Beam transport control system

Abstract

The present application relates to the technical field of beam transport control, and discloses a kind of beam transport control systems, comprising: analysis unit, correction unit, calculation unit, analysis unit, compensation unit and calibration unit, through analysis unit accurately capture superconducting magnet vibration characteristics and generate attenuation factor, vibration-voltage response matrix and space constraint matrix constructed in conjunction with correction unit, can effectively eliminate the zero drift generated by high-frequency vibration of quantum sensor, avoid frequent calibration, improve continuous working stability, simultaneously through three-dimensional current matrix decomposition magnetic field component, dynamic response surface analysis deviation, accurately process the nonlinear magnetic field distortion at the intersection of multiple magnets, solve the problem of insufficient prediction accuracy of deep learning algorithm, compensation and calibration unit linkage, realize real-time accurate compensation through phase optimization and energy constraint, substantially reduce lateral deviation error, ensure long-distance beam transmission accuracy, especially adapt to the continuous, high-precision control demand of medical radiotherapy.

Description

Technical Field

[0001] This invention relates to the field of beam transport control technology, and more specifically to a beam transport control system. Background Technology

[0002] Currently, in beam transportation, high-precision control is usually achieved by combining superconducting magnet arrays with AI-driven dynamic field distortion compensation technology. Specifically, a high-precision magnetic field is generated by embedding superconducting coils, and deep learning algorithms are used to predict beam path deviations in real time. Quantum sensor arrays are used to monitor spatial magnetic field fluctuations, and the magnet current is adjusted through a backpropagation algorithm to control the error of the beam trajectory and ensure that the beam maintains stable accuracy during long-distance transmission.

[0003] However, in medical radiotherapy, the above-mentioned control technology still has the following significant drawbacks: When working continuously, the quantum sensor will experience zero-point drift due to the vibration of the superconducting magnet (frequency > 100 Hz), which is particularly noticeable after continuous use. It needs to be manually calibrated after a period of time. At the same time, when the beam passes through multiple intersecting magnets, the nonlinear distortion formed by the superposition of magnetic fields is difficult to be accurately predicted by deep learning algorithms. The combined effect of these two factors leads to a decrease in the accuracy of beam compensation, such as an increase in lateral offset compensation error, resulting in a deterioration in control performance. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a beam transport control system that solves the aforementioned problems.

[0005] The above-mentioned technical objective of the present invention is achieved through the following technical solution:

[0006] A beam transport control system, comprising:

[0007] The analytical unit is used to acquire the mechanical vibration signal of the superconducting magnet array in real time, calculate the mechanical vibration signal, and obtain the mechanical vibration attenuation factor.

[0008] The correction unit is used to correct the drift of the original voltage of the quantum sensor array based on the mechanical vibration attenuation factor, and generate an anti-disturbance magnetic field strength value.

[0009] The computing unit is used to acquire the spatial coordinates and current direction of the superconducting magnet array in real time, and calculate the magnetic field coupling distortion intensity based on the spatial coordinates and current direction combined with the anti-disturbance magnetic field strength value.

[0010] The analysis unit is used to perform dynamic response analysis on the magnetic field coupling distortion intensity and generate a beam position offset estimate.

[0011] The compensation unit is used to calculate the estimated beam position offset to obtain the magnet current compensation amount;

[0012] The calibration unit is used to superimpose the magnet current compensation amount onto the driving current of the superconducting magnet array and calibrate the beam trajectory in real time.

[0013] Furthermore, the mechanical vibration signal is calculated to obtain the mechanical vibration attenuation factor, including:

[0014] The mechanical vibration signal of the superconducting magnet array is analyzed, and the characteristic frequency bands related to the resonance of the superconducting magnet array are screened out to generate vibration feature vectors.

[0015] Based on the vibration eigenvector, the relationship between vibration transmission time difference and amplitude attenuation among different superconducting magnets is analyzed to obtain the vibration transmission coefficient.

[0016] The temperature of the superconducting magnet array is acquired in real time. Based on the current direction and temperature of the superconducting magnet array, the vibration transmission coefficient is corrected to obtain the attribute correlation correction value.

[0017] The vibration characteristic vector is calculated to generate the initial attenuation coefficient;

[0018] Real-time environmental vibration data of the superconducting magnet array is acquired, and the initial attenuation coefficient is calibrated based on the real-time environmental vibration to obtain the mechanical vibration attenuation factor.

[0019] Furthermore, the original voltage of the quantum sensor array is drift-corrected based on the mechanical vibration attenuation factor to generate an anti-disturbance magnetic field strength value, including:

[0020] Based on the mechanical vibration attenuation factor, the mapping relationship between the voltage drift of the quantum sensor array and the vibration of the superconducting magnet array is analyzed, and a vibration-voltage response matrix is ​​generated.

[0021] The original voltage signal of the quantum sensor array is separated into multiple channels to obtain a vibration noise feature set;

[0022] Based on the vibration-voltage response matrix and vibration noise feature set, the dynamic drift coefficient of each quantum sensor channel is calculated;

[0023] Based on the voltage drift constraint condition between adjacent quantum sensors, the spatial constraint matrix is ​​obtained;

[0024] Based on the dynamic drift coefficient and spatial constraint matrix, the original voltage signal of the quantum sensor array is corrected to obtain the anti-disturbance magnetic field strength value.

[0025] Furthermore, based on the voltage drift constraint conditions between adjacent sensors, a spatial constraint matrix is ​​obtained, including:

[0026] By analyzing the real-time output signals of adjacent sensors in a quantum sensor array, the quantum entanglement correlation characteristics are obtained.

[0027] Based on the quantum entanglement correlation characteristics, voltage drift constraints between adjacent sensors are constructed, and a spatial constraint matrix is ​​obtained.

[0028] Furthermore, based on the spatial coordinates and current direction combined with the anti-disturbance magnetic field strength value, the magnetic field coupling distortion strength is calculated, including:

[0029] A three-dimensional current direction correlation matrix is ​​constructed based on the spatial coordinates and current direction of the superconducting magnet array.

[0030] The anti-disturbance magnetic field strength value is combined with the three-dimensional current direction correlation matrix and decomposed to generate unidirectional magnetic field components and orthogonal magnetic field components.

[0031] The vibration coupling distortion factor is obtained by analyzing the co-directional magnetic field components and the orthogonal magnetic field components and combining them with the mechanical vibration attenuation factor.

[0032] The vibration coupling distortion factor is optimized based on the attribute association correction value to generate the critical safety margin factor.

[0033] The magnetic field coupling distortion intensity is generated by analyzing the three-dimensional current direction correlation matrix and the critical safety margin factor.

[0034] Furthermore, dynamic response analysis is performed on the magnetic field coupling distortion intensity to generate a beam position shift estimate, including:

[0035] A dynamic distortion spectrum is generated by mapping the magnetic field coupling distortion intensity with the spatial coordinates and current direction of the superconducting magnet array.

[0036] Based on the dynamic distortion map, the velocity field and acceleration field of distortion propagation are determined, and the spatiotemporal propagation matrix is ​​generated.

[0037] The spatiotemporal propagation matrix is ​​constrained and optimized based on the spatial constraint matrix. The voltage drift correlation characteristics between adjacent quantum sensors are analyzed, and constraint correlation factors are generated.

[0038] Based on the dynamic distortion map, spatiotemporal propagation matrix, and constraint correlation factor, a dynamic response surface for beam position offset is constructed;

[0039] The dynamic response surface is analyzed to generate a beam position offset estimate.

[0040] Furthermore, the dynamic response surface is analyzed to generate a beam position offset estimate, including:

[0041] The distribution of extreme points and saddle points on the dynamic response surface is analyzed to determine the sensitive region of the influence of magnetic field distortion on the beam trajectory and generate the response sensitivity index.

[0042] The dynamic response surface is processed based on the response sensitivity index to generate a beam position offset estimate.

[0043] Furthermore, the estimated beam position offset is calculated to obtain the magnet current compensation amount, including:

[0044] By combining the anti-interference magnetic field strength, magnetic field coupling distortion strength, and real-time temperature of the superconducting magnet array, orthogonal interference characteristics are obtained.

[0045] The spatial coordinates and current direction of the superconducting magnet array are analyzed to generate a spatiotemporal correlation response matrix;

[0046] Based on the quantum entanglement correlation characteristics, the spatiotemporal correlation response matrix is ​​optimized and decomposed to generate a quantum constraint compensation sequence;

[0047] The magnet current compensation amount is obtained by calculating the orthogonal interference characteristics, the spatiotemporal correlation response matrix, and the quantum constraint compensation sequence.

