Error calibration method for radio frequency therapy equipment under complex magnetic field interference

Abstract

The application discloses a radio physiotherapy equipment error calibration method under complex magnetic field interference, comprising the following steps: collecting magnetic field time sequence data and multi-source electromagnetic interference signals when the equipment is running, obtaining a feature set through magnetic field time sequence distortion feature extraction, splitting different interference source signals by using multi-source electromagnetic interference separation and decoupling technology, establishing a frequency offset calibration model through an equipment frequency offset intelligent convergence calibration algorithm, generating a calibration parameter set in combination with a public parameter extraction and a localization parameter adaptation mechanism, and completing dynamic adjustment of equipment parameters through a physiotherapy electric wave error traceability analysis platform. The method solves the problems of insufficient adaptability and lack of dynamic adaptation capability of traditional calibration methods under complex magnetic field interference, improves calibration accuracy and stability, does not need to rely on extensive interference suppression means, adapts to individual differences of different equipment and dynamic changes of interference, and provides a reliable error calibration solution for radio physiotherapy equipment.

Description

Technical Field

[0001] This invention relates to the field of radio therapy equipment calibration technology, and in particular to a method for calibrating errors in radio therapy equipment under complex magnetic field interference. Background Technology

[0002] Radio-guided physiotherapy equipment is widely used in clinical treatment and rehabilitation, and its operational accuracy directly affects treatment efficacy and safety. However, various electromagnetic interference sources exist in the operating environment of these devices, creating complex magnetic field interference scenarios. Magnetic field timing distortions and the superposition of multi-source electromagnetic interference can lead to deviations in the frequency, phase, and amplitude of the output radio waves, affecting the accuracy of physiotherapy parameters. Therefore, targeted error calibration technology is urgently needed to solve this problem. Complex magnetic field interference is characterized by diverse sources, complex propagation paths, and variable timing characteristics. Traditional calibration techniques struggle to achieve accurate interference separation and error tracing, becoming a key bottleneck restricting the performance improvement of radio-guided physiotherapy equipment.

[0003] Currently, error calibration for radio-controlled physiotherapy equipment often employs a combination of a single interference suppression strategy and a fixed-parameter calibration mode. This involves reducing external interference through general electromagnetic shielding devices and using preset calibration parameter tables or simple linear calibration algorithms to statically adjust parameters such as the frequency and phase of the equipment's output radio waves. Some technologies collect partial magnetic field data during equipment operation, remove significant interference signals through basic signal filtering, and then determine the error range based on empirical thresholds before performing calibration. Overall, this approach relies on fixed calibration logic and limited interference handling methods, failing to establish a systematic mechanism for interference separation, feature extraction, and dynamic calibration.

[0004] The existing technology has two main drawbacks: First, it is not adaptable to complex magnetic field interference, fails to effectively distinguish electromagnetic interference signals from different sources, and cannot accurately capture the core characteristics of magnetic field temporal distortion. Simple filtering or shielding is insufficient to completely eliminate the errors caused by the superposition of multiple interference sources, resulting in limited calibration accuracy. Second, the calibration process lacks dynamic adaptability. It uses fixed parameters or static calibration models, which cannot adjust the calibration strategy according to the real-time changes in magnetic field interference and individual differences of equipment. Furthermore, it has not established a sound error tracing and parameter adaptation mechanism, making it difficult to achieve accurate compensation for equipment errors in complex scenarios, resulting in poor stability of calibration results. Summary of the Invention

[0005] In order to overcome the shortcomings and deficiencies of the existing technology, the present invention provides a method for calibrating errors in radio therapy equipment under complex magnetic field interference.

