A vacuum chamber resistance determination method, device, equipment, storage medium and product

Abstract

The application discloses a vacuum chamber resistance determination method, device, equipment, storage medium and product, a feature matrix is obtained by performing spectral embedding transformation on a mutual inductance sub-matrix of a to-be-recognized loop set, loops with close electromagnetic coupling are classified into one category according to the feature matrix, only one shared gain coefficient needs to be recognized in one partition, an original high-dimensional resistance recognition problem is reduced to a low-dimensional gain optimization problem, deviation between a measured vacuum field magnetic probe signal and a simulation magnetic probe signal corresponding to the gain coefficient is minimized as an optimization target, an optimal gain coefficient is solved, and an initial recognition resistance of the vacuum chamber is determined based on the optimal gain coefficient, overfitting risk is effectively suppressed while physical rationality is ensured, and vacuum chamber resistance recognition accuracy is improved.

Description

Technical Field

[0001] This invention relates to the field of plasma control technology, and more specifically, to a method, apparatus, equipment, storage medium, and product for determining the resistance of a vacuum chamber. Background Technology

[0002] In magnetic confinement fusion devices, rapid changes in external coil current and plasma current during plasma discharge induce significant eddy currents on the vacuum chamber walls. Active control of the plasma's vertical displacement relies on the adjustment of the fast control coil, while the eddy current mirror effect generated by passive structures is used to suppress vertical instabilities. The stabilization effect of the eddy currents is determined by their decay rate, which essentially depends on the equivalent resistance of the vacuum chamber.

[0003] In plasma control, the magnitude of eddy currents is a critical variable that must be accurately measured, and the output current of the fast control coil needs to be compensated in real time based on this. However, errors in the identification of vacuum chamber resistance directly lead to inaccurate eddy current estimation, preventing the fast control coil from outputting an accurate compensation current. This weakens the effectiveness of vertical displacement control and increases the risk of vertical displacement events (VDEs). Once a VDE occurs, it may cause plasma rupture and impact the device structure. Especially in long-pulse high-confinement mode operation scenarios, eddy current compensation errors accumulate continuously during the hundreds of seconds of discharge, causing slow drift of control parameters. In severe cases, this will trigger the system's protective termination, preventing the device from reaching its designed operating range. Therefore, accurate identification of vacuum chamber resistance is not only a parameter acquisition issue but also a critical prerequisite for the safe and stable operation of the device.

[0004] The common method for resistance identification is to fit the experimental data by treating the resistance of each loop in the vacuum chamber as an independent parameter. However, there are far more parameters to be identified than independent observations, and the contributions of eddy currents in different loops to the same probe are highly coupled. Directly fitting and solving the resistance parameters of all loops faces the problem of mathematical ill-posedness, which can easily lead to overfitting or even non-physical solutions. Summary of the Invention

[0005] Based on this, the present invention provides a method, apparatus, device, storage medium and product for determining the resistance of a vacuum chamber, in order to solve the defects of the prior art that the identification results are overfitted or produce non-physical results due to too many identification parameters of the resistance of each circuit of the vacuum chamber and severe signal coupling.

[0006] To achieve the above objectives, embodiments of the present invention provide a method for determining the resistance of a vacuum chamber, comprising: Obtain the prior resistance vector of the vacuum chamber circuit, the mutual inductance submatrix of the circuit set to be identified, and the measured magnetic probe signal of the vacuum field; Perform a spectral embedding transformation on the mutual inductance submatrix to obtain the feature matrix; Based on the feature matrix, all the circuits to be identified in the set of circuits to be identified are clustered to obtain several partitions; Each partition is assigned an independent resistance gain coefficient to be solved, and the simulated magnetic probe signal corresponding to the resistance gain coefficient to be solved is determined. With the optimization objective of minimizing the deviation between the measured magnetic probe signal in the vacuum field and the simulated magnetic probe signal, the gain coefficient of the resistor to be solved is obtained to obtain the optimal gain coefficient. The initial identification resistance of the vacuum chamber is obtained by correcting the prior resistance vector based on the optimal gain coefficient.

[0007] To achieve the above objectives, embodiments of the present invention also provide a vacuum chamber resistance determination device, comprising: The data acquisition module is used to acquire the prior resistance vector of the vacuum chamber circuit, the mutual inductance submatrix of the circuit set to be identified, and the measured magnetic probe signal of the vacuum field. The feature extraction module is used to perform spectral embedding transformation on the mutual inductance submatrix to obtain the feature matrix; The loop clustering module is used to cluster all the loops to be identified in the set of loops to be identified based on the feature matrix, and obtain several partitions. The signal simulation module is used to assign an independent resistance gain coefficient to be solved for each partition and to determine the simulated magnetic probe signal corresponding to the resistance gain coefficient to be solved. The gain calculation module is used to solve for the gain coefficient of the resistor to be solved with the optimization objective of minimizing the deviation between the measured magnetic probe signal of the vacuum field and the simulated magnetic probe signal, and to obtain the optimal gain coefficient. The resistance determination module is used to correct the prior resistance vector based on the optimal gain coefficient to obtain the initial identification resistance of the vacuum chamber.