[0048] Furthermore, the spatial coordinates and current direction of the superconducting magnet array are analyzed to generate a spatiotemporal correlation response matrix, including:

[0049] Based on the spatial coordinates and current direction of the superconducting magnet array, a four-dimensional tensor reflecting the spatiotemporal propagation characteristics of magnetic field disturbances is constructed.

[0050] The electromagnetic induction relationship between adjacent magnets in a superconducting magnet array is analyzed, and the four-dimensional tensor is reduced in dimension to obtain the spatiotemporal correlation response matrix.

[0051] Furthermore, the magnet current compensation is superimposed on the driving current of the superconducting magnet array, and the beam trajectory is calibrated in real time, including:

[0052] The quantum phase synchronization factor between the compensation current and the distorted magnetic field was obtained by analyzing the strength of the anti-disturbance magnetic field, the magnetic field coupling distortion strength, and the entanglement correlation characteristics.

[0053] The energy confinement threshold is determined based on the current direction, temperature, and beam position offset estimates of the superconducting magnet array.

[0054] Phase modulation of the spatiotemporal correlation response matrix is ​​performed based on the quantum phase synchronization factor to generate a phase-optimized response matrix;

[0055] The magnet current compensation amount is trimmed according to the energy constraint threshold to obtain the constrained current compensation amount.

[0056] Based on the phase-optimized response matrix and the constrained current compensation, a curvature correction tensor is constructed, and the beam trajectory is calibrated in real time using the curvature correction tensor.

[0057] In summary, the present invention has the following main beneficial effects:

[0058] By capturing the mechanical vibration signal of the superconducting magnet array in real time through the analytical unit, accurately selecting the resonant characteristic frequency band and generating the vibration characteristic vector, and combining dynamic parameters such as temperature and environmental vibration to correct the vibration transmission coefficient, the mechanical vibration attenuation factor is finally obtained. Furthermore, the correction unit constructs a vibration-voltage response matrix based on the mechanical vibration attenuation factor. Through multi-channel separation and spatial constraint matrix, the original voltage of the quantum sensor is corrected for drift, effectively eliminating zero-point drift caused by high-frequency vibration (frequency > 100 Hz). Compared with the traditional technology that relies on manual periodic calibration, this application achieves dynamic compensation for vibration interference, enabling the quantum sensor to maintain stable magnetic field measurement accuracy during continuous operation, avoiding the increase in beam control error caused by drift accumulation. Especially in long-term continuous irradiation scenarios in medical radiotherapy, it can reduce the frequency of manual intervention, ensure the real-time reliability of magnetic field monitoring data, and guarantee the precise control of beam transport.

[0059] By constructing a three-dimensional current direction correlation matrix, the anti-disturbance magnetic field strength value is decomposed into unidirectional and orthogonal magnetic field components by combining the matrix with the matrix. By combining the vibration coupling distortion factor and the critical safety margin factor, the magnetic field coupling distortion intensity is accurately quantified. Through dynamic response surface analysis and combined with quantum entanglement constraints, a beam position offset prediction value is generated, which overcomes the shortcomings of traditional deep learning algorithms in predicting nonlinear distortions. At the same time, based on the orthogonal interference characteristics, the spatiotemporal correlation response matrix and the quantum constraint compensation sequence, the current compensation amount is calculated, realizing targeted compensation for complex magnetic field distortions. This further reduces the beam lateral offset compensation error and maintains high-precision control even in the multi-magnet intersection region, meeting the high-precision positioning requirements of tumor target areas in medical radiotherapy and reducing the radiation dose to normal tissues.

[0060] By controlling the phase synchronization of the compensation current and the distorted magnetic field through quantum phase synchronization factors and energy constraint thresholds, and simultaneously calibrating the beam trajectory in real time based on curvature correction tensors, dynamic response optimization from magnetic field measurement to current adjustment is achieved. Compared with the lag of traditional backpropagation algorithms, this application shortens the response time of distortion identification and compensation through vibration-voltage mapping relationship and spatiotemporal propagation matrix, ensuring real-time calibration of the beam trajectory. In addition, the spatial constraint matrix and quantum constraint compensation sequence constructed through quantum entanglement correlation characteristics improve the system's robustness to environmental disturbances (such as temperature fluctuations and external vibrations). In medical radiotherapy scenarios, it can effectively cope with dynamic target area changes caused by human respiration and organ movement, and ensure accurate beam focusing by rapidly compensating for beam deviation. Attached Figure Description

[0061] Figure 1 This is a schematic diagram of the beam transport control system of the present invention. Detailed Implementation

[0062] 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.

[0063] refer to Figure 1 A beam transport control system, comprising:

[0064] The analytical unit is used to acquire the mechanical vibration signal of the superconducting magnet array in real time, calculate the mechanical vibration signal, and obtain the mechanical vibration attenuation factor.

[0065] The correction unit is used to correct the drift of the original voltage of the quantum sensor array based on the mechanical vibration attenuation factor, and generate an anti-disturbance magnetic field strength value.

[0066] The computing unit is used to acquire the spatial coordinates and current direction of the superconducting magnet array in real time, and calculate the magnetic field coupling distortion intensity based on the spatial coordinates and current direction combined with the anti-disturbance magnetic field strength value.

[0067] The analysis unit is used to perform dynamic response analysis on the magnetic field coupling distortion intensity and generate a beam position offset estimate.

[0068] The compensation unit is used to calculate the estimated beam position offset to obtain the magnet current compensation amount;

[0069] The calibration unit is used to superimpose the magnet current compensation amount onto the driving current of the superconducting magnet array and calibrate the beam trajectory in real time.

[0070] The system employs six interconnected units—analysis, correction, calculation, analysis, compensation, and calibration—to form a closed-loop control system. The analysis unit eliminates superconducting magnet vibration interference exceeding 100Hz in real time; the correction unit dynamically suppresses quantum sensor zero-point drift without manual calibration; the calculation and analysis unit accurately captures the nonlinear magnetic field distortion of multi-crossing magnets, overcoming the limitations of deep learning prediction; and the compensation and calibration unit achieves instantaneous response for current compensation and orbit calibration through quantum phase synchronization and curvature correction. Overall, this reduces the error of beam lateral offset during continuous operation, stably maintains high-precision beam transmission, and meets the needs of medical radiotherapy for uninterrupted and highly precise beam control.

[0071] In one embodiment, the mechanical vibration signal is calculated to obtain the mechanical vibration attenuation factor, including:

[0072] The mechanical vibration signal of the superconducting magnet array is analyzed to identify characteristic frequency bands related to the resonance of the superconducting magnet array and generate vibration feature vectors. Specifically, this involves: setting a threshold value for high-frequency interference in the mechanical vibration signal of the superconducting magnet array; using low-pass filtering to retain signals below this threshold value while filtering out unwanted interference at higher frequencies; then using Fourier transform to convert the time-varying vibration signal into a frequency-ordered signal to obtain the vibration energy value corresponding to each frequency; then, starting from the lowest frequency, the energy of each frequency is checked sequentially. When the energy of a frequency is higher than the average energy of its five adjacent frequencies on either side by a preset percentage (50%), that frequency and its consecutive adjacent frequencies that meet this condition are grouped into an interval. These intervals are the resonance-related characteristic frequency bands, and each interval is then extracted. The maximum vibration amplitude, average vibration amplitude, duration, and total number of occurrences are used as points in a multi-dimensional space. Each point represents a combination of parameters for a characteristic frequency band. The similarity (cosine similarity) between parameter points is calculated, and parameter points with high similarity are merged into clusters. A threshold is set (the average similarity of parameter points within a cluster must reach more than 80%) to select the main clusters that meet the criteria. For each main cluster, the parameter value corresponding to its center point is calculated as the representative value of the cluster. These representative values ​​are the key comprehensive parameters. Finally, these comprehensive parameters are sorted according to the importance of the clusters (the maximum vibration amplitude in each cluster is calculated, and the total number of occurrences of the corresponding vibration in the cluster is counted. The two are multiplied to obtain a comprehensive score, and the clusters with higher scores are more important) to form a vibration feature vector.