[0006] The technical solution adopted in this invention is a method for calibrating errors in radio-controlled physiotherapy equipment under complex magnetic field interference, comprising the following steps: S1, collecting magnetic field time-series data and multi-source electromagnetic interference signals during the operation of the radio-controlled physiotherapy equipment through a physiotherapy wave error tracing and analysis platform, and simultaneously recording the frequency, phase, and amplitude correlation data of the equipment's output waves; S2, using a magnetic field time-series distortion feature extraction algorithm to identify distortion features in the collected magnetic field time-series data, and obtaining a magnetic field time-series distortion feature set through time-series correlation analysis, distortion extreme point location, and feature vector construction; S3, using a multi-source electromagnetic interference separation and decoupling algorithm to separate and process the multi-source electromagnetic interference signals, based on... S4. Based on the differences in the spectral characteristics and propagation paths of the interference signals, different interference source signals are decoupled and separated; S5. The frequency offset of the output radio waves of the radio therapy equipment is dynamically calculated using the intelligent convergence calibration algorithm of the equipment frequency offset, and a frequency offset calibration model is established by combining the magnetic field timing distortion feature set and the decoupled interference signal; S6. Through the common parameter extraction and local parameter adaptation mechanism, the feature data, decoupled signals and frequency offset calibration model parameters obtained in S2-S4 are integrated to generate the equipment error calibration parameter set; S7. Based on the radio therapy wave error tracing analysis platform, the calibration parameter set is imported into the radio therapy equipment control system to complete the dynamic adjustment of the equipment operating parameters and error calibration.

[0007] Furthermore, the expression for the magnetic field temporal distortion feature extraction algorithm is as follows: ,in The time-series correlation weighting coefficient. This is the original magnetic field time series data. For distorted reference time series data, The length of the feature extraction time window. For time delay variables, This is the current observation time; The formula for constructing the distorted feature vector is calculated as follows: ,in, This represents the maximum value of the distortion characteristic function. This represents the minimum value of the distortion characteristic function. The integral value of the distortion characteristic function. The variance of the distortion characteristic function. This is the distorted feature vector.

[0008] Furthermore, the expression for the multi-source electromagnetic interference separation and decoupling algorithm is as follows: ,in The separated single-source interference signal matrix, For interference separation weight matrix, The collected mixed interference signal matrix; The decoupling optimization formula is calculated as follows: ,in The optimized interference separation weight matrix, For the first Information entropy of the path-separated signal For the number of interference sources, The regularization coefficient is . Let L be the L2 norm of the weight matrix.

[0009] Furthermore, the expression for the device frequency offset intelligent convergence calibration algorithm is as follows: ,in This is the frequency offset calibration value. To calibrate the step size factor, Let the frequency offset error be the objective function. For the error model parameter vector, Let be the partial derivative of the objective function with respect to frequency. The historical calibration weighting coefficient. This is the frequency offset calibration value from the previous moment. The convergence decay coefficient is... To calibrate the iteration time.

[0010] Furthermore, the physiotherapy electro-wave error tracing and analysis platform includes an FPGA processing module, a multi-channel high-speed data acquisition module, a millimeter-wave magnetic field sensing array, a real-time signal processing module, and a calibration parameter storage module. The FPGA processing module uses a 40nm Cyclone V series chip and integrates a hardware acceleration engine for parallel algorithm computation. The multi-channel high-speed data acquisition module has a sampling rate of 1.25GSps and a quantization bit depth of 16 bits, supporting 8-channel synchronous acquisition. The millimeter-wave magnetic field sensing array consists of 16 sensing units arranged in a tetrahedral pattern, with a sensing unit detection bandwidth covering 30GHz-300GHz. The real-time signal processing module has a built-in embedded Linux system with a custom signal processing kernel, and the data processing latency is controlled within 50μs. The calibration parameter storage module uses an NVMe interface solid-state drive with a storage capacity of no less than 1TB, supporting real-time writing and fast reading of calibration parameters.

[0011] Further, S2 includes the following sub-steps: S21, performing time-series segmentation processing on the collected magnetic field time-series data, dividing the continuous time-series data into several data segments according to a fixed time interval, each data segment including the same number of sampling points, and using sliding window technology to achieve non-overlapping coverage of the data segments; S22, calculating the time-series correlation coefficient of each data segment, and initially screening out data segments with distortion characteristics by comparing the amplitude change trend and phase synchronization of adjacent data segments; S23, locating distortion extrema points, processing the screened data segments using a difference algorithm, determining the location of extrema points by the sign change of the difference between adjacent sampling points, and statistically analyzing the number and distribution density of extrema points within each data segment; S24, constructing a distortion feature vector, extracting the extrema point amplitude, extrema point spacing, data segment mean, and data segment standard deviation of each data segment, and arranging the parameters in a set order to form a magnetic field time-series distortion feature set.

