the input is the time range of the current time period, the characteristic temperature, characteristic pressure, and characteristic liquid level obtained each time within the time period, and the time range of the next time period. The output is the reference data prediction model for the characteristic temperature, characteristic pressure, and characteristic liquid level at each time point in the next time period.

[0175] It should be noted that the differences in characteristic temperature, characteristic pressure, and characteristic liquid level at each time point are: the difference between the corresponding characteristic temperature at each time point and the corresponding characteristic temperature in the test data; the difference between the corresponding characteristic pressure at each time point and the corresponding characteristic pressure in the test data; and the difference between the corresponding characteristic liquid level at each time point and the corresponding characteristic liquid level in the test data. A single time point includes the differences in characteristic temperature, characteristic pressure, and characteristic liquid level.

[0176] It should be noted that, for the characteristic temperature, characteristic pressure and characteristic liquid level height at each time point in the next time period, the characteristic temperature, characteristic pressure and characteristic liquid level height at each time point are the characteristic temperature, characteristic pressure and characteristic liquid level height most recently acquired at the corresponding time point. If there are two times that are most recently acquired at the corresponding time point, the characteristic temperature, characteristic pressure and characteristic liquid level height acquired later will be used.

[0177] For example: if the next time period corresponding to several time periods in the historical reference data is from 15:00 to 15:10 on September 2, 2025, and one of the time points is 15:06:05 on September 2, 2025, and characteristic temperature, characteristic pressure, and characteristic liquid level height were obtained before and after this time point at 15:05:50 and 15:06:10 on September 2, 2025, then the characteristic temperature, characteristic pressure, and characteristic liquid level height at 15:06:10 on September 2, 2025 will be used as the characteristic temperature, characteristic pressure, and characteristic liquid level height at 15:06:05 on September 2, 2025.

[0178] If one of the time points is 15:06:00 on September 2, 2025, and characteristic temperature, characteristic pressure, and characteristic liquid level are obtained before and after this time point at 15:05:50 on September 2, 2025 and 15:06:10 on September 2, 2025, then the characteristic temperature, characteristic pressure, and characteristic liquid level at 15:06:10 on September 2, 2025 will be used as the characteristic temperature, characteristic pressure, and characteristic liquid level at 15:06:00 on September 2, 2025.

[0179] In this application, a health factor one for the high-temperature superconducting reactor at the current time is obtained based on the feature values ​​of reference data, and a health factor two for the high-temperature superconducting reactor in the next time period is obtained based on the predicted reference data, including:

[0180] Real-time extraction of the characteristic temperature TW, characteristic pressure TP, and characteristic liquid level height TG of the high-temperature superconducting reactor at the current time; determination of whether there is one or more of the following: characteristic temperature TW greater than temperature threshold WY, or the absolute value of the difference between characteristic liquid level height TG and the standard value of liquid level height CTG greater than the standard deviation BTC, or characteristic pressure TP greater than the pressure threshold YY; if yes, mark the current time health factor KZ1 as 0; if no, determine the current time health factor KZ1 based on the calculation formula (9); where the temperature threshold, the standard value of liquid level height, the standard deviation, and the pressure threshold are all set based on experience;

[0181] Extract the characteristic temperature YTW of the high-temperature superconducting reactor at each time point in the next time period. h Characteristic pressure YTP h and characteristic liquid level height YTG h ; sequentially determine whether the characteristic temperature YTW exists at each time point in the next time period. h The temperature is greater than the threshold value WY, or the characteristic liquid level height YTG. h The absolute value of the difference between the liquid level height and the standard value (CTG) is greater than the standard deviation (BTC), or the characteristic pressure (YTP) is greater. hIf one or more of the stress thresholds YY are greater than the threshold; then the corresponding health factor KZ2 at the specified time point will be considered. h Mark as 0; otherwise, determine the health factor KZ2 at the corresponding time point based on formula (10). h ;

[0182] The calculation formula (9) is:

[0183]

[0184] In the formula, α1, α2, and α3 are proportional adjustment coefficients determined based on the number of times that characteristic temperature, characteristic pressure, and characteristic liquid level height in history caused the health factor to be marked as 0; β is an artificially set standard constant, generally taken as 100.