[0008] To achieve the above objectives, embodiments of the present invention also provide a vacuum chamber resistance determination device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the vacuum chamber resistance determination method as described in any of the above embodiments.

[0009] To achieve the above objectives, embodiments of the present invention also provide a computer-readable storage medium, the computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to perform the vacuum chamber resistance determination method as described in any of the above embodiments.

[0010] To achieve the above objectives, embodiments of the present invention also provide a computer program product, including a computer program / instructions, which, when executed by a processor, implement the vacuum chamber resistance determination method as described in any of the above embodiments.

[0011] Compared with the prior art, the vacuum chamber resistance determination method, apparatus, device, storage medium, and product disclosed in this invention first acquire the mutual inductance submatrix of the set of circuits to be identified and the measured magnetic probe signal of the vacuum field; then, perform spectral embedding transformation on the mutual inductance submatrix to obtain a feature matrix, and cluster all circuits to be identified in the set of circuits to be identified based on the feature matrix to obtain several partitions; next, assign an independent resistance gain coefficient to each partition, and determine the simulated magnetic probe signal corresponding to the resistance gain coefficient; finally, with the optimization objective of minimizing the deviation between the measured magnetic probe signal of the vacuum field and the simulated magnetic probe signal, solve the resistance gain coefficient to obtain the optimal gain coefficient, and determine the initial identification resistance of the vacuum chamber based on the optimal gain coefficient. Therefore, the embodiments of the present invention obtain the feature matrix by performing spectral embedding transformation on the mutual inductance submatrix of the set of circuits to be identified. Based on the feature matrix, the electromagnetically tightly coupled circuits are classified into one class. Only one shared gain coefficient needs to be identified for each partition. The original high-dimensional resistance identification problem is reduced to a low-dimensional gain optimization problem. While ensuring physical rationality, the risk of overfitting is effectively suppressed, and the accuracy of vacuum chamber resistance identification is improved. Attached Figure Description

[0012] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments 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.

[0013] Figure 1 This is a schematic flowchart of a method for determining the resistance of a vacuum chamber according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the mapping of partitioning results on the poloidal section of a vacuum chamber according to an embodiment of the present invention; Figure 3 This is a bar chart illustrating the optimal gain coefficients of each partition according to an embodiment of the present invention; Figure 4 This is a fitting curve of the vertical displacement growth rate exponential model provided in an embodiment of the present invention; Figure 5 This is a schematic diagram of a vacuum chamber resistance determination device provided in an embodiment of the present invention; Figure 6This is a schematic diagram of a vacuum chamber resistance determination device provided in an embodiment of the present invention. Detailed Implementation

[0014] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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.

[0015] See Figure 1 This is a flowchart illustrating a method for determining the resistance of a vacuum chamber according to an embodiment of the present invention. Specifically, the method for determining the resistance of a vacuum chamber includes steps S1 to S6: S1. Obtain the prior resistance vector of the vacuum chamber circuit, the mutual inductance submatrix of the circuit set to be identified, and the measured magnetic probe signal of the vacuum field.

[0016] Specifically, regarding the mutual inductance sub-matrix: from the full-loop mutual inductance matrix of the vacuum chamber (i.e., the mutual inductance matrix between the vacuum chamber loop and the external coil), the portion of mutual inductance data belonging to the set of loops to be identified (not all loops) is extracted; the mutual inductance sub-matrix reflects the electromagnetic coupling strength between the loops to be identified and serves as the physical basis for subsequent spectral clustering dimensionality reduction. The measured magnetic probe signal in the vacuum field is a time series of magnetic probe signals from the vacuum field testing experiment.

[0017] S2. Perform spectral embedding transformation on the mutual inductance submatrix to obtain the feature matrix.

[0018] Specifically, in step S2, the mutual inductance submatrix is ​​normalized to obtain a normalized electromagnetic coupling weight matrix; a degree matrix and a graph Laplace matrix are constructed based on the normalized electromagnetic coupling weight matrix; and the feature matrix of the set of loops to be identified is obtained based on the degree matrix and the graph Laplace matrix.

[0019] For example, the feature matrix is ​​determined as follows: Determine the subset of the loop index to be identified Extract the corresponding mutual inductance submatrix ,in, The number of loops to be identified; based on Constructing a normalized electromagnetic coupling weight matrix ,in: ; diagonal elements =0; It is the normalized electromagnetic coupling weight matrix No. Line number Column elements; It is the mutual inductance submatrix of the th Line number Column elements; It is the mutual inductance submatrix of the th Line number Column elements; It is the mutual inductance submatrix of the th Line number The elements of the column.