[0012] Further, step S3 includes the following sub-steps: S31, performing spectral analysis on the mixed interference signal, obtaining the frequency distribution characteristics of the signal through fast Fourier transform, identifying the signal energy peaks in different frequency ranges, and initially determining the possible frequency range of the interference source; S32, constructing an interference signal propagation path model, analyzing the propagation path of the interference signal from the source to the acquisition end based on the structural parameters and working environment layout of the radio therapy equipment, and determining the path attenuation coefficient and phase offset; S33, establishing a multi-source interference separation objective function, aiming to maximize the independence of the separated signals, and constructing the objective function by combining the spectral analysis results and path model parameters; S34, solving the interference separation weight matrix, using the gradient descent optimization algorithm to iteratively solve the objective function, obtaining the optimal weight matrix, and performing matrix operations to separate and decouple the mixed interference signal.

[0013] Further, S4 includes the following sub-steps: S41, collecting frequency data of the radio waves output by the radio therapy device, obtaining the output frequency values ​​at different times through a high-precision frequency counter, and forming a frequency time series; S42, calculating the frequency offset, comparing the collected frequency time series with the rated operating frequency of the device, obtaining the frequency offset data at each time, and statistically analyzing the maximum, minimum, and fluctuation range of the frequency offset; S43, constructing a frequency offset error objective function, combining the magnetic field time series distortion feature set and the decoupled interference signal, and establishing a correlation model between the frequency offset and the distortion feature parameters and the amplitude of the interference signal; S44, solving the objective function through the device's intelligent frequency offset convergence calibration algorithm, obtaining the frequency offset calibration amount, adjusting the device's frequency control parameters according to the calibration amount, and performing dynamic calibration of the frequency offset.

[0014] Further, S5 includes the following sub-steps: S51, extract common parameters, and select common parameters that are not affected by the equipment model and working environment from the magnetic field time-series distortion feature set obtained in S2 to form a common parameter set; S52, perform localized parameter adaptation, and combine the specific model, rated power, and operating frequency range of the radio-controlled physiotherapy equipment with the environmental parameters of the magnetic field strength and interference source type of the usage environment to adaptively adjust the common parameters; S53, integrate feature data and interference signals, and fuse the adapted localized parameters with the single-source interference signal decoupled in S3 and the frequency offset calibration model parameters obtained in S4 to form a multi-dimensional calibration parameter matrix; S54, optimize the calibration parameter set, and use principal component analysis to reduce the dimensionality of the multi-dimensional calibration parameter matrix, retain key calibration parameters, remove redundant information, and generate the final equipment error calibration parameter set.

[0015] A method for calibrating errors in radio-controlled physiotherapy equipment under complex magnetic field interference is disclosed. This method is implemented through a system for calibrating errors in radio-controlled physiotherapy equipment under complex magnetic field interference, comprising: a magnetic field time-series data acquisition and preprocessing unit, a multi-source electromagnetic interference separation and decoupling unit, a magnetic field time-series distortion feature extraction unit, a device frequency offset intelligent convergence calibration unit, a physiotherapy radio wave error tracing and analysis unit, and a calibration parameter adaptation and execution unit. The magnetic field time-series data acquisition and preprocessing unit is connected to the multi-source electromagnetic interference separation and decoupling unit and the magnetic field time-series distortion feature extraction unit via a high-speed data bus, used to acquire magnetic field time-series data and mixed interference signals and transmit them to the corresponding processing units. The multi-source electromagnetic interference separation and decoupling unit is connected to the device frequency offset intelligent convergence calibration unit via a signal interface. The separated single-source interference signal is transmitted to the device frequency offset intelligent convergence calibration unit; the magnetic field timing distortion feature extraction unit is connected to the device frequency offset intelligent convergence calibration unit through a data link, providing it with a magnetic field timing distortion feature set; the device frequency offset intelligent convergence calibration unit communicates bidirectionally with the physiotherapy wave error tracing analysis unit, receiving the error tracing analysis results and feeding back calibration progress data; the physiotherapy wave error tracing analysis unit is connected to the calibration parameter adaptation and execution unit through a control bus, transmitting the error tracing data to the calibration parameter adaptation and execution unit; the calibration parameter adaptation and execution unit is connected to the control system of the radio-controlled physiotherapy equipment, dynamically adjusting the equipment operating parameters based on the calibration parameter set, and all units work together to calibrate the equipment error under complex magnetic field interference.