[0185] The calculation formula (10) is:

[0186]

[0187] It is worth noting that, in the process of assessing and predicting the health status of high-temperature superconducting reactors, this invention can form a complete health factor system by performing real-time predictive analysis of key characteristic parameters for the current and future time periods. It not only provides a static assessment of the current operating status but also incorporates dynamic analysis of predictive reference data into its logical design, organically combining health factor one and health factor two, thereby achieving end-to-end health management from real-time monitoring to future trends.

[0188] During the current time period, the system accurately extracts characteristic temperature TW, characteristic pressure TP, and characteristic liquid level height TG. Combined with the temperature threshold WY, the standard value of liquid level height and its standard deviation BTC, and the pressure threshold YY parameters set by experience, a multi-condition anomaly identification mechanism is constructed. As long as any one or more indicators exceed the set range, the health factor KZ1 can be quickly set to 0 to ensure that high-risk states are responded to in real time. If all indicators are within the safe range, the formula (9) is used in combination with the proportional adjustment coefficients of α1, α2, and α3 and the standard constant β to calculate the health factor, so that the health factor has a delicate ability to be continuously quantified and avoids the information loss caused by simple binary judgment.

[0189] Meanwhile, in predicting health in future time periods, this method extracts the characteristic temperature YTW at each time point in the next time period. h Characteristic pressure YTP h and characteristic liquid level height YTG h For each future time point, a judgment is made individually to ensure the accuracy and temporal resolution of the prediction. The process for judging abnormal conditions is the same as for the current time; any situation exceeding the threshold will cause the corresponding health factor KZ2 to be affected. hSet to 0 to lock in future risks in advance; if the predicted data is normal, use formula (10) to calculate, so that health factor 2 and health factor 1 are structurally consistent, which facilitates horizontal comparison and trend analysis of the data. This design can form a two-way safety barrier in actual operation: on the one hand, real-time monitoring ensures the safe operation of the equipment at present; on the other hand, prediction and assessment provide maintenance personnel with the opportunity to intervene in advance, effectively reducing the probability of sudden failures and improving the safety level and controllability of the equipment throughout its entire life cycle.

[0190] Furthermore, the health factor, based on multi-parameter input and nonlinear calculation formulas, avoids misjudgments caused by fluctuations in a single parameter. By adjusting the proportional weight of the α coefficient, it can dynamically reflect the importance of different parameters in historical data, making the model closer to actual operational experience. The introduction of the standard constant β makes the numerical range of the calculation results more stable and interpretable, facilitating unified management across different systems. In addition, this method can be combined with cloud platforms or edge computing nodes during application to achieve one-click calculation and automatic recording of health factors, forming a closed loop between real-time monitoring and predictive analysis. This provides data-driven decision-making support for the operational safety of high-temperature superconducting reactors and can be further extended to the health management of other power equipment, realizing intelligent, predictive, and refined operation and maintenance models, thereby significantly improving grid stability and power supply reliability.

[0191] In this application, α1, α2, and α3 are proportional adjustment coefficients determined based on the number of times characteristic temperatures, characteristic pressures, and characteristic liquid levels in history resulted in a health factor being marked as 0. The specific interpretation of these coefficients is as follows:

[0192] Obtain the following historical records of the current high-temperature superconducting reactor after it has been put into operation: C1, the number of times the characteristic temperature TW is greater than the temperature threshold WY; C2, the number of times the characteristic pressure TP is greater than the pressure threshold YY; and C3, the number of times the absolute value of the difference between the characteristic liquid level height TG and the standard value of the liquid level height CTG is greater than the standard deviation BTC.