[0020] Construct the degree matrix And the Laplace matrix Solve for the generalized eigenvalues: ; in, This represents the diagonal matrix generating function. Represents generalized eigenvalues. It is a generalized eigenvector; Take before The eigenvectors corresponding to the smallest non-zero eigenvalues ​​constitute the feature matrix. ; For the characteristic matrix of Execute each row vector mean ( Clustering divides the circuits to be identified into... There are 8 partitions; the number of partitions can be preset, for example, the default setting is 8.

[0021] S3. Based on the feature matrix, cluster all the circuits to be identified in the set of circuits to be identified to obtain several partitions.

[0022] Specifically, in step S3, clustering is performed in the spectral space based on the row vectors of the feature matrix. Each partition is naturally derived from the spectral structure of the mutual inductance submatrix. Loops within the same partition may not be spatially adjacent, but they are equivalent in the sense of electromagnetic coupling, ensuring the physical rationality of the loop partitioning. Based on this, this partitioning method can reflect the resistance differences in different regions of the vacuum chamber, such as the difference between densely welded areas and continuous wall areas, and between areas near and far from the port, thus maintaining the ability to express the spatial non-uniformity of resistance. This dimensionality reduction strategy effectively solves the ill-posed problem of independent fitting of the entire loop, transforming the high-dimensional resistance identification problem into a low-dimensional gain optimization problem, significantly improving numerical stability and generalization ability.

[0023] S4. Assign an independent resistance gain coefficient to each partition and determine the simulated magnetic probe signal corresponding to the resistance gain coefficient.

[0024] Specifically, each partition is assigned a gain coefficient to be solved for resistance, and all loops within the same partition share this gain coefficient. Based on the current gain coefficient, the simulated magnetic probe signal is determined according to the law of electromagnetic induction, circuit theory, and the principle of linear superposition of eddy current fields. Specifically, based on the current gain coefficient, the resistance value of each loop is calculated in combination with the prior resistance vector of the vacuum chamber. The resistance value is substituted into the eddy current dynamic state-space model, and the simulated eddy current time series is obtained through time-domain numerical integration. Then, it is multiplied by the Green's function to obtain the simulated magnetic probe signal.

[0025] S5. With minimizing the deviation between the measured magnetic probe signal in the vacuum field and the simulated magnetic probe signal as the optimization objective, solve for the gain coefficient of the resistor to be solved, and obtain the optimal gain coefficient.

[0026] Specifically, in step S5, an optimization problem is constructed with the resistance gain coefficient to be solved as the independent variable and the deviation between the simulated magnetic probe signal and the measured magnetic probe signal in the vacuum field as the dependent variable. The optimization objective is to minimize the deviation. The measured magnetic probe signal in the vacuum field is equal to the magnetic field strength measured by the magnetic probe signal minus the magnetic field strength contributed by the external coil.

[0027] S6. Based on the optimal gain coefficient, correct the prior resistance vector to obtain the initial identification resistance of the vacuum chamber.

[0028] Specifically, in step S6, the initial identification resistor is obtained by fitting the magnetic probe signal under vacuum field conditions, and its spatial distribution pattern (i.e., the relative relationship between each partition) has been calibrated.

[0029] It is worth noting that steps S1 to S6 determine the relative gain distribution of the resistance of each zone under vacuum conditions by fitting the magnetic probe signal, thus obtaining the initial identification resistance. It is understandable that in vacuum field testing, there is no plasma, and the eddy currents are determined solely by the coupling between the coil and the vacuum chamber; therefore, the relative relationships between the zones can stably converge to a reasonable spatial distribution pattern.

[0030] It is worth noting that the method is applicable to the identification of vacuum chamber resistance in tokamak and other axisymmetric magnetic confinement devices, including the HL-3, Xuanlong-50U, and Experimental Advanced Superconducting Tokamak (EAST). The vacuum chamber of a tokamak device is a ring-shaped, double-layered, sealed container made of materials such as stainless steel, responsible for maintaining high vacuum and supporting the first wall assembly. Taking EAST as an example, its vacuum chamber adopts a double-layered D-shaped cross-section structure, with both inner and outer wall thicknesses of approximately 8 mm, and includes numerous windows, bellows sections, and welded connections; its actual geometric complexity far exceeds that of the idealized model. During plasma discharge, the rapid changes in external coil current and plasma current induce significant eddy currents on the vacuum chamber wall. During the EAST plasma current ramp-up phase, the peak value of the induced eddy current in the vertical section outside the vacuum chamber can reach approximately 65 kA, about 10% of the plasma current.