[0016] Beneficial Effects: This invention proposes an error calibration method for radio-controlled physiotherapy equipment under complex magnetic field interference. By accurately capturing the core characteristics of magnetic field temporal distortion, it achieves effective separation and decoupling of multi-source electromagnetic interference. Combined with a dynamic frequency offset calibration strategy and comprehensive error source analysis, it overcomes the shortcomings of existing technologies in adapting to complex magnetic field interference. Instead of relying on simple filtering or shielding methods, it eliminates the calibration accuracy limitations caused by the superposition of multiple interference sources at the source through targeted feature extraction and interference decomposition. Simultaneously, through the adaptation mechanism of common and local parameters and intelligent convergence calibration logic, a dynamically adjusted calibration system is constructed. This system can respond in real time to changes in magnetic field interference and individual equipment differences, establishing a complete error source tracing and parameter adaptation process. This effectively solves the defects of existing technologies, such as lack of dynamic adaptation capability and poor calibration effect stability. It improves the accuracy and reliability of error calibration for radio-controlled physiotherapy equipment under complex magnetic field interference scenarios, ensures the accuracy of the equipment's output physiotherapy parameters, and provides stable technical support for clinical treatment and rehabilitation applications. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the overall process of the method of the present invention. Figure 2 This is a flowchart of method step S2 of the present invention; Figure 3 This is a flowchart of method step S3 of the present invention; Figure 4 This is a flowchart of method step S4 of the present invention; Figure 5 This is a flowchart of step S5 of the method of the present invention. Detailed Implementation

[0018] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0019] like Figure 1As shown, a method for calibrating errors in radio-controlled physiotherapy equipment under complex magnetic field interference is characterized by the following steps: S1, collecting magnetic field time-series data and multi-source electromagnetic interference signals during the operation of the radio-controlled physiotherapy equipment through a physiotherapy radio wave error tracing and analysis platform, and simultaneously recording the frequency, phase, and amplitude correlation data of the equipment's output radio waves; S2, using a magnetic field time-series distortion feature extraction algorithm to identify distortion features in the collected magnetic field time-series data, and obtaining a magnetic field time-series distortion feature set through time-series correlation analysis, distortion extreme point location, and feature vector construction; S3, using a multi-source electromagnetic interference separation and decoupling algorithm to separate and process the multi-source electromagnetic interference signals, based on the interference... S4. Decouple and separate different interference source signals based on the spectral characteristics and propagation path differences of the interference signals; S5. Utilize the intelligent convergence calibration algorithm for frequency offset to dynamically calculate the frequency offset of the output radio waves of the radio therapy equipment, and establish a frequency offset calibration model by combining the magnetic field timing distortion feature set and the decoupled interference signal; S6. Through common parameter extraction and localized parameter adaptation mechanisms, integrate the feature data, decoupled signals, and frequency offset calibration model parameters obtained in S2-S4 to generate a set of equipment error calibration parameters; S7. Based on the radio therapy wave error tracing and analysis platform, import the calibration parameter set into the radio therapy equipment control system to complete the dynamic adjustment of equipment operating parameters and error calibration.

[0020] Step S1 involves multi-dimensional data acquisition using a radiofrequency wave error tracing and analysis platform. This platform is equipped with a 16-channel high-speed data acquisition module, which simultaneously acquires magnetic field timing data and multi-source electromagnetic interference signals during the operation of the radiofrequency physiotherapy equipment. The acquisition time window is set to 10 milliseconds, and 200 sets of data are continuously acquired to ensure data integrity. Simultaneously, the frequency, phase, and amplitude correlation data of the equipment's output radio waves are recorded synchronously through the device's built-in frequency detection module, phase sensor, and amplitude acquisition unit. The frequency recording accuracy reaches 0.001 Hz, the phase recording accuracy is 0.01 degrees, and the amplitude recording accuracy is 0.001 volts. The acquired data is processed in real time. The data is stored in an NVMe interface solid-state drive with a capacity of at least 1024 gigabytes and a storage rate of at least 2000 megabytes per second, ensuring that there is no data loss or delay during the data transmission and storage process. This step provides comprehensive and accurate raw data support for subsequent magnetic field timing distortion feature extraction, multi-source electromagnetic interference separation and decoupling, and equipment frequency offset calibration. It is the foundation of the entire error calibration process. Through high-frequency and high-precision data acquisition, the dynamic characteristics of magnetic field timing changes and electromagnetic interference signals can be fully captured, avoiding deviations in subsequent calibration results due to data loss or insufficient accuracy. This lays the data foundation for achieving accurate error calibration under complex magnetic field interference.