[0073] Based on vibration eigenvectors, the relationship between vibration transmission time difference and amplitude attenuation between different superconducting magnets is analyzed to obtain the vibration transmission coefficient. Specifically, this involves: for every two different magnets in the superconducting magnet array, one is designated as the source magnet and the other as the target magnet, forming a magnet pair. Vibration signals of the same resonant characteristic frequency band are extracted from each pair of magnets. The moment when the source magnet has the largest vibration amplitude in the resonant characteristic frequency band is found and recorded as the source peak moment. The moment when the target magnet has the largest vibration amplitude in the resonant characteristic frequency band is found and recorded as the target peak moment. The transmission time difference between the two magnets in the resonant characteristic frequency band is obtained by subtracting the source peak moment from the target peak moment. The ratio of the maximum vibration amplitude of the target magnet in the resonant characteristic frequency band to the maximum vibration amplitude of the source magnet in the resonant characteristic frequency band is calculated to obtain the amplitude attenuation ratio of the resonant characteristic frequency band. The amplitude attenuation ratio is divided by the transmission time difference to obtain the vibration transmission coefficient of the resonant characteristic frequency band. The average value of the vibration transmission coefficients of all resonant characteristic frequency bands is calculated and used as the vibration transmission coefficient of the pair of magnets.

[0074] The temperature of the superconducting magnet array is acquired in real time. Based on the current direction and temperature of the superconducting magnet array, the vibration transmission coefficient is corrected to obtain the attribute-related correction value. Specifically, the real-time temperature of the superconducting magnet array is collected at a frequency of 100Hz, and the average real-time temperature of 10 consecutive sampling points is taken as the current stable temperature T. The current direction of the superconducting magnet is marked as +1 for positive and -1 for negative. The reference temperature is set as T0, T0=4.2K, where the base value of the temperature correction coefficient is 1.0 (i.e., when T=T0, the temperature correction coefficient is 1.0). When T is higher than T0, the temperature correction coefficient increases by 0.02 for every 1K increase; when T is lower than T0, the temperature correction coefficient decreases by 0.015 for every 1K decrease. Specifically, the current correction coefficient is 1.05 for positive current and 0.95 for negative current. The temperature correction coefficient is multiplied by the current direction correction coefficient to obtain the comprehensive correction coefficient. The vibration transmission coefficient is multiplied by the comprehensive correction coefficient to obtain the attribute-related correction value.

[0075] The initial attenuation coefficient is generated by calculating the vibration feature vector. Specifically, this involves: assigning weights to each key comprehensive parameter in the vibration feature vector according to the proportion of the comprehensive score of the cluster (the comprehensive scores of all clusters are added together to get the total score, and the comprehensive score of each cluster is divided by the total score to get the proportion of the comprehensive score of that cluster. This proportion is used as the allocation coefficient of the key comprehensive parameter corresponding to that cluster. For example, if the comprehensive score of a cluster is 20 and the total score is 100, then the allocation coefficient of the parameter of that cluster is 0.2). Each cluster parameter is multiplied by its corresponding weight and then summed up. The sum is then divided by the total number of cluster parameters to get the base value. Finally, the base value is multiplied by the ratio of the average vibration amplitude to the maximum amplitude of each parameter to obtain the initial attenuation coefficient.

[0076] Real-time acquisition of environmental vibration data from the superconducting magnet array is used to calibrate the initial attenuation coefficient based on the real-time environmental vibration, resulting in a mechanical vibration attenuation factor. Specifically, this involves: for the acquired environmental vibration signals, first establishing a threshold value identical to the high-frequency interference threshold of the superconducting magnet; using low-pass filtering to retain signals below this threshold value; then converting the signals into a frequency-ordered form using Fourier transform to obtain the vibration energy corresponding to each frequency; starting from the lowest frequency, checking the energy of each frequency sequentially; when the energy of a frequency is 50% or more higher than the average energy of its five adjacent frequencies on either side, grouping that frequency and its consecutively matching adjacent frequencies into an interval, which are the characteristic frequency bands of the environmental vibration; comparing these environmental characteristic frequency bands with the previously determined superconducting magnet resonance-related characteristic frequency bands; and identifying signals corresponding to intervals that completely overlap or partially overlap, which are the overlapping signals to be retained; calculating the average amplitude of these overlapping signals and the total average amplitude of all environmental vibration signals; dividing the average amplitude by the total average amplitude to obtain the environmental influence ratio; and multiplying the initial attenuation coefficient by (1 minus the environmental influence ratio) to obtain the mechanical vibration attenuation factor.

[0077] By analyzing the vibration signal of the superconducting magnet, accurately selecting the resonant characteristic frequency band and generating the vibration characteristic vector, and combining temperature and current direction to dynamically correct the vibration transmission coefficient, and then integrating environmental vibration calibration to generate a mechanical vibration attenuation factor, the interference of high-frequency vibration >100Hz on the quantum sensor can be offset in real time. This effectively suppresses zero-point drift during continuous operation, avoids manual interval calibration, and ensures that the quantum sensor maintains high measurement accuracy during long-term use, providing stable and reliable magnetic field data for beam control.

[0078] By coupling calculations of spatial coordinates, current direction, and anti-disturbance magnetic field strength, the magnetic field coupling distortion intensity is directly generated, overcoming the prediction limitations of deep learning algorithms. Dynamic response analysis makes the beam position offset prediction more consistent with the actual magnetic field superposition law, reducing the prediction deviation caused by nonlinear distortion, significantly improving the lateral offset compensation effect, and enhancing the beam control stability.

[0079] In one embodiment, the original voltage of the quantum sensor array is drift-corrected based on the mechanical vibration attenuation factor to generate an anti-disturbance magnetic field strength value, including:

[0080] Based on the mechanical vibration attenuation factor, the mapping relationship between the voltage drift of the quantum sensor array and the vibration of the superconducting magnet array is analyzed, generating a vibration-voltage response matrix. Specifically, this involves: aligning the vibration signal of the superconducting magnet and the voltage signal of the quantum sensor by timestamp; filtering the two signals using a 10-point moving average method; shifting the sliding window by one sampling point each time to eliminate random noise; dividing the vibration signal into 20 frequency bands (each band width 10Hz); calculating the total energy of the vibration signal in each band and its percentage of the total energy of the entire vibration signal; combining multiple percentages to form a vibration vector; and dividing the voltage signal into 5ms intervals. The voltage signal is segmented (each segment overlapping by 2.5ms), and the root mean square (RMS) value of each segment is calculated. Multiple RMS values ​​are arranged to obtain the voltage feature sequence. The voltage signal time axis is moved within a ±50ms range in 1ms increments to create various time-delay scenarios. For each scenario, the correlation between the vibration vector and the voltage feature sequence is calculated: the changing directions of the vibration vector and the voltage feature sequence are compared point-by-point. If more than 7 out of 10 consecutive sampling points show the same changing direction, it is counted as a correlation point. The proportion of correlation points in each scenario to the total number of points is calculated, and this proportion is used as the correlation evaluation value. Scenarios with a correlation evaluation value greater than 0.7 are selected from multiple scenarios. In this scenario, the corresponding time delay is identified as the effective time delay point. For each effective time delay point, the difference in vibration energy ratio between two adjacent sampling points is divided by the value of the previous sampling point, and the resulting value is the vibration vector change rate. The difference in root mean square values ​​of two adjacent voltages is divided by the root mean square value of the previous sampling point, and the resulting value is the voltage characteristic sequence change rate. The vibration vector change rate of each frequency band is divided by the voltage characteristic sequence change rate to obtain the correlation coefficient between vibration and voltage. The correlation coefficients of the 20 frequency bands corresponding to each effective time delay point are arranged into a 20x20 matrix. A weighted average is then performed on the matrices obtained from multiple effective time delay points: first, the correlation coefficients of each effective time delay point are calculated... The square of the correlation assessment value at the time delay point is calculated, and then each element of each matrix is ​​multiplied by the corresponding squared correlation value at the time delay point. Finally, the elements at corresponding positions in all matrices are summed and divided by the number of time delay points to obtain a comprehensive 20×20 response coefficient matrix. The ratio of the standard deviation to the mean of each element over 5 consecutive sampling periods (coefficient of variation) is calculated. Elements with a coefficient of variation greater than 0.3 are removed, and the median of the remaining elements in the same column is used to fill the missing positions. Then, the elements in each row of the matrix are normalized, and all elements in each row are summed to obtain the total. Each element in each row is then divided by the total of the row to make the sum of the elements in each row equal to 1, thus forming the vibration-voltage response matrix.