[0193] Extract the initial values ​​CZ1 for α1, CZ2 for α2, and CZ3 for α3; where the initial value of α1 can be 0.45, the initial value of α2 can be 0.2, and the initial value of α3 can be 0.35.

[0194] The reference scaling factor CB is determined by formula (11) based on the number of iterations C1, C2, C3, initial values ​​CZ1, CZ2, and CZ3. y Where y takes the values ​​1, 2, and 3; when y = 1, CB y α1 represents the reference scaling factor, Cy represents the number of iterations C1, and αy represents the scaling factor α1; when y = 2, CB yα represents the reference scaling factor α2, Cy represents the number of iterations C2, and αy represents the scaling factor α2; when y = 3, CB y α represents the reference scaling factor α3, Cy represents the number of times C3, and αy represents the scaling factor α3.

[0195] Based on the reference scaling factor CB y The values ​​of the proportional adjustment coefficients α1, α2 and α3 are determined by calculation formula (12);

[0196] The calculation formula (11) is:

[0197]

[0198] The calculation formula (12) is:

[0199]

[0200] It is worth noting that the dynamic value selection mechanism of α1, α2, and α3 can adaptively optimize the proportional adjustment coefficients based on the actual historical operating conditions of the high-temperature superconducting reactor, thus making the calculation of health factors more closely reflect the actual operating characteristics of the equipment. By statistically analyzing the number of times the health factor is marked as 0 for characteristic temperature, characteristic pressure, and characteristic liquid level, and then normalizing it using initial values ​​and reference proportional coefficients, the weight allocation can be continuously adjusted according to changes in the long-term operating data of the equipment. This method avoids the insufficient adaptability problem of traditional fixed coefficient methods. Especially in scenarios where the equipment operating environment or load conditions change significantly, the fluctuation of a certain parameter may have different risk contributions at different stages. The dynamically adjusted α can reflect such risk changes in a timely manner, improving both the sensitivity and accuracy of the health factor model. This not only improves the accuracy of anomaly detection but also reduces the possibility of false positives and false negatives, making real-time and predictive health status assessments more scientific and reliable, ultimately effectively enhancing the safe operation capability and predictive prevention level of high-temperature superconducting reactors.

[0201] This application issues warning information based on health factor one and health factor two, including:

[0202] Extract health factor KZ1. When health factor KZ1 is not greater than alarm threshold 1, perform an emergency shutdown and send an alarm message to the administrator requiring immediate maintenance. When health factor KZ1 is greater than alarm threshold 1 but less than alarm threshold 2, activate redundant cooling and send an alarm message to the administrator requiring maintenance. When health factor KZ1 is not less than alarm threshold 2, do not send an alarm message. Alarm threshold 1 and alarm threshold 2 are manually set, and alarm threshold 1 is less than alarm threshold 2.

[0203] Extract health factor 2 KZ2 at each time point in the next time period h When the health factor KZ1 at a given time point is not greater than alarm threshold 1, an alarm message is sent to the manager indicating that the high-temperature superconducting reactor has a fault at the corresponding time and needs to be repaired in advance; when the health factor KZ1 at a given time point is greater than alarm threshold 1 but less than alarm threshold 2, an alarm message is sent to the manager indicating that the high-temperature superconducting reactor may have a fault at the corresponding time and needs to reduce its operating frequency.

[0204] Please see Figure 2 A second aspect of the present invention provides an intelligent monitoring system for a high-temperature superconducting reactor, comprising: an intelligent fusion module, and a data processing module and a status analysis module connected to the intelligent fusion module;

[0205] Data processing module: Adaptively sets the installation location of the temperature sensor, and acquires reference data for the operation of the high-temperature superconducting reactor based on the temperature sensor and preset monitoring devices; wherein, the preset monitoring devices include a pressure transmitter and a capacitive level gauge; the reference data includes temperature, pressure and liquid level.