[0021] Step S2 employs magnetic field temporal distortion feature extraction technology to perform deep processing on the collected magnetic field time series data. First, the continuous time series data is divided into 40 data segments at fixed time intervals of 5 milliseconds, each segment containing 62,500 sampling points. A sliding window technique is used to achieve non-overlapping coverage of the data segments. Then, the temporal correlation coefficient of each data segment is calculated, and a correlation threshold of 0.85 is set. Twenty suspected distorted data segments with correlation coefficients below the threshold are selected. Next, a difference algorithm is used to process the suspected distorted data segments. The location of extreme points is determined by the sign change of the difference between adjacent sampling points. The number and distribution density of extreme points within each data segment are statistically analyzed. With a set extreme point density threshold of 1000 per millisecond, 15 data segments with significant distortion were further screened out. Finally, the extreme point amplitude, extreme point spacing, data segment mean, and data segment standard deviation of each valid data segment were extracted. The four types of parameters were arranged in descending order of extreme point amplitude to construct a core feature set of magnetic field time series distortion including 60 feature parameters. This step, through multi-dimensional feature extraction, can accurately identify the distortion patterns in magnetic field time series data, providing targeted feature basis for subsequent multi-source electromagnetic interference separation and frequency offset calibration, effectively improving the efficiency and accuracy of subsequent processing steps, and ensuring that the calibration process can accurately focus on the core influencing factors of magnetic field distortion.

[0022] Step S3 employs multi-source electromagnetic interference separation and decoupling technology to decompose the collected mixed interference signal. First, a fast Fourier transform is used to perform spectral analysis on the mixed interference signal, covering a frequency range from 30 Hz to 300 GHz. Eight signal energy peaks in different frequency intervals are identified, initially determining the presence of six main interference sources, with frequencies concentrated around 50 Hz, 1000 Hz, 100000 Hz, 1 GHz, 10 GHz, and 100 GHz. Then, based on the structural parameters and working environment layout of the radio-controlled physiotherapy equipment, six interference signal propagation path models are constructed, and the attenuation coefficient and phase shift of each path are calculated. The attenuation coefficient ranges from... The phase offset is between 0 and 180 degrees, with values ​​ranging from 0.1 to 0.9. Next, a multi-source interference separation objective function is established, aiming to maximize the independence of the separated signals. The objective function weights are set based on spectral analysis results and path model parameters. Finally, a gradient descent optimization algorithm is used to iteratively solve the objective function, with 1000 iterations and a learning rate of 0.001, to obtain the optimal interference separation weight matrix. Matrix operations are then used to split the mixed interference signal into six single-source interference signals. This step achieves accurate separation of multi-source electromagnetic interference, completely solving the problem of different interference source signals superimposing and affecting calibration accuracy, providing a clean signal environment for subsequent frequency offset calibration.

[0023] Step S4 utilizes intelligent convergence calibration technology to dynamically calibrate the frequency offset of the output radio waves from the radio therapy equipment. First, a high-precision frequency counter collects frequency data of the output radio waves at a frequency of 1000 times per second for 100 seconds, forming a frequency time-series sequence of 100,000 data points. Then, the collected frequency time-series sequence is compared with the equipment's rated operating frequency, and the frequency offset data at each moment is calculated. The maximum frequency offset is found to be 50 Hz, the minimum to 5 Hz, and the fluctuation range to be 45 Hz. Next, combining the core feature set of magnetic field time-series distortion and the decoupled six single-source interference signals, the frequency offset is compared with the distortion... A correlation model is established by varying characteristic parameters and interference signal amplitude. The model training sample size is set to 80,000, the test sample size to 20,000, and the model training iterations to 500. Finally, the objective function is solved using an intelligent convergence calibration algorithm to obtain the real-time frequency offset calibration value. The calibration value adjustment step size is dynamically set according to the frequency offset fluctuation, ranging from 0.01 Hz to 1 Hz. The frequency control parameters of the device are adjusted in real time according to the calibration value to achieve dynamic calibration of the frequency offset. This step, through the dynamic calibration strategy, can accurately compensate for the frequency offset caused by magnetic field interference, improve the frequency stability of the output radio wave of the device, and provide core technical support for error calibration.