[0081] The raw voltage signal of the quantum sensor array is subjected to multi-channel separation to obtain a vibration noise feature set. Specifically, this includes: applying a 10th-order Butterworth low-pass filter to the raw voltage signal of each channel of the quantum sensor array, with the cutoff frequency set to the high-frequency interference threshold (200Hz) of the superconducting magnet mechanical vibration signal to filter out high-frequency noise; smoothing the signal using a 5-point moving average method to eliminate random pulse interference; arranging the signals of each channel of the quantum sensor array in chronological order of acquisition time to form a two-dimensional data matrix, where rows represent channel numbers and columns represent time sampling points; setting a separation matrix with the same dimension as the number of channels; and iteratively optimizing the separation matrix: in each iteration, calculating the currently separated signal components... The kurtosis value (kurtosis is the ratio of the square of the fourth moment to the square of the second moment of the signal minus 3) is used. The maximum number of iterations is set to 200. Iteration stops when the change in kurtosis value for each component is less than 0.001 in five consecutive iterations. The resulting separation matrix is ​​multiplied by the two-dimensional data matrix to obtain multiple independent signal components, each representing a potential independent signal source. The cosine similarity between each independent signal component and the vibration eigenvector of the superconducting magnet is calculated. Independent components with a similarity greater than 0.7 are selected as candidate vibration noise components. Simultaneously, the power spectral density of each candidate vibration noise component is calculated. If the energy proportion of the power spectral density within the resonant characteristic frequency band of the superconducting magnet exceeds 60%, it is considered a valid vibration noise component; otherwise... These are not considered valid vibration noise components. For each valid vibration noise component, the bandwidth is divided into 20 frequency bands (covering 0-200Hz) with a 10Hz bandwidth. The characteristic parameters of each frequency band are calculated, including: band energy proportion, time-domain characteristics (root mean square value), frequency-domain characteristics (center frequency, bandwidth, spectral kurtosis), and time-frequency characteristics (rate of change of each parameter within 5 consecutive sampling periods). The characteristic parameters of all valid vibration noise components are arranged by channel number to form an initial feature matrix. Singular value decomposition (SVD) is performed on the initial feature matrix, and the singular vectors of the first 80% of the energy are retained. Performing SVD on the initial feature matrix yields three matrices: a left singular vector matrix, a singular value diagonal matrix (diagonal... The matrix consists of singular values ​​(sorted from largest to smallest) and a right singular vector matrix. The sum of squares of all singular values ​​is calculated and defined as the total energy of the matrix. The squares of individual singular values ​​are accumulated sequentially from largest to smallest. Accumulation stops when the sum reaches 80% of the total energy. The singular vectors (including the corresponding left or right singular vectors) corresponding to the singular values ​​involved in the accumulation process are the key singular vectors to be retained. The matrix is ​​then reconstructed to eliminate noise. The reconstructed feature matrix is ​​normalized row-wise so that the sum of elements in each row is 1. Each row then represents the standardized feature distribution of a vibration noise component. The standardized feature distributions of all vibration noise components are combined to form a vibration noise feature set.

[0082] Based on the vibration-voltage response matrix and vibration noise feature set, the dynamic drift coefficient of each quantum sensor channel is calculated. Specifically, for each quantum sensor channel, the standardized feature distribution of the corresponding row in its vibration noise feature set is taken and multiplied element-wise with the vibration-voltage response matrix. The product results are accumulated to obtain the initial drift value. For each sampling point, the absolute difference between the initial drift value and the actual voltage drift value is calculated. The average of the absolute differences of all sampling points is taken as the total deviation. The initial weight is set to 1.0, and the weight is adjusted successively in steps of 0.01. After each adjustment, the total deviation is recalculated. If the total deviation increases compared to the previous one, the step size is adjusted in the opposite direction. If the total deviation decreases, the original direction is maintained and the adjustment continues. The iteration continues until the total deviation is less than 0.01, and the weight value at this time is recorded. The average of the weight values ​​obtained from three consecutive iterations is calculated. This average value is the dynamic drift coefficient of the quantum sensor channel.

[0083] Based on the voltage drift constraint condition between adjacent quantum sensors, the spatial constraint matrix is ​​obtained;

[0084] Based on the dynamic drift coefficient and the spatial constraint matrix, the original voltage signal of the quantum sensor array is corrected to obtain the anti-disturbance magnetic field strength value. Specifically, this includes: for each quantum sensor channel, multiplying the dynamic drift coefficient of the channel by each eigenvalue of the corresponding row in the vibration noise feature set, and then summing all the products to obtain the total voltage drift of the channel; subtracting the total voltage drift from the original voltage signal of the channel to obtain the preliminary correction voltage; multiplying the preliminary correction voltage by the element of the corresponding row in the spatial constraint matrix point by point, and then summing all the products and dividing by the number of rows and columns to obtain the spatially constrained correction voltage; and multiplying the correction voltage by the voltage-magnetic field conversion coefficient (0.05) to obtain the anti-disturbance magnetic field strength value of the channel.

[0085] By constructing a vibration-voltage response matrix, the mapping relationship between the superconducting magnet's vibration (>100Hz) and the quantum sensor's voltage drift is accurately captured. Combined with the vibration noise feature set, interference signals are dynamically separated, and then optimized and corrected by a spatial constraint matrix. This allows for real-time cancellation of zero-point drift caused by vibration, eliminating the need for manual interval calibration. The voltage drift of the quantum sensor can be controlled after continuous operation. At the same time, the anti-interference magnetic field strength value after drift correction can effectively filter out the interference components caused by the superconducting magnet's vibration. The nonlinear distortion analysis based on this data is more consistent with the actual magnetic field superposition law, reducing the deviation between the estimated and measured values ​​of the beam position offset, improving the targeting of current compensation, reducing the lateral offset compensation deviation caused by magnetic field measurement errors, and enhancing the stability of beam control.

[0086] In one embodiment, based on the voltage drift constraint condition between adjacent sensors, a spatial constraint matrix is ​​obtained, including:

[0087] The quantum entanglement correlation characteristics are obtained by analyzing the real-time output signals of adjacent sensors in the quantum sensor array. Specifically, for each pair of adjacent sensors in the quantum sensor array, the real-time output signal is acquired, and within a 10ms time window, the proportion of the number of times the two signals change in the same direction is calculated. If the proportion exceeds 80%, it is recorded as strong correlation. The average of the strong correlation proportions over 20 consecutive windows is taken as the quantum entanglement correlation characteristics of the pair of quantum sensors.

[0088] Based on the quantum entanglement correlation characteristics, voltage drift constraints between adjacent sensors are constructed to obtain a spatial constraint matrix. Specifically, the number of rows and columns of the spatial constraint matrix are set to be equal to the total number of quantum sensors. Each position corresponds to a pair of sensors. In the row and column corresponding to each sensor, the average proportion of the quantum entanglement correlation characteristics of this pair of sensors (i.e., the average proportion of 20 consecutive windows) is filled in as the constraint strength of their voltage drift. The column positions corresponding to non-adjacent sensors are filled with 0. Then, the elements of each row of the matrix are processed: all non-zero elements in each row are added to obtain a sum. Each element in each row is divided by this sum to make the sum of the elements in each row equal to 1, thus completing the normalization. The final matrix is ​​the spatial constraint matrix.

[0089] By analyzing the quantum entanglement correlation characteristics of adjacent quantum sensors, a spatial constraint matrix is ​​constructed, which transforms the correlation of voltage drift of adjacent sensors into quantitative constraint conditions. This makes the drift correction of a single sensor no longer isolated, and with the help of the collaborative verification of adjacent sensors, the zero-point drift error caused by the vibration of the superconducting magnet >100Hz is further reduced. Especially after continuous operation, the magnetic field measurement deviation caused by the accumulation of drift of a single sensor can be reduced, and the spatial consistency of the anti-disturbance magnetic field strength value can be improved.

[0090] In one embodiment, the magnetic field coupling distortion intensity is calculated based on the spatial coordinates and current direction combined with the anti-interference magnetic field strength value, including:

[0091] Based on the spatial coordinates and current direction of the superconducting magnet array, a three-dimensional current direction correlation matrix is ​​constructed. Specifically, the number of rows, columns, and layers of the three-dimensional current direction correlation matrix are set to be equal to the total number of superconducting magnets. The positive current direction of each magnet is marked as 1, and the negative current direction is marked as -1. For any two magnets, the straight-line distance between their spatial coordinates is calculated. The distance coefficient is obtained by dividing 1 by the straight-line distance. The distance coefficient is multiplied by the current direction values ​​of the two magnets to obtain the product. The product is used as the element value of the corresponding position of the pair of magnets in the three-dimensional current direction correlation matrix, where the diagonal elements are set to 1. After all elements are calculated, the elements of each row and layer of the three-dimensional current direction correlation matrix are normalized so that the sum of the absolute values ​​of each row is 1, thus obtaining the three-dimensional current direction correlation matrix.