[0031] Accurate estimation of eddy currents is particularly crucial for confinement performance. EAST's active vertical displacement control is achieved through fast control coils, while also relying on passive stabilization of vertical instability by a passive structure. The mirror current generated by eddy currents in the wall resists changes in plasma vertical displacement; its stabilization effect depends on the decay rate of the eddy currents. The lower the vacuum chamber resistance, the slower the eddy current decay and the stronger the stabilization effect. During EAST's active vertical displacement control, if there is a misidentification of the vacuum chamber resistance, the feedforward eddy current compensation will be inaccurate, preventing the fast control coils from outputting accurate compensation current. This directly increases the risk of Vertical Displacement Events (VDEs). A VDE can lead to plasma rupture and impact the device's first wall and vacuum chamber structure. Simultaneously, the magnetic field generated by the eddy currents directly superimposes with the coil magnetic field and the plasma current magnetic field. This means that diagnostic signals from magnetic probes and flux loops do not purely reflect the contribution of the plasma itself. If the plasma equilibrium reconstruction cannot accurately subtract the eddy current component, the resulting plasma parameters will exhibit systematic deviations. For devices like EAST, which primarily operate in long-pulse H-mode, eddy current compensation errors accumulate continuously during discharges lasting hundreds of seconds, causing slow parameter drift and potentially triggering system protective termination in severe cases. Here, H-mode represents the high-constraint mode.

[0032] Compared with existing technologies, the embodiments of the present invention obtain a feature matrix by performing a spectral embedding transformation on the mutual inductance submatrix of the set of vacuum chamber circuits to be identified. Based on the feature matrix, circuits with tight electromagnetic coupling are grouped into one category. Only one shared gain coefficient needs to be identified for each partition, reducing the original high-dimensional resistance identification problem to a low-dimensional gain optimization problem. This effectively suppresses the risk of overfitting while ensuring physical rationality, thus improving the accuracy of vacuum chamber resistance identification. It can be understood that in active vertical displacement control, the identified vacuum chamber resistance can be used to accurately calculate the eddy current in the vacuum chamber, and based on this eddy current, an accurate compensation current can be provided to the fast control coil, reducing the risk of vertical displacement events and thus reducing the risk of plasma rupture, thereby protecting the device structure.

[0033] In the second preferred embodiment, based on steps S1 to S6, the method further includes: Obtain the number of magnetic probes and the number of circuits to be identified; The number of partitions is determined by the minimum of the square root of the number of circuits to be identified and half the number of magnetic probes.

[0034] Specifically, the number of partitions required for clustering It can be predetermined using an empirical formula, as follows: .

[0035] It is understood that the embodiments of the present invention are based on the number of circuits to be identified. The square root and the number of magnetic probes The minimum of the two values ​​(half and half) determines the number of partitions. This ensures that the number of partitions does not exceed the observation information capacity to avoid overfitting, while also guaranteeing that the number of partitions is sufficient to represent the resistance differences between different regions of the vacuum chamber, achieving a balance between identification accuracy and computational complexity. Furthermore, this method relies only on two basic parameters: the number of loops and the number of magnetic probes, exhibiting good device versatility.

[0036] In a third preferred embodiment, based on any of the above embodiments, it further includes: Substitute the initial identification resistor into the pre-constructed rigid plasma radial-vertical motion model to determine the plasma vertical displacement growth rate corresponding to the initial identification resistor; Acquire the free drift data of the plasma vertical displacement, and determine the vertical displacement instability growth rate based on the free drift data of the plasma vertical displacement; The initial identification resistor is scaled by multiplying the ratio of the vertical displacement instability growth rate to the plasma vertical displacement growth rate to obtain the final identification resistor.

[0037] For example, after obtaining the initial identification resistor, the resistor is corrected in the following way: During EAST's normal discharge, the vertical displacement controller was turned off, and data on the free drift portion of the plasma vertical displacement was extracted. An exponential model was then applied to this portion of the signal. Boundary constraints ( ≥0) Nonlinear least squares fitting, calculate the vertical displacement instability growth rate of experimental data. ;in, Indicates time item, Indicates time The actual observed values ​​(i.e., the data of the free drift of plasma vertical displacement). , and The coefficients are to be fitted; after the fitting converges, The solution result is denoted as the vertical displacement instability growth rate. .

[0038] Initial identification resistor Substituting into the rigid plasma radial-vertical motion model (RZIp), calculate the plasma vertical displacement growth rate corresponding to this resistance. In terms of ratio As a multiplier, for Scaling is performed to obtain the final identification resistor. .

[0039] It is worth noting that the identification obtained in this embodiment It directly serves the eddy current field compensation for real-time plasma equilibrium reconstruction, the construction of vertical displacement response models, and the environmental dynamics modeling of plasma configuration controllers based on reinforcement learning, thereby achieving stable and reliable operation of the device.

[0040] Furthermore, the method also includes: Acquire multi-shot discharge data; Using the multi-shot discharge data, the steps of determining the plasma vertical displacement growth rate and the vertical displacement instability growth rate, as well as the scaling operation of the initial identification resistor, are repeated to verify the prediction accuracy of the final identification resistor. Furthermore, the optimal gain coefficient, vertical displacement instability growth rate, and final identification resistor for each partition are output.

[0041] Specifically, the calculation of plasma vertical displacement growth rate, vertical displacement instability growth rate, and correction of initial identification resistance are repeated using multi-shot discharge data (such as EAST discharge data). The resulting multiple final identification resistances are then analyzed to determine the prediction accuracy of the final identification resistance.