[0024] Step S5 integrates and optimizes the calibration parameter set through a common parameter extraction and localized parameter adaptation mechanism. First, 15 common parameters unaffected by equipment model and operating environment are selected from the core feature set of magnetic field temporal distortion. These include the mean distance between extreme points and the mean standard deviation of data segments, forming a common parameter set. Then, considering equipment parameters such as the specific model, rated power, and operating frequency range of the radiotherapy equipment, as well as environmental parameters such as magnetic field strength and interference source type, the common parameters are adaptively adjusted. The adjustment coefficient is set between 0.8 and 1.2 based on the differences in equipment parameters. Finally, the 15 adapted localized parameters are... The calibration parameters are fused with the characteristic parameters of the six decoupled single-source interference signals and the eight core parameters of the frequency offset calibration model to form a multi-dimensional calibration parameter matrix with 29 parameters. Finally, principal component analysis is used to reduce the dimensionality of the multi-dimensional calibration parameter matrix, retaining 12 key calibration parameters with a cumulative variance contribution rate of 95%, removing redundant information, and generating the final equipment error calibration parameter set. This step, through parameter screening, adaptation, and fusion, ensures that the calibration parameter set is both universal and adaptable to specific equipment and environments, providing accurate and efficient parameter support for subsequent equipment parameter adjustments and improving the pertinence and reliability of calibration operations.

[0025] Step S6, based on the physiotherapy radio wave error tracing and analysis platform, completes the dynamic adjustment and error calibration of equipment parameters. First, the final generated set of 12 key calibration parameters is imported into the radio-controlled physiotherapy equipment via a high-speed data bus. The data transmission rate is no less than 1000 megabytes per second to ensure real-time parameter transmission. Then, the equipment control system dynamically adjusts the frequency generator, phase adjuster, and amplitude controller of the equipment according to the frequency calibration parameters, phase calibration parameters, and amplitude calibration parameters in the calibration parameter set. The frequency adjustment accuracy is set to 0.001 Hz, the phase adjustment accuracy to 0.01 degrees, and the amplitude adjustment accuracy to 0.001 volts. During the adjustment process, the frequency, phase, and amplitude data of the equipment's output radio waves are collected in real time and compared with... The calibration target value is compared, and the calibration error is calculated. If the frequency error is greater than 0.01 Hz, the phase error is greater than 0.1 degrees, or the amplitude error is greater than 0.01 volts, the error tracing function of the physiotherapy wave error tracing analysis platform is used to locate the specific link in the error. The corresponding calibration parameters are then readjusted until the calibration error meets the requirements, namely, the frequency error does not exceed 0.01 Hz, the phase error does not exceed 0.1 degrees, and the amplitude error does not exceed 0.01 volts. This completes the dynamic adjustment and error calibration of the equipment's operating parameters. This step, through a closed-loop calibration process, ensures that the equipment can continuously output accurate physiotherapy wave parameters in complex magnetic field interference environments, improves the equipment's working accuracy and stability, and provides reliable protection for clinical treatment and rehabilitation applications.