[0092] The anti-disturbance magnetic field strength value is combined with the three-dimensional current direction correlation matrix and decomposed to generate a co-directional magnetic field component and an orthogonal magnetic field component. The co-directional magnetic field component represents the effective magnetic field generated by the main current, and the orthogonal magnetic field component represents the distorted magnetic field caused by the interference. Specifically, the anti-disturbance magnetic field strength value is arranged into a vector according to the magnet number and multiplied element-wise with the three-dimensional current direction correlation matrix to obtain a new three-dimensional matrix. The elements at corresponding positions in each layer of the new three-dimensional matrix are added to obtain a two-dimensional intermediate matrix. In the intermediate matrix, elements with the same direction as the magnet current are selected and these elements are added to obtain the co-directional magnetic field component. The co-directional magnetic field component is subtracted from the anti-disturbance magnetic field strength value to obtain the orthogonal magnetic field component.

[0093] The vibration coupling distortion factor is obtained by analyzing the co-directional magnetic field component and the orthogonal magnetic field component and combining them with the mechanical vibration attenuation factor. Specifically, the ratio of the orthogonal magnetic field component to the co-directional magnetic field component is calculated, the ratio is multiplied by the mechanical vibration attenuation factor to obtain the preliminary distortion value, the average value of the preliminary distortion value of 10 consecutive sampling points is taken, the average value is divided by the maximum value of the co-directional magnetic field component, and the calculation result is normalized to the range of 0-1 to obtain the vibration coupling distortion factor.

[0094] The vibration coupling distortion factor is optimized based on the attribute association correction value to generate a critical safety margin factor. Specifically, the optimization process includes: multiplying the attribute association correction value by the vibration coupling distortion factor to obtain a preliminary optimized value; calculating the average value of the preliminary optimized values ​​from 5 consecutive sampling points; comparing the average value with a safety threshold (0.8); if the average value exceeds the threshold, subtracting the excess from the average value; if the average value does not exceed or is equal to the threshold, keeping the original value; and normalizing the average value after the above processing to make it fall within the range of 0-1 to obtain the critical safety margin factor.

[0095] The magnetic field coupling distortion intensity is generated by analyzing the three-dimensional current direction correlation matrix and the critical safety margin factor. Specifically, the following steps are taken: multiply each element of the three-dimensional current direction correlation matrix by the critical safety margin factor to obtain the coupling matrix; sum all elements of each layer in the three-dimensional coupling matrix to obtain a two-dimensional matrix; calculate the sum of the absolute values ​​of all elements in the two-dimensional matrix; divide the sum by the total number of superconducting magnets to obtain the average coupling value; and normalize the difference between the average coupling value and 1 to obtain the magnetic field coupling distortion intensity.

[0096] By constructing a three-dimensional current direction correlation matrix, the spatial coordinates of the superconducting magnet are quantitatively correlated with the current direction, accurately capturing the magnetic field coupling law of multiple intersecting magnets. By decomposing the unidirectional and orthogonal magnetic field components, the interfering distorted magnetic field is directly separated, improving the analytical accuracy of distortion characteristics under complex magnetic field environments. By integrating the mechanical vibration attenuation factor and the attribute correlation correction value, interferences such as superconducting magnet vibration and temperature drift are included in the distortion assessment. By dynamically adapting different working states through the critical safety margin factor, the normalized magnetic field coupling distortion intensity can accurately reflect the actual distortion degree after the superposition of multiple magnets. Especially in continuous working scenarios, it effectively weakens the synergistic interference of vibration and nonlinear superposition.

[0097] In one embodiment, dynamic response analysis is performed on the magnetic field coupling distortion intensity to generate a beam position offset estimate, including:

[0098] A dynamic distortion map is generated by mapping the magnetic field coupling distortion intensity to the spatial coordinates and current direction of the superconducting magnet array. Specifically, the following steps are taken: a three-dimensional mesh is established based on the spatial coordinates of the superconducting magnet array, with each mesh node corresponding to a magnet position. The magnetic field coupling distortion intensity is bound to the corresponding mesh node according to the magnet number. The positive direction of the current direction of the superconducting magnet is marked as 1 and the negative direction as -1, which is then used as an additional attribute of the corresponding magnet position node in the three-dimensional mesh. The four nearest magnet nodes around the blank node in the three-dimensional mesh are selected, and the inverse distance between each magnet node and the blank node is used as the weight. The magnetic field coupling distortion intensity of the four nodes is multiplied by the weight and then summed to obtain the magnetic field coupling distortion intensity of the blank node. The magnetic field coupling distortion intensity of all nodes is updated according to the time series. The corresponding color is assigned based on the different numerical ranges of the magnetic field coupling distortion intensity. The three-dimensional mesh is dynamically rendered using a color gradient (e.g., red to blue represents strong to weak) to form a dynamic distortion map.

[0099] Based on the dynamic distortion map, the velocity and acceleration fields of distortion propagation are determined, and a spatiotemporal propagation matrix is ​​generated. Specifically, this involves: selecting image frames from two adjacent moments at fixed time intervals from the dynamic distortion map; identifying the points with the highest magnetic field coupling distortion intensity in the first frame as peak points and marking each peak point; finding the peak points corresponding to these marked points in the second frame; calculating the straight-line distance each peak point moves between the two frames; dividing this distance by the time interval between the two frames to obtain the magnitude of the point's movement velocity; combining the velocity magnitude with the direction of movement to form a velocity vector; for blank nodes without magnets in the 3D mesh, finding the four nearest magnet nodes with peak points; assigning weights to the velocity vectors of these four nodes according to the inverse of the distance between the blank node and them (the larger the inverse distance, the greater the weight); summing the four weights to obtain the velocity vector of the blank node, thus filling the entire mesh. A complete velocity field is generated by dividing the grid into two grids. The velocity fields of two adjacent time points are selected. For the same node, the velocity vector of the later time point is subtracted from the velocity vector of the previous time point to obtain the velocity change. The velocity change is divided by the time interval between the two time points to obtain the acceleration vector of that node. The same calculation method is used to calculate the acceleration vector of blank nodes to generate an acceleration field. The number of rows in the spatiotemporal propagation matrix is ​​set to the total number of time steps multiplied by the total number of grid nodes, and the number of columns is equal to the total number of grid nodes. Each element corresponds to "a certain time point + a certain node". The velocity vector value and acceleration vector value of that grid node at that time are each weighted 50%, and summed to obtain the initial value of this element. Then, each row of the matrix is ​​processed by summing the values ​​of all elements in that row to obtain a total. The initial value of each element is then divided by this total to make the sum of the values ​​of the elements in each row equal to 1, thus obtaining the spatiotemporal propagation matrix.

[0100] The spatial constraint matrix is ​​used to constrain and optimize the spatiotemporal propagation matrix. The voltage drift correlation characteristics between adjacent quantum sensors are analyzed, and a constraint correlation factor is generated. Specifically, the spatial constraint matrix and the spatiotemporal propagation matrix are multiplied element by element at corresponding node positions to obtain a new matrix. For adjacent quantum sensors, the elements at corresponding positions in the new matrix after multiplication are extracted, and the average value of the element over 10 consecutive sampling periods is calculated. The average value is used as the basic value of the voltage drift correlation characteristics. The basic value of the voltage drift correlation characteristics is divided by the maximum difference between the two sensors in that time period (the voltage drift values ​​of the pair of sensors are recorded in 10 sampling periods, and the largest voltage drift value is taken as the maximum difference), and then multiplied by 0.8 (normalization coefficient) to make the result fall within the range of 0-1, thus obtaining the constraint correlation factor.

[0101] Based on the dynamic distortion map, spatiotemporal propagation matrix, and constraint correlation factor, a dynamic response surface for beam position offset is constructed. Specifically, the following steps are taken: the magnetic field coupling distortion intensity of the node in the dynamic distortion map is used as the base value; the base value is multiplied by the element value of the corresponding node in the spatiotemporal propagation matrix and then multiplied by the constraint correlation factor to obtain the beam position offset value of the node; for blank nodes in the dynamic distortion map, the four nearest nodes are found, and the offset values ​​of the four nodes are multiplied by the weights according to the reciprocal of the distance and then summed to obtain the offset value of the blank node; the offset values ​​of all nodes are updated according to the time series; and all offset values ​​are combined to form a dynamic response surface.