[0206] Intelligent fusion module: Synchronously matches the reference data to obtain the feature values ​​of the reference data, divides the high-temperature superconducting reactor into time periods, and performs trend prediction based on the feature values ​​of the reference data to obtain the predicted reference data for the next time period of the current time; wherein, the feature values ​​of the reference data include characteristic temperature, characteristic pressure and characteristic liquid level height;

[0207] The status analysis module obtains the health factor 1 of the high-temperature superconducting reactor at the current time based on the feature values ​​of the reference data, and the health factor 2 of the high-temperature superconducting reactor in the next time period based on the predicted reference data; and issues early warning information based on health factor 1 and health factor 2.

[0208] A third aspect of the present invention provides a storage medium, characterized in that it is used to store a computer program, which, when executed, implements an intelligent monitoring method for a high-temperature superconducting reactor.

[0209] Some of the data in the above formula are calculated by removing dimensions and taking their numerical values. The formula is the closest to the real situation obtained by software simulation of a large amount of collected data. The preset parameters and preset thresholds in the formula are set by those skilled in the art according to the actual situation or obtained through simulation of a large amount of data.

[0210] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A smart monitoring method for high-temperature superconducting reactors, characterized in that, include: The installation point of the temperature sensor is adaptively set, and reference data for the operation of the high-temperature superconducting reactor is obtained based on the temperature sensor and the preset monitoring device; the preset monitoring device includes a pressure transmitter and a capacitive level gauge; the reference data includes temperature, pressure and liquid level. The feature values ​​of the reference data are obtained by synchronous matching of the reference data, and the high-temperature superconducting reactor is divided into time periods. Trend prediction is performed based on the feature values ​​of the reference data to obtain the predicted reference data for the next time period of the current time; the feature values ​​of the reference data include feature temperature, feature pressure, and feature liquid level height; Based on the feature values ​​of the reference data, the health factor one of the high-temperature superconducting reactor at the current time is obtained, and the health factor two of the high-temperature superconducting reactor in the next time period is obtained based on the predicted reference data; an early warning message is issued based on the health factor one and the health factor two. The step of synchronously matching the reference data to obtain the feature values ​​of the reference data includes: The temperature sensors are sequentially labeled as target temperature sensors, and the temperatures monitored by the target temperature sensors within the last n minutes are obtained. Regarding temperature Perform linear fitting to determine temperature trend Use temperature in sequence Subtract temperature trend Obtain the temperature residual Obtain several temperature residuals Standard deviation The confidence weight of the current target sensor is determined by calculation formula (2). ;in, The target temperature sensor is numbered, and The value is , The maximum value for the target temperature sensor number; The time number monitored within the last n minutes; Based on the temperature most recently acquired by the target temperature sensor. And credibility weight The weighted median is determined by formula (3). Based on weighted median ,temperature And credibility weight The weighted median absolute deviation is determined by formula (4). Based on weighted median absolute deviation The rejection threshold is determined by formula (5). ; Based on the removal threshold Temperature Outlier removal yields a temperature retention set. The temperature retention set Satisfy the calculation formula (6); retain the set of temperatures. Retained in and the corresponding credibility weight The characteristic temperature is obtained by weighted calculation. ; Characteristic temperatures were extracted from a pressure level reference database. Corresponding reference pressure and reference liquid level height ; reference pressure The characteristic pressure is calculated by weighting the maximum and average pressure values ​​from the most recent pressure data obtained in the reference data. ; Reference liquid level height The feature liquid level height is calculated by weighting the minimum and average liquid level heights most recently acquired in the reference data. ; The calculation formula (2) is: ； In the formula, To prevent division by zero of small constants; The calculation formula (3) is: ； In the formula, for The corresponding target temperature sensor most recently acquired temperature; The calculation formula (4) is: ； The calculation formula (5) is: ； In the formula, This is determined based on the temperature fluctuation range within the time interval from 2n to n minutes from the current time, and The value range of is [2,3]; The calculation formula (6) is: 。 2. The intelligent monitoring method for a high-temperature superconducting reactor according to claim 1, characterized in that, The adaptive setting of the installation point of the temperature sensor includes: Select the physical properties of solid and liquid materials, set the boundary conditions for convective and radiative heat transfer in actual use, as well as the Dewar inlet velocity and outlet pressure; The gas-liquid two-phase flow was simulated using the VOF multiphase flow model in finite element simulation software. The temperature field was calculated using the energy equation, and turbulence was simulated using the K-epsilon viscosity equation. The temperature and flow fields at the Dewar and reactor frame were simulated and calculated. The physical properties included density, thermal conductivity, and viscosity. Based on the temperature and flow fields at the Dewar and reactor frame, fiber optic sensors are arranged in large-area flow dead zones, and single-point temperature sensors are arranged in small-area flow dead zones. The temperature sensors include fiber optic sensors and single-point temperature sensors. Flow dead zones with a diameter of more than 1m are considered large-area flow dead zones, and flow dead zones with a diameter of less than 1m are considered small-area flow dead zones.