[0042] It is worth noting that, to reduce the risk of systematic deviations in the absolute magnitude of the identified vacuum chamber resistance after the introduction of plasma, this embodiment compares the vertical displacement growth rate calculated by the vertical displacement free drift experiment with the value calculated by the rigid model, and performs a scaling factor on the initial identified resistance. This physical-level correction of its absolute value yields the final identified resistance. This multi-stage adjustment method ensures that the identification results have both spatial resolution and physical-scale accuracy. Simultaneously, multi-shot data is used to verify the reliability of the final identified resistance.

[0043] In a fourth preferred embodiment, based on any of the above embodiments, the step of assigning an independent unsolved resistance gain coefficient to each partition and determining the simulated magnetic probe signal corresponding to the unsolved resistance gain coefficient includes: Obtain the self-inductance matrix of the vacuum chamber circuit, the mutual inductance matrix between the vacuum chamber circuit and the external coil, the coil current time series, and the Green's function from the vacuum chamber circuit to the magnetic probe; Assign an independent resistance gain coefficient to be solved for each partition; The resistance gain coefficient to be solved in each partition is assigned to all the circuits to be identified in the corresponding partition to construct a full-loop gain vector; wherein, the gain value corresponding to the non-identified circuit in the full-loop gain vector is a preset reference value. The identification resistance to be solved is determined based on the prior resistance vector and the full-loop gain vector. The simulated magnetic probe signal is determined based on the self-inductance matrix of the vacuum chamber circuit, the mutual inductance matrix between the vacuum chamber circuit and the external coil, the coil current time series, the identification resistor to be solved, and the Green's function.

[0044] The coil current time series refers to the current time series of the external coil in the vacuum field test experiment. The simulated magnetic probe signal is determined as follows: an eddy current dynamic state-space model is constructed based on the self-inductance matrix, the mutual inductance matrix, the coil current time series, and the identification resistor to be solved; the eddy current dynamic state-space model is numerically integrated in the time domain to obtain the simulated eddy current time series; and the simulated magnetic probe signal is determined based on the simulated eddy current time series and the Green's function.

[0045] Specifically, regarding the self-induction matrix With Green's function Based on the geometric model of the EAST device, and using the TokSys tokamak plasma control simulation platform, the self-inductance matrix was determined according to the Biot-Savart law. With Green's function Regarding the prior resistance vector: It is the initial resistance estimate of each equivalent loop in the vacuum chamber obtained from theoretical calculations. It serves as the starting point and reference for subsequent identification work. Specifically, it is calculated using the resistivity of stainless steel and the cross-sectional area of ​​each loop segment.

[0046] For example, the simulated magnetic probe signal is determined as follows: Taking the EAST device as the implementation object, the vacuum chamber loop is discretized into N=182 surface elements. The last 82 equivalent loops (physical numbers 101 to 182, corresponding to the limiter, divertor, passive structure, and first wall region) are selected as the loops to be identified. The index subset of the loops to be identified is... Record the physical number of the circuit to be identified. The first 100 loops are treated as non-identifiable loops, and their gain values ​​are fixed at a preset reference value (set to 1 for ease of calculation), keeping the prior resistance vector unchanged.

[0047] Based on spectral embedding transformation and clustering algorithms, the circuits to be identified are classified to obtain, as follows: Figure 2 The diagram shows eight zones, with zones 1 through 8 represented by different colors. The area not included refers to the outer vacuum chamber. Figure 2 This is a schematic diagram of the partitioning results mapped onto the poloidal section of the vacuum chamber. The horizontal axis R (m) represents the distance in the large radius direction, and the vertical axis Z (m) represents the height in the vertical direction, with units in meters. Each of the eight partitions is assigned an independent resistance gain coefficient. This forms an 8-dimensional parameter vector. And construct a 182-dimensional gain vector through partition mapping. (i.e., the total loop gain vector); where, for loops outside the subset of the loop index to be identified, The value is 1. Prior resistance vector. The element-wise multiplication with the full-loop gain vector yields the identification resistance to be solved. Understandably, the above process reduces the original 182-dimensional identification problem to 8 dimensions.

[0048] Based on self-induction matrix Mutual inductance matrix Coil current time series, identification resistor to be solved and Green's function A forward simulation of the Ordinary Differential Equation (ODE) in the time domain is performed to determine the simulated magnetic probe signal. The specific principle is as follows: Vacuum Chamber Vortex With coil current The coupling relationship is described by the circuit equations: ; Magnetic probe signal generated by eddy current for: ; in, This is the mutual inductance matrix between the vacuum chamber circuit and the external coil. Let Green's function be the loop-to-magnetic probe. It is the self-inductance matrix of the vacuum chamber circuit.

[0049] By performing an identity transformation on the above circuit equations, a state-space model of eddy current dynamics is constructed: ; ; ; Driver item The coil current change rate is determined by the coil current change rate, which is calculated based on the coil current time series. The simulated eddy current is obtained through time-domain numerical integration, and then the simulated magnetic probe signal is calculated. This is the substitution term for the identification resistance to be solved.