[0026] The physiotherapy electromagnetic wave error tracing and analysis platform is the core hardware carrier supporting the entire error calibration process. It consists of an FPGA processing module using Cyclone V series chips with a 40nm process, an 8-channel multi-channel high-speed data acquisition module with a sampling rate of 1.25GSps, a millimeter-wave magnetic field sensor array with 16 tetrahedral distributions, a real-time signal processing module with a processing latency of less than 50μs, and a calibration parameter storage module with more than 1TB of NVMe storage. It undertakes the functions of magnetic field time-series data and interference signal acquisition, algorithm hardware acceleration calculation, calibration parameter storage and distribution, and error closed-loop tracing. It is the physical foundation for realizing the implementation of the entire process of data acquisition, analysis, and calibration, and can accurately locate the error generation link and dynamically provide feedback and adjustment. The magnetic field time-series distortion feature extraction algorithm is a distortion identification technique for magnetic field time-series data. Based on the principle of time-series signal correlation analysis, it quantifies the delay correlation between the original magnetic field time-series data and the distorted reference time-series data through integral operations. It introduces a time-series correlation weight coefficient to balance the contribution of different delays, and then extracts the maximum, minimum, integral, and variance of the distortion feature function to construct a four-dimensional feature vector. This enables precise screening and feature quantification of magnetic field time-series distortion regions, extracting distortion patterns from the time-series data and providing targeted feature inputs for subsequent interference separation and frequency offset calibration, avoiding the problem of incomplete distortion identification caused by single features. The multi-source electromagnetic interference separation and decoupling algorithm is the core technology for achieving mixed interference splitting. Based on the linear superposition assumption, it maps the collected mixed interference signal matrix to a single-source interference signal matrix through an interference separation weight matrix. Simultaneously, it constructs an optimization objective function by minimizing the information entropy of the separated signals combined with L2 regularization, and solves the optimal weight matrix iteratively through gradient descent. This enables the splitting of interference signals from different sources based on spectral characteristics and propagation path differences, breaking through the calibration accuracy bottleneck caused by multi-source interference superposition and providing a clean signal analysis environment for subsequent frequency offset calibration. The intelligent convergence calibration algorithm for frequency offset is the core algorithm for dynamic frequency offset compensation. It integrates the gradient information of the objective function of frequency offset error with the attenuation effect of historical calibration, controls the adjustment range through the calibration step size coefficient, and balances the weight of historical calibration experience with the exponential decay mechanism. It can dynamically calculate the real-time frequency offset calibration amount, ensuring both calibration response speed and avoiding calibration oscillation, thus achieving intelligent convergence and dynamic compensation of frequency offset, and ultimately improving the frequency stability and overall calibration accuracy of the equipment output radio waves.

[0027] Preferably, the expression for the magnetic field temporal distortion feature extraction algorithm is: ,in The time-series correlation weighting coefficient. This is the original magnetic field time series data. For distorted reference time series data, The length of the feature extraction time window. For time delay variables, This is the current observation time; The formula for constructing the distorted feature vector is calculated as follows: ,in, This represents the maximum value of the distortion characteristic function. This represents the minimum value of the distortion characteristic function. The integral value of the distortion characteristic function. The variance of the distortion characteristic function. This is the distorted feature vector.

[0028] Specifically, the magnetic field temporal distortion feature extraction algorithm is based on the requirements of temporal signal correlation analysis and distortion feature quantification. By analyzing the dynamic changes in magnetic field temporal data and considering the time delay correlation between the original magnetic field temporal data and the distorted reference temporal data, it uses integral operations to describe the correlation between the two within a specific time window. A temporal correlation weight coefficient is introduced to balance the correlation contribution under different time delays, thus establishing an initial feature extraction expression. To comprehensively characterize distortion features, extreme value features, energy features, and stability features are further extracted from the output of the initial expression. Feature vectors are constructed using four key indicators: maximum value, minimum value, integral value, and variance, forming a complete distortion feature description system. The algorithm's parameter values ​​have been verified through extensive experiments. The temporal correlation weight coefficient is set between 0.7 and 0.9, the feature extraction time window length is set to 10 milliseconds, and the time delay variable ranges from 0 to 5 milliseconds. During implementation, data preprocessing is first used to remove accidental noise from the collected data. Then, the parameters are substituted into the data to perform integral operations to obtain the distortion feature function. Subsequently, the four key indicators of this function are calculated and arranged in order to form a feature vector. Magnetic field temporal distortion can cause changes in the correlation between the original data and the reference data and abnormal feature parameters. By quantifying these changes, the core features of the distortion can be accurately captured, providing precise feature input for interference separation and error calibration. This avoids the problem of incomplete distortion identification