[0102] The dynamic response surface is analyzed to generate a beam position offset estimate.

[0103] The spatiotemporal distribution of magnetic field coupling distortion is visually presented through dynamic distortion maps. Combined with velocity and acceleration field analysis, the distortion propagation law is accurately captured. The spatiotemporal propagation matrix can quantify the dynamic characteristics of nonlinear distortion, enabling the beam position offset prediction to keep up with the dynamic changes of magnetic field distortion in real time. Especially in the region of multiple intersecting magnets, the prediction deviation is reduced. Furthermore, the voltage drift correlation characteristics of adjacent quantum sensors are used to generate constraint correlation factors, which further correct the dynamic response surface. This allows the beam position offset prediction to simultaneously adapt to the nonlinear characteristics of magnetic field distortion and the spatial correlation of sensors. Under continuous vibration environment, the prediction value is more consistent with the actual offset, reducing the compensation lag caused by inaccurate distortion prediction and ensuring the accuracy of beam control.

[0104] In one embodiment, the dynamic response surface is analyzed to generate a beam position offset estimate, including:

[0105] The distribution of extreme points and saddle points on the dynamic response surface is analyzed to determine the sensitive region of the influence of magnetic field distortion on the beam trajectory, and a response sensitivity index is generated. Specifically, this includes: for each node on the dynamic response surface, calculating the rate of change of its beam position offset value relative to its six adjacent nodes, and marking nodes with a rate of change of 0 as candidate points; selecting two adjacent nodes along each of the x, y, and z coordinate axes in the three-dimensional mesh (i.e., one on each side of each axis, for a total of six nodes); calculating the difference between the candidate point and these six nodes to obtain the first-order difference along the three axes; and calculating the difference between two adjacent first-order differences along each axis to obtain the response sensitivity index. The second-order difference along the x, y, and z axes is squared, summed, and the square root is taken to obtain the curvature value of the candidate point. If the curvature value is positive, it is an extreme point, indicating that the magnetic field coupling distortion intensity at that point has the strongest influence on the beam trajectory. If the curvature value is both positive and negative (i.e., the signs of the second-order difference along different axes are opposite), it is a saddle point, indicating that this is a critical node for distortion propagation. The dense areas of extreme points and saddle points are identified as sensitive areas. Finally, the magnetic field coupling distortion intensity of all nodes in the sensitive area is multiplied by the estimated beam position offset, and the sum is divided by the total number of nodes in the area and normalized to the range of 0-1 to obtain the response sensitivity index.

[0106] The dynamic response surface is processed based on the response sensitivity index to generate a beam position offset estimate. Specifically, this involves: setting a response sensitivity index threshold of 0.7; marking nodes with response sensitivity indices above the threshold as critical nodes and the rest as non-critical nodes; and processing the critical nodes by selecting the offset values ​​of the eight neighboring nodes around each critical node, assigning weights based on the square of the response sensitivity index of each node (the larger the square, the larger the weight); multiplying the beam position offset value of each neighboring node by its corresponding weight; summing these eight products to obtain a weighted offset value; and dividing the weighted offset value by the total weight. The weighted average of the eight adjacent nodes is obtained. The weighted average is multiplied by the reciprocal of the corresponding response sensitivity index to obtain the enhanced offset value. Non-critical nodes are processed by multiplying the beam position offset value of the non-critical node by (1 minus the corresponding response sensitivity index) to obtain the adjusted offset value. The enhanced offset value and adjusted offset value of all nodes are collected. The offset values ​​of the three most recent time steps are selected. For each node, these three values ​​are assigned weights in chronological order (the weight of the most recent time step is 0.5, the middle time step is 0.3, and the earliest time step is 0.2) and then weighted and summed to obtain the estimated beam position offset value for the next time step.

[0107] By analyzing the extreme points and saddle points of the dynamic response surface, sensitive areas where magnetic field distortion affects the beam trajectory are accurately located. Differential processing of critical and non-critical nodes is achieved using the response sensitivity index. The accuracy of offset calculation is enhanced for critical nodes, while the weight of non-critical nodes is appropriately weakened. Combined with time-series weighted optimization, the relevance and timeliness of beam position offset predictions are significantly improved. Especially in nonlinear distortion environments with multiple intersecting magnets, it can effectively capture instantaneous offsets caused by high-frequency vibrations, reducing lateral offset compensation errors.

[0108] In one embodiment, the estimated beam position offset is calculated to obtain the magnet current compensation amount, including:

[0109] The orthogonal interference characteristics are obtained by fusing the anti-interference magnetic field strength, magnetic field coupling distortion strength, and real-time temperature of the superconducting magnet array. Specifically, the anti-interference magnetic field strength, magnetic field coupling distortion strength, and real-time temperature are normalized to 0-1, and then multiplied by their respective weights according to their influence weights (anti-interference magnetic field strength is 0.4, magnetic field coupling distortion strength is 0.3, and real-time temperature is 0.3). The sum of these weights is the orthogonal interference characteristics.

[0110] The spatial coordinates and current direction of the superconducting magnet array are analyzed to generate a spatiotemporal correlation response matrix;

[0111] Based on the quantum entanglement correlation characteristics, the spatiotemporal correlation response matrix is ​​optimized and decomposed to generate a quantum constraint compensation sequence. Specifically, the spatiotemporal correlation response matrix and the quantum entanglement correlation characteristics are matched according to the corresponding positions of the quantum sensors. The cosine similarity between each feature vector in the matrix and the quantum entanglement correlation characteristics is calculated. Feature vectors with similarity greater than 0.6 are selected. The proportion of the vibration signal energy corresponding to each of these vectors to the total energy is used as the weight. These feature vectors are multiplied by their corresponding weights and then added together to obtain a new vector, which is the quantum constraint compensation sequence.

[0112] The magnet current compensation amount is obtained by calculating the orthogonal interference characteristics, the spatiotemporal correlation response matrix, and the quantum constraint compensation sequence. Specifically, this involves: multiplying the orthogonal interference characteristics element-wise by the spatiotemporal correlation response matrix to obtain the interference correction matrix; adding the interference correction matrix to the quantum constraint compensation sequence at corresponding positions to form a comprehensive adjustment matrix; summing the absolute values ​​of all elements in the comprehensive adjustment matrix and dividing by the total number of superconducting magnets to obtain the average adjustment value; and allocating compensation shares based on the average adjustment value according to the proportion of each element to the total (the larger the element value, the higher the allocation proportion). The upper and lower limits of the rated current of each superconducting magnet are used as the magnet safety... The preset values ​​are used to process the allocation of each element in the comprehensive adjustment matrix: each share is multiplied by the total rated current range of the corresponding magnet (i.e., the difference between the upper and lower limits, such as 50-(-50)=100A) to obtain the preliminary current compensation value; if the preliminary current compensation value exceeds the upper limit of the preset value, the upper limit value is taken as the preliminary current compensation value; if the preliminary current compensation value is lower than the lower limit of the preset value, the lower limit value is taken as the preliminary current compensation value; if it is within the range, the original value of the preliminary current compensation value is retained; multiple preliminary current compensation values ​​are arranged in order according to the magnet number to obtain a set of compensation amounts, which are the magnet current compensation amounts.

[0113] By integrating the anti-disturbance magnetic field strength, magnetic field coupling distortion strength, and real-time temperature to generate orthogonal interference characteristics, this method comprehensively integrates multi-dimensional interference factors, avoiding the limitations of single-parameter analysis. By combining the spatiotemporal correlation response matrix and quantum constraint compensation sequence, the quantum entanglement characteristics are transformed into quantitative constraints for current compensation, making the compensation calculation more consistent with the nonlinear distortion law of the magnetic field and significantly improving the targeting of compensation. Especially in high-frequency vibration environments, it can accurately offset the current deviation caused by orthogonal interference.

[0114] By controlling the safety threshold, the magnet current compensation is ensured to be within the rated range, which avoids secondary magnetic field distortion caused by overcompensation. The compensation share is allocated proportionally to achieve fine adjustment. The compensation sequence optimized by quantum constraint can quickly respond to the predicted changes in beam position offset, shorten the current adjustment lag time, effectively reduce the lateral offset compensation error, and ensure the continuous stability of the beam trajectory in medical radiotherapy.