[0050] In the state-space model of eddy current dynamics, numerical integration methods that automatically switch between rigid and non-rigid (such as the LSODA integrator) are used to solve the eddy current state equations. The simulated eddy current time series was obtained. The simulated magnetic probe signal is then obtained using the following formula. : ; The Chinese name for the LSODA integrator is Livermore Solver for Ordinary Differential Equations.

[0051] Finally, with the optimization objective of minimizing the deviation between the measured magnetic probe signal in the vacuum field and the simulated magnetic probe signal, the gain coefficient of the resistor to be solved is obtained, as shown below. Figure 3 The optimal gain coefficients for the eight partitions are shown, and the corresponding initial identification resistors are determined. Figure 3 In the diagram, the horizontal axis 1 to 8 refer to partitions 1 to 8 respectively, the vertical axis represents the resistance gain coefficient, and the dashed line represents the preset reference value.

[0052] Furthermore, the final identification resistance is obtained by employing the resistance adjustment method shown in the third embodiment; wherein, the data change of the vertical displacement growth rate can be found in [reference needed]. Figure 4 The horizontal axis represents the discharge time in seconds; the vertical axis represents the vertical displacement in meters; the blue dots represent the data of the free drift of the plasma vertical displacement, and the red solid line represents the fitting curve of the vertical displacement growth rate exponential model.

[0053] It is worth noting that this implementation method uses the time-domain excitation signal (i.e., the coil current time series) in the vacuum field test to achieve resistance identification. This method of directly using the conventional operating data of the device to achieve resistance identification is more time-saving than the finite element simulation method.

[0054] Furthermore, the step of minimizing the deviation between the measured magnetic probe signal in the vacuum field and the simulated magnetic probe signal, and solving for the resistance gain coefficient to obtain the optimal gain coefficient, includes: The first optimization objective is to minimize the deviation between the measured magnetic probe signal and the simulated magnetic probe signal in the vacuum field, and the second optimization objective is to minimize the degree to which the unsolved resistance gain coefficient of each partition deviates from the preset reference value, thus constructing a bi-objective optimization problem. Under the preset feasible region constraint of gain coefficient, the bi-objective optimization problem is solved to obtain the optimal gain coefficient of each partition.

[0055] Specifically, the cost function corresponding to the bi-objective optimization problem is: ; in, This is the signal from a magnetic probe measured in a vacuum field. The regularization coefficient is . The Frobenius norm is used. The regularization term anchors the gain coefficient to a preset reference value (1 in this embodiment), assuming that the prior resistance is generally accurate and only making relative adjustments between different partitions. The gain coefficient of the resistor to be solved is defined by the preset gain coefficient boundary. Under the constraints, a nonlinear least squares optimization algorithm (such as the Trust-Region Reflective algorithm (TRF)) is used to solve for the optimal gain coefficients in order to obtain the initial identification resistance of each partition. .

[0056] Compared with the prior art, the vacuum chamber resistance determination method disclosed in this invention first obtains the mutual inductance sub-matrix of the set of circuits to be identified and the measured magnetic probe signal of the vacuum field; then, it performs spectral embedding transformation on the mutual inductance sub-matrix to obtain a feature matrix, and clusters all circuits to be identified in the set of circuits to be identified based on the feature matrix to obtain several partitions; next, it assigns an independent resistance gain coefficient to each partition and determines the simulated magnetic probe signal corresponding to the resistance gain coefficient; finally, with the optimization objective of minimizing the deviation between the measured magnetic probe signal of the vacuum field and the simulated magnetic probe signal, it solves for the resistance gain coefficient to obtain the optimal gain coefficient, and determines the initial identification resistance of the vacuum chamber based on the optimal gain coefficient. Therefore, this invention, through spectral embedding transformation of the mutual inductance submatrix of the set of circuits to be identified to obtain a feature matrix, groups electromagnetically tightly coupled circuits into one category based on the feature matrix. Each partition only needs to identify one shared gain coefficient, reducing the original high-dimensional resistance identification problem to a low-dimensional gain optimization problem. This effectively suppresses overfitting risk while ensuring physical rationality, thus improving the accuracy of vacuum chamber resistance identification. Furthermore, the method relies only on standard diagnostic data, i.e., only on the existing magnetic probe and coil current related data of the device, without requiring additional hardware.

[0057] See Figure 5 , Figure 5 This is a schematic diagram of a vacuum chamber resistance determination device provided in an embodiment of the present invention. The vacuum chamber resistance determination device 20 includes: Data acquisition module 21 is used to acquire the prior resistance vector of the vacuum chamber circuit, the mutual inductance submatrix of the circuit set to be identified, and the measured magnetic probe signal of the vacuum field. Feature extraction module 22 is used to perform spectral embedding transformation on the mutual inductance submatrix to obtain the feature matrix; The loop clustering module 23 is used to cluster all the loops to be identified in the set of loops to be identified based on the feature matrix, and obtain several partitions. The signal simulation module 24 is used to assign an independent resistance gain coefficient to be solved for each partition and to determine the simulated magnetic probe signal corresponding to the resistance gain coefficient to be solved. The gain calculation module 25 is used to solve the gain coefficient of the resistor to be solved with the optimization objective of minimizing the deviation between the measured magnetic probe signal of the vacuum field and the simulated magnetic probe signal, so as to obtain the optimal gain coefficient. The resistance determination module 26 is used to correct the prior resistance vector based on the optimal gain coefficient to obtain the initial identification resistance of the vacuum chamber.