3. The intelligent monitoring method for a high-temperature superconducting reactor according to claim 1, characterized in that, The reference data obtained based on temperature sensors and preset monitoring devices during the operation of the high-temperature superconducting reactor includes: A1: Obtain the number of warnings YC during the current operation of the high-temperature superconducting reactor, and determine whether the number of warnings YC exceeds the warning number threshold YZ; if yes, set the current dynamic monitoring frequency DP of the high-temperature superconducting reactor to the minimum value of the standard frequency range; if no, proceed to A2; wherein, the standard frequency range is the frequency range set according to the safety monitoring accuracy of the high-temperature superconducting reactor; the warning number threshold YZ is obtained according to the service life of the high-temperature superconducting reactor; A2: Based on the number of warnings YC, the current dynamic monitoring frequency DP of the high-temperature superconducting reactor is determined by formula (1); The calculation formula (1) is: ； In the formula, BPZ is the maximum value of the standard frequency range, and BPX is the minimum value of the standard frequency range. A3: Based on the dynamic monitoring frequency DP, the temperature inside the Dewar of the high-temperature superconducting reactor is obtained through a temperature sensor during operation; based on the dynamic monitoring frequency DP, the pressure inside the Dewar of the high-temperature superconducting reactor is obtained through a pressure transmitter during operation; based on the dynamic monitoring frequency DP, the liquid level inside the Dewar of the high-temperature superconducting reactor is obtained through a capacitive level gauge during operation.

4. The intelligent monitoring method for a high-temperature superconducting reactor according to claim 1, characterized in that, The time period division of the high-temperature superconducting reactor includes: Extract the most recent u-time characteristic temperature of the high-temperature superconducting reactor at the first time point of the current time period. And based on the calculation formula (7), the two adjacent characteristic temperatures are determined. Rate of temperature change between Obtain several temperature change rates average ; Standard duration for extracting time periods Based on average and standard duration The dynamic duration of the next time period is determined by formula (8). ; Combine the last time point of the current time period with the dynamic duration Add the times together to get the time range for the next time period; The calculation formula (7) is: ； The calculation formula (8) is: ； In the formula, The standard rate of change is set based on the service life of the high-temperature superconducting reactor.