[0115] In one embodiment, the spatial coordinates and current direction of the superconducting magnet array are analyzed to generate a spatiotemporal correlation response matrix, including:

[0116] Based on the spatial coordinates and current direction of a superconducting magnet array, a four-dimensional tensor reflecting the spatiotemporal propagation characteristics of magnetic field disturbances is constructed. Specifically, this involves: setting the dimensions of the four-dimensional tensor to x, y, z (space) and t (time); calculating the spatial components and the time-varying rate of change of the current of each magnet at all spatiotemporal points (x, y, z, t); calculating the linear distance between any two magnets; dividing 1 by the linear distance to obtain a distance coefficient; multiplying the distance coefficient by the current direction values ​​of the two magnets (1 for positive, -1 for negative) to obtain the spatial components; subtracting the previous value from the current value at adjacent moments and dividing by the time interval to obtain the time-varying rate of change of the current of each magnet; arranging the product of the spatial components and the time-varying rate of change of the current of each magnet at all spatiotemporal points (x, y, z, t) according to the x, y, z, t dimensions to form an initial four-dimensional data array; setting the decomposition rank (to 1 / 3 of the total number of magnets); and then... The decomposition is divided into four factor matrices, corresponding to the x, y, z, and t dimensions respectively. The row index of each matrix is ​​the coordinate / time point of the corresponding dimension, and the column index is the decomposition rank. Iterative optimization is performed using alternating least squares: the y, z, and t factor matrices are fixed, and the x factor matrix is ​​updated by minimizing the squared error between the initial array and the reconstructed value; then the x, z, and t factor matrices are fixed again, and the y factor matrix is ​​updated, and the z and t factor matrices are updated alternately until the error change of 5 consecutive iterations is less than 0.001. Finally, the column vectors corresponding to the x, y, z, and t coordinates in the four factor matrices are taken, and the length of each vector is the same as the decomposition rank. The elements at the same index position in the four vectors are multiplied respectively to obtain a product value with the same number of decomposition ranks. These product values ​​are added together, and the sum is the magnetic field perturbation value of the spatiotemporal point. All spatiotemporal points are traversed in this way, and then these magnetic field perturbation values ​​are arranged in the order of x, y, z spatial dimensions and t time dimension to form a four-dimensional tensor.

[0117] The electromagnetic induction relationship between adjacent magnets in a superconducting magnet array is analyzed, and the four-dimensional tensor is reduced in dimension to obtain the spatiotemporal correlation response matrix. Specifically, for adjacent magnets in the superconducting magnet array, the rate of change of induced current of each adjacent magnet pair (the current value of the next moment minus the current value of the previous moment, and then divided by the time interval) is multiplied by the distance coefficient to obtain the electromagnetic induction coefficient of the adjacent magnet pair. The electromagnetic induction coefficient is used as a weight to perform a weighted average on the time dimension of the four-dimensional tensor: the three-dimensional spatial data of each time point is multiplied by the corresponding weight, the product results of all time points are added together, and then divided by the sum of the weights to obtain a three-dimensional array after merging the time dimension. The three-dimensional array is arranged into rows and columns of a two-dimensional matrix in the order of x-coordinate from smallest to largest, y-coordinate from smallest to largest under the same x-coordinate, and z-coordinate from smallest to largest under the same y-coordinate, thus obtaining the spatiotemporal correlation response matrix.

[0118] By constructing a four-dimensional tensor to accurately characterize the spatiotemporal propagation characteristics of magnetic field disturbances, the spatial coordinates of superconducting magnets, current direction, and time rate of change are quantitatively correlated, ensuring the analytical accuracy of complex dynamic changes in magnetic fields. At the same time, the four-dimensional tensor is reduced in dimensionality based on the electromagnetic induction coefficients of adjacent magnets, and the generated spatiotemporal correlation response matrix retains the key features of magnetic field disturbances, making it easier to capture the spatiotemporal correlation law of nonlinear distortion of magnetic fields and improving the accuracy of compensation.

[0119] In one embodiment, the magnet current compensation is superimposed on the driving current of the superconducting magnet array, and the beam trajectory is calibrated in real time, including:

[0120] The analysis of the anti-disturbance magnetic field strength, magnetic field coupling distortion strength, and entanglement correlation characteristics yields the quantum phase synchronization factor between the compensation current and the distorted magnetic field. Specifically, this involves: normalizing the anti-disturbance magnetic field strength, magnetic field coupling distortion strength, and entanglement correlation characteristics; calculating the linear correlation between the anti-disturbance magnetic field strength and the magnetic field coupling distortion strength: taking 50 consecutive sampling points for each, calculating the sum of the products of their deviations from their respective average values, and then dividing by the product of their standard deviations and the number of sampling points to obtain the correlation. This correlation is then normalized to 0-1 to obtain the linear correlation between the anti-disturbance magnetic field strength and the magnetic field coupling distortion strength; calculating the matching degree between the linear correlation and the normalized entanglement correlation characteristics: subtracting the absolute value of the difference between the two from 1, and normalizing the result to 0-1 to obtain the matching degree; calculating the difference between the compensation current phase and the distorted magnetic field phase, taking its absolute value, and dividing it by 1 to obtain the weight (if the absolute value is 0, the weight is 1); and multiplying the matching degree by the weight to obtain the quantum phase synchronization factor.

[0121] Based on the current direction, temperature, and beam position offset prediction of the superconducting magnet array, the energy constraint threshold is determined. Specifically, this includes: normalizing the current direction (positive (+1) and negative (-1) to 0-1 (positive direction is 1, negative direction is 0); using 4.2K as the reference value, the difference between the current temperature and the reference value is divided by the maximum temperature difference and normalized to 0-1; normalizing the beam position offset prediction to 0-1; setting the weight of the current direction to 0.3, the weight of the temperature to 0.2, and the weight of the beam position offset prediction to 0.5; multiplying the normalized current direction, temperature, and beam position offset prediction by their respective weights and then adding them together; and normalizing the sum to (0-1) to obtain the energy constraint threshold.

[0122] Phase modulation of the spatiotemporal correlation response matrix based on the quantum phase synchronization factor generates a phase-optimized response matrix. Specifically, this involves multiplying each element of the spatiotemporal correlation response matrix by the quantum phase synchronization factor to obtain a preliminary modulation matrix, summing the elements of each row of the preliminary modulation matrix, dividing each element of each row by the sum of the elements of that row to make the sum of the elements of each row equal to 1, and then normalizing the matrix. The normalized matrix is ​​the phase-optimized response matrix.

[0123] The current compensation amount of the magnets is trimmed according to the energy constraint threshold to obtain the constrained current compensation amount. Specifically, this includes: multiplying the current compensation amount of each magnet by the energy constraint threshold to obtain the preliminary trimming value of each magnet; setting an upper limit and a lower limit for the rated current of each magnet (upper limit 50A, lower limit -50A); if the preliminary trimming value is higher than the upper limit, the upper limit value is taken as the preliminary trimming value of the magnet; if it is lower than the lower limit, the lower limit value is taken as the preliminary trimming value of the magnet; if it is between the upper and lower limits, the preliminary trimming value is retained; and the preliminary trimming values ​​of all magnets are arranged in order of their numbers to obtain the constrained current compensation amount.

[0124] Based on the phase-optimized response matrix and the constrained current compensation, a curvature correction tensor is constructed, and the beam trajectory is calibrated in real time using the curvature correction tensor. Specifically, this involves: multiplying the phase-optimized response matrix and the constrained current compensation by magnet number, element by element, to obtain an intermediate matrix; calculating the average value of all elements in each row of the intermediate matrix, using this average value as the tangential curvature correction coefficient for the corresponding position in that row; calculating the standard deviation of the elements in that row, using this as the normal curvature correction coefficient; combining the tangential and normal curvature correction coefficients to form the curvature correction tensor; and using the tangential and normal curvature correction coefficients of each trajectory point in the curvature correction tensor to adjust the driving current of the corresponding superconducting magnet, thereby correcting the curvature of the beam trajectory in real time. Specifically, for each trajectory... The calibration process involves identifying the corresponding superconducting magnet. When the tangential curvature correction factor is positive, the driving current of the magnet is increased to reduce the curvature of the track in the tangential direction. When the tangential curvature correction factor is negative, the driving current of the magnet is decreased to increase the curvature of the track in the tangential direction. When the normal curvature correction factor is positive, the driving current of the magnet is finely adjusted to the side perpendicular to the track tangent, causing the track to shift towards that side to correct the deviation. When the normal curvature correction factor is negative, the current is finely adjusted to the opposite side. After each adjustment, the actual curvature change of the beam track is monitored in real time. If the actual curvature change does not meet the requirements, the above adjustment process is repeated according to the new tangential and normal curvature correction factors until the track curvature meets the requirements, thus completing the calibration.