[0058] It is worth noting that the specific working process of the vacuum chamber resistance determination device can be referred to the working process of the vacuum chamber resistance determination method described in the above embodiments, and will not be repeated here.

[0059] Compared with the prior art, the vacuum chamber resistance determination device disclosed in this embodiment of the invention obtains the feature matrix by performing spectral embedding transformation on the mutual inductance submatrix of the set of circuits to be identified. Based on the feature matrix, the electromagnetically tightly coupled circuits are classified into one category. Only one shared gain coefficient needs to be identified for each partition. This reduces the original high-dimensional resistance identification problem to a low-dimensional gain optimization problem. While ensuring physical rationality, it effectively suppresses the risk of overfitting and improves the accuracy of vacuum chamber resistance identification.

[0060] See Figure 6 , Figure 6 This is a schematic diagram of a vacuum chamber resistance determination device 30 provided in an embodiment of the present invention. The vacuum chamber resistance determination device 30 includes a processor 31, a memory 32, and a computer program stored in the memory 32 and configured to be executed by the processor 31. When the processor 31 executes the computer program, it implements the steps as described in the above embodiment of the vacuum chamber resistance determination method, for example... Figure 1 The steps S1 to S6 described above; or, when the processor 31 executes the computer program, it implements the functions of each module in the above-described device embodiments.

[0061] For example, the computer program can be divided into one or more modules, which are stored in the memory 32 and executed by the processor 31 to complete the present invention. The one or more modules can be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program in the vacuum chamber resistance determination device. For example, the computer program can be divided into multiple modules, and the specific working process of each module can be referred to the working process of the vacuum chamber resistance determination device described in the above embodiments, which will not be repeated here.

[0062] The vacuum chamber resistance determination device can be a desktop computer, laptop, handheld computer, or cloud server, etc. The vacuum chamber resistance determination device may include, but is not limited to, a processor 31 and a memory 32. Those skilled in the art will understand that the vacuum chamber resistance determination device may also include input / output devices, network access devices, buses, etc.

[0063] The processor 31 can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor. The processor 31 is the control center of the vacuum chamber resistance determination device, connecting various parts of the device via various interfaces and lines.

[0064] The memory 32 can be used to store the computer program and / or modules. The processor 31 implements various functions of the vacuum chamber resistance determination device by running or executing the computer program and / or modules stored in the memory 32 and calling the data stored in the memory 32. The memory 32 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function, etc.; the data storage area may store data created based on the use of the mobile phone, etc. In addition, the memory 32 may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0065] If the module integrated into the vacuum chamber resistance determination device is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by the processor 31, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0066] This invention also provides a computer program product, including a computer program / instructions, which, when executed by a processor, implement the vacuum chamber resistance determination method as described in any of the above embodiments.

[0067] Compared with the prior art, the vacuum chamber resistance determination method, apparatus, device, storage medium, and product disclosed in this invention first acquire the mutual inductance submatrix of the set of circuits to be identified and the measured magnetic probe signal of the vacuum field; then, perform spectral embedding transformation on the mutual inductance submatrix to obtain a feature matrix, and cluster all circuits to be identified in the set of circuits to be identified based on the feature matrix to obtain several partitions; next, assign an independent resistance gain coefficient to each partition, and determine the simulated magnetic probe signal corresponding to the resistance gain coefficient; finally, with the optimization objective of minimizing the deviation between the measured magnetic probe signal of the vacuum field and the simulated magnetic probe signal, solve the resistance gain coefficient to obtain the optimal gain coefficient, and determine the initial identification resistance of the vacuum chamber based on the optimal gain coefficient. Therefore, the embodiments of the present invention obtain the feature matrix by performing spectral embedding transformation on the mutual inductance submatrix of the set of circuits to be identified. Based on the feature matrix, the electromagnetically tightly coupled circuits are classified into one class. Only one shared gain coefficient needs to be identified for each partition. The original high-dimensional resistance identification problem is reduced to a low-dimensional gain optimization problem. While ensuring physical rationality, the risk of overfitting is effectively suppressed, and the accuracy of vacuum chamber resistance identification is improved.