5. The intelligent monitoring method for a high-temperature superconducting reactor according to claim 1, characterized in that, The method of obtaining the prediction reference data for the next time period based on the feature values ​​of the reference data includes: Extract the time range of several time periods from the historical reference data, the characteristic temperature, characteristic pressure and characteristic liquid level height obtained each time within the time period, and the time range of the next time period, the characteristic temperature, characteristic pressure and characteristic liquid level height at each time point within the time period; The system integrates several time ranges, characteristic temperatures, pressures, and liquid levels acquired at each time point within a given time period, the corresponding time range of the next time period, and the characteristic temperatures, pressures, and liquid levels at each time point within that time period into training and testing data. The training data is used to train the AI ​​model, and the testing data is used to test the trained AI model. The AI ​​model is then adjusted based on the testing results. The final result is a reference data prediction model that takes the current time range, the characteristic temperatures, pressures, and liquid levels acquired at each time point within that time period, and the time range of the next time period as input, and outputs the characteristic temperatures, pressures, and liquid levels at each time point within the next time period. The AI ​​model includes both a BP neural network model and an RBF neural network model. By inputting the time range of the current time period, the characteristic temperature, characteristic pressure, and characteristic liquid level height acquired each time within the time period, and the time range of the next time period into the reference data prediction model, the characteristic temperature, characteristic pressure, and characteristic liquid level height at each time point in the next time period can be obtained.

6. The intelligent monitoring method for a high-temperature superconducting reactor according to claim 1, characterized in that, The health factor one of the high-temperature superconducting reactor at the current time is obtained based on the feature values ​​of the reference data, and the health factor two of the high-temperature superconducting reactor at the next time period is obtained based on the predicted reference data, including: Real-time extraction of the characteristic temperature of the high-temperature superconducting reactor at the current time Characteristic pressure and characteristic liquid level height Determine if a characteristic temperature exists. greater than the temperature threshold or characteristic liquid level height The absolute value of the difference between the liquid level and the standard value Greater than the standard deviation or characteristic pressure greater than the pressure threshold One or more of the following; yes, the health factors at the current time. Mark as 0; otherwise, determine the current time's health factor based on formula (9). ; Extract the characteristic temperatures of the high-temperature superconducting reactor at each time point in the next time period. Characteristic pressure and characteristic liquid level height ; sequentially determine whether characteristic temperatures exist at each time point in the next time period. greater than the temperature threshold or characteristic liquid level height The absolute value of the difference between the liquid level and the standard value Greater than the standard deviation or characteristic pressure greater than the pressure threshold One or more of the following; yes, the corresponding health factor II at the corresponding time point. Mark as 0; otherwise, determine the health factor II at the corresponding time point based on formula (10). ; The calculation formula (9) is: ； In the formula, , and It is a proportional adjustment coefficient determined based on the number of times in history that characteristic temperatures, characteristic pressures, and characteristic liquid levels have caused the health factor to be marked as 0; These are standard constants; The calculation formula (10) is: 。 7. The intelligent monitoring method for a high-temperature superconducting reactor according to claim 6, characterized in that, The , and The proportional adjustment coefficient, determined based on the number of times historical characteristic temperatures, pressures, and liquid levels resulted in a health factor being marked as 0, can be explained as follows: Obtain the characteristic temperature from the historical records of the current high-temperature superconducting reactor since it was put into operation. greater than the temperature threshold Number of times Characteristic pressure greater than the pressure threshold Number of times Characteristic liquid level height The absolute value of the difference between the liquid level and the standard value Greater than the standard deviation Number of times ; extract initial value , initial value and initial value ; Based on number ,frequency ,frequency Initial value Initial value and initial value The reference scaling factor is determined by formula (11). ;in, The values ​​are 1, 2, and 3; Based on reference scaling factor The proportional adjustment coefficient is determined by formula (12). , and The value; The calculation formula (11) is: ； The calculation formula (12) is: 。 8. The intelligent monitoring method for a high-temperature superconducting reactor according to claim 1, characterized in that, The warning information issued based on health factor one and health factor two includes: Extracting health factors When health factors If the health factor is not greater than the alarm threshold, an emergency shutdown will be initiated, and an alarm message requiring immediate maintenance will be sent to the administrator; when the health factor is... When the value is greater than alarm threshold one and less than alarm threshold two, redundant cooling is activated, and an alarm message requiring maintenance is sent to the administrator; when health factor one... No alarm message is issued when the value is not less than alarm threshold two; where alarm threshold one is less than alarm threshold two. Extracting health factors at each time point in the next time period (Part 2) When there are health factors at a certain point in time When the value is not greater than the alarm threshold, an alarm message is sent to the manager indicating that the high-temperature superconducting reactor has malfunctioned at the corresponding time and requires prior maintenance; when there is a health factor at the time point... When the value exceeds alarm threshold one but is less than alarm threshold two, an alarm message is sent to the administrator indicating that the high-temperature superconducting reactor may malfunction at the corresponding time and that the operating frequency needs to be reduced.