[0125] Precise phase matching between the compensation current and the distorted magnetic field is achieved through quantum phase synchronization factors. Combined with the energy constraint threshold for dynamic trimming of the compensation amount, it ensures that the superimposed driving current can both offset the magnetic field distortion caused by high-frequency vibration and not exceed the safe operating range of the magnet. The phase-optimized response matrix strengthens the spatiotemporal correlation between current adjustment and magnetic field change, enabling the superconducting magnet to output a stable magnetic field even in a vibration environment >100Hz, avoiding compensation failure caused by phase mismatch. Furthermore, the orbit curvature and offset direction are controlled by tangential and normal coefficients respectively, achieving an instantaneous response to nonlinear distortion. Compared with traditional technologies, this shortens the lag time of beam orbit calibration, maintains long-term accuracy without manual intervention, and perfectly meets the requirements of beam stability in medical radiotherapy.

[0126] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A beam transport control system, characterized in that, include: The analysis unit is used to acquire the mechanical vibration signal of the superconducting magnet array in real time, calculate the mechanical vibration attenuation factor from the mechanical vibration signal, including: The mechanical vibration signal of the superconducting magnet array is analyzed, and the characteristic frequency bands related to the resonance of the superconducting magnet array are screened out to generate vibration feature vectors. Based on the vibration eigenvector, the relationship between vibration transmission time difference and amplitude attenuation among different superconducting magnets is analyzed to obtain the vibration transmission coefficient. The temperature of the superconducting magnet array is acquired in real time. Based on the current direction and temperature of the superconducting magnet array, the vibration transmission coefficient is corrected to obtain the attribute correlation correction value. The vibration characteristic vector is calculated to generate the initial attenuation coefficient; Real-time acquisition of environmental vibration data of superconducting magnet array; calibration of initial attenuation coefficient based on real-time environmental vibration; and deriving mechanical vibration attenuation factor. The correction unit is used to correct the drift of the original voltage of the quantum sensor array based on the mechanical vibration attenuation factor, and to generate an anti-disturbance magnetic field strength value, including: Based on the mechanical vibration attenuation factor, the mapping relationship between the voltage drift of the quantum sensor array and the vibration of the superconducting magnet array is analyzed, and a vibration-voltage response matrix is ​​generated. The original voltage signal of the quantum sensor array is separated into multiple channels to obtain a vibration noise feature set; Based on the vibration-voltage response matrix and vibration noise feature set, the dynamic drift coefficient of each quantum sensor channel is calculated; Based on the voltage drift constraint condition between adjacent quantum sensors, the spatial constraint matrix is ​​obtained; Based on the dynamic drift coefficient and spatial constraint matrix, the original voltage signal of the quantum sensor array is corrected to obtain the anti-disturbance magnetic field strength value; The computing unit is used to acquire the spatial coordinates and current direction of the superconducting magnet array in real time, and calculate the magnetic field coupling distortion intensity based on the spatial coordinates and current direction combined with the anti-disturbance magnetic field strength value. The analysis unit is used to perform dynamic response analysis on the magnetic field coupling distortion intensity and generate a beam position offset estimate. The compensation unit is used to calculate the estimated beam position offset to obtain the magnet current compensation amount; The calibration unit is used to superimpose the magnet current compensation amount onto the driving current of the superconducting magnet array and calibrate the beam trajectory in real time.

2. The beam transport control system according to claim 1, characterized in that, Based on the voltage drift constraint between adjacent sensors, the spatial constraint matrix is ​​obtained, including: By analyzing the real-time output signals of adjacent sensors in a quantum sensor array, the quantum entanglement correlation characteristics are obtained. Based on the quantum entanglement correlation characteristics, voltage drift constraints between adjacent sensors are constructed, and a spatial constraint matrix is ​​obtained.

3. The beam transport control system according to claim 1, characterized in that, Based on the spatial coordinates and current direction combined with the anti-interference magnetic field strength value, the magnetic field coupling distortion strength is calculated, including: A three-dimensional current direction correlation matrix is ​​constructed based on the spatial coordinates and current direction of the superconducting magnet array. The anti-disturbance magnetic field strength value is combined with the three-dimensional current direction correlation matrix and decomposed to generate unidirectional magnetic field components and orthogonal magnetic field components. The vibration coupling distortion factor is obtained by analyzing the co-directional magnetic field components and the orthogonal magnetic field components and combining them with the mechanical vibration attenuation factor. The vibration coupling distortion factor is optimized based on the attribute association correction value to generate the critical safety margin factor. The magnetic field coupling distortion intensity is generated by analyzing the three-dimensional current direction correlation matrix and the critical safety margin factor.

4. A beam transport control system according to claim 3, characterized in that, Dynamic response analysis of the magnetic field coupling distortion intensity is performed to generate a beam position shift estimate, including: A dynamic distortion spectrum is generated by mapping the magnetic field coupling distortion intensity with the spatial coordinates and current direction of the superconducting magnet array. Based on the dynamic distortion map, the velocity field and acceleration field of distortion propagation are determined, and the spatiotemporal propagation matrix is ​​generated. The spatiotemporal propagation matrix is ​​constrained and optimized based on the spatial constraint matrix. The voltage drift correlation characteristics between adjacent quantum sensors are analyzed, and constraint correlation factors are generated. Based on the dynamic distortion map, spatiotemporal propagation matrix, and constraint correlation factor, a dynamic response surface for beam position offset is constructed; The dynamic response surface is analyzed to generate a beam position offset estimate.

5. A beam transport control system according to claim 4, characterized in that, The dynamic response surface is analyzed to generate a beam position offset estimate, including: The distribution of extreme points and saddle points on the dynamic response surface is analyzed to determine the sensitive region of the influence of magnetic field distortion on the beam trajectory and generate the response sensitivity index. The dynamic response surface is processed based on the response sensitivity index to generate a beam position offset estimate.

6. A beam transport control system according to claim 5, characterized in that, The estimated beam position offset is calculated to obtain the magnet current compensation amount, including: By combining the anti-interference magnetic field strength, magnetic field coupling distortion strength, and real-time temperature of the superconducting magnet array, orthogonal interference characteristics are obtained. The spatial coordinates and current direction of the superconducting magnet array are analyzed to generate a spatiotemporal correlation response matrix; Based on the quantum entanglement correlation characteristics, the spatiotemporal correlation response matrix is ​​optimized and decomposed to generate a quantum constraint compensation sequence; The magnet current compensation amount is obtained by calculating the orthogonal interference characteristics, the spatiotemporal correlation response matrix, and the quantum constraint compensation sequence.

7. A beam transport control system according to claim 6, characterized in that, The spatial coordinates and current direction of the superconducting magnet array are analyzed to generate a spatiotemporal correlation response matrix, including: Based on the spatial coordinates and current direction of the superconducting magnet array, a four-dimensional tensor reflecting the spatiotemporal propagation characteristics of magnetic field disturbances is constructed. The electromagnetic induction relationship between adjacent magnets in a superconducting magnet array is analyzed, and the four-dimensional tensor is reduced in dimension to obtain the spatiotemporal correlation response matrix.

8. A beam transport control system according to claim 7, characterized in that, The magnet current compensation is superimposed on the driving current of the superconducting magnet array, and the beam trajectory is calibrated in real time, including: The quantum phase synchronization factor between the compensation current and the distorted magnetic field was obtained by analyzing the strength of the anti-disturbance magnetic field, the magnetic field coupling distortion strength, and the entanglement correlation characteristics. The energy confinement threshold is determined based on the current direction, temperature, and beam position offset estimates of the superconducting magnet array. Phase modulation of the spatiotemporal correlation response matrix is ​​performed based on the quantum phase synchronization factor to generate a phase-optimized response matrix; The magnet current compensation amount is trimmed according to the energy constraint threshold to obtain the constrained current compensation amount. Based on the phase-optimized response matrix and the constrained current compensation, a curvature correction tensor is constructed, and the beam trajectory is calibrated in real time using the curvature correction tensor.

Images