[0068] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for determining the resistance of a vacuum chamber, characterized in that, include: Obtain the prior resistance vector of the vacuum chamber circuit, the mutual inductance submatrix of the circuit set to be identified, and the measured magnetic probe signal of the vacuum field; Perform a spectral embedding transformation on the mutual inductance submatrix to obtain the feature matrix; Based on the feature matrix, all the circuits to be identified in the set of circuits to be identified are clustered to obtain several partitions; Each partition is assigned an independent resistance gain coefficient to be solved, and the simulated magnetic probe signal corresponding to the resistance gain coefficient to be solved is determined. With the optimization objective of minimizing the deviation between the measured magnetic probe signal in the vacuum field and the simulated magnetic probe signal, the gain coefficient of the resistor to be solved is obtained to obtain the optimal gain coefficient. The initial identification resistance of the vacuum chamber is obtained by correcting the prior resistance vector based on the optimal gain coefficient.

2. The method for determining the resistance of a vacuum chamber as described in claim 1, characterized in that, Also includes: Obtain the number of magnetic probes and the number of circuits to be identified; The number of partitions is determined by the minimum of the square root of the number of circuits to be identified and half the number of magnetic probes.

3. The method for determining the resistance of a vacuum chamber as described in claim 1, characterized in that, Also includes: Substitute the initial identification resistor into the pre-constructed rigid plasma radial-vertical motion model to determine the plasma vertical displacement growth rate corresponding to the initial identification resistor; Acquire the free drift data of the plasma vertical displacement, and determine the vertical displacement instability growth rate based on the free drift data of the plasma vertical displacement; The initial identification resistor is scaled by multiplying the ratio of the vertical displacement instability growth rate to the plasma vertical displacement growth rate to obtain the final identification resistor.

4. The method for determining the resistance of a vacuum chamber as described in claim 3, characterized in that, Also includes: Acquire multi-shot discharge data; Using the multi-shot discharge data, the steps of determining the plasma vertical displacement growth rate and the vertical displacement instability growth rate, as well as the scaling operation of the initial identification resistor, are repeated to verify the prediction accuracy of the final identification resistor.

5. The method for determining the resistance of a vacuum chamber as described in claim 1, characterized in that, Assigning an independent unsolved resistance gain coefficient to each partition and determining the simulated magnetic probe signal corresponding to the unsolved resistance gain coefficient includes: Obtain the self-inductance matrix of the vacuum chamber circuit, the mutual inductance matrix between the vacuum chamber circuit and the external coil, the coil current time series, and the Green's function from the vacuum chamber circuit to the magnetic probe; Assign an independent resistance gain coefficient to be solved for each partition; The resistance gain coefficient to be solved in each partition is assigned to all the circuits to be identified in the corresponding partition to construct a full-loop gain vector; wherein, the gain value corresponding to the non-identified circuit in the full-loop gain vector is a preset reference value. The identification resistance to be solved is determined based on the prior resistance vector and the full-loop gain vector. The simulated magnetic probe signal is determined based on the self-inductance matrix of the vacuum chamber circuit, the mutual inductance matrix between the vacuum chamber circuit and the external coil, the coil current time series, the identification resistor to be solved, and the Green's function.

6. The method for determining the resistance of a vacuum chamber as described in claim 5, characterized in that, The optimization objective is to minimize the deviation between the measured magnetic probe signal in the vacuum field and the simulated magnetic probe signal. The process involves solving for the gain coefficient of the resistor to be solved, yielding the optimal gain coefficient, including: The first optimization objective is to minimize the deviation between the measured magnetic probe signal and the simulated magnetic probe signal in the vacuum field, and the second optimization objective is to minimize the deviation of the unsolved resistance gain coefficient of each partition from the preset reference value. A dual-objective optimization problem is constructed. Under the preset feasible region constraint of gain coefficient, the bi-objective optimization problem is solved to obtain the optimal gain coefficient of each partition.

7. A device for determining the resistance of a vacuum chamber, characterized in that, include: The data acquisition module is used to acquire the prior resistance vector of the vacuum chamber circuit, the mutual inductance submatrix of the circuit set to be identified, and the measured magnetic probe signal of the vacuum field. The feature extraction module is used to perform spectral embedding transformation on the mutual inductance submatrix to obtain the feature matrix; The loop clustering module is used to cluster all the loops to be identified in the set of loops to be identified based on the feature matrix, and obtain several partitions. The signal simulation module is used to assign an independent resistance gain coefficient to be solved for each partition and to determine the simulated magnetic probe signal corresponding to the resistance gain coefficient to be solved. The gain calculation module is used to solve for the gain coefficient of the resistor to be solved with the optimization objective of minimizing the deviation between the measured magnetic probe signal of the vacuum field and the simulated magnetic probe signal, so as to obtain the optimal gain coefficient. The resistance determination module is used to correct the prior resistance vector based on the optimal gain coefficient to obtain the initial identification resistance of the vacuum chamber.

8. A device for determining the resistance of a vacuum chamber, characterized in that, It includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor, when executing the computer program, implements the vacuum chamber resistance determination method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device containing the computer-readable storage medium to perform the vacuum chamber resistance determination method as described in any one of claims 1 to 6.

10. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the vacuum chamber resistance determination method as described in any one of claims 1 to 6.

Images