9. An intelligent monitoring system for a high-temperature superconducting reactor, used to operate the intelligent monitoring method for a high-temperature superconducting reactor according to any one of claims 1 to 8, characterized in that, include: The intelligent fusion module, and the data processing module and status analysis module connected to the intelligent fusion module; The data processing module adaptively sets the installation location of the temperature sensor and acquires reference data for the operation of the high-temperature superconducting reactor based on the temperature sensor and a preset monitoring device; wherein, the preset monitoring device includes a pressure transmitter and a capacitive level gauge; the reference data includes temperature, pressure and liquid level. The intelligent fusion module: synchronously matches the reference data to obtain the feature values ​​of the reference data, divides the high-temperature superconducting reactor into time periods, and performs trend prediction based on the feature values ​​of the reference data to obtain the predicted reference data for the next time period of the current time; wherein, the feature values ​​of the reference data include feature temperature, feature pressure, and feature liquid level height; The status analysis module: obtains the health factor one of the high-temperature superconducting reactor at the current time based on the feature value of the reference data, and obtains the health factor two of the high-temperature superconducting reactor in the next time period based on the predicted reference data; and issues early warning information based on health factor one and health factor two. The step of synchronously matching the reference data to obtain the feature values ​​of the reference data includes: The temperature sensors are sequentially labeled as target temperature sensors, and the temperatures monitored by the target temperature sensors within the last n minutes are obtained. Regarding temperature Perform linear fitting to determine temperature trend Use temperature in sequence Subtract temperature trend Obtain the temperature residual Obtain several temperature residuals Standard deviation The confidence weight of the current target sensor is determined by calculation formula (2). ;in, The target temperature sensor is numbered, and The value is , The maximum value for the target temperature sensor number; The time number monitored within the last n minutes; Based on the temperature most recently acquired by the target temperature sensor. And credibility weight The weighted median is determined by formula (3). Based on weighted median ,temperature And credibility weight The weighted median absolute deviation is determined by formula (4). Based on weighted median absolute deviation The rejection threshold is determined by formula (5). ; Based on the removal threshold Temperature Outlier removal yields a temperature retention set. The temperature retention set Satisfy the calculation formula (6); retain the set of temperatures. Retained in and the corresponding credibility weight The characteristic temperature is obtained by weighted calculation. ; Characteristic temperatures were extracted from a pressure level reference database. Corresponding reference pressure and reference liquid level height ; reference pressure The characteristic pressure is calculated by weighting the maximum and average pressure values ​​from the most recent pressure data obtained in the reference data. ; Reference liquid level height The feature liquid level height is calculated by weighting the minimum and average liquid level heights most recently acquired in the reference data. ; The calculation formula (2) is: ； In the formula, To prevent division by zero of small constants; The calculation formula (3) is: ； In the formula, for The corresponding target temperature sensor most recently acquired temperature; The calculation formula (4) is: ； The calculation formula (5) is: ； In the formula, This is determined based on the temperature fluctuation range within the time interval from 2n to n minutes from the current time, and The value range of is [2,3]; The calculation formula (6) is: 。 10. A storage medium, characterized in that, Used to store a computer program, which, when executed, implements the intelligent monitoring method for a high-temperature superconducting reactor as described in any one of claims 1 to 8.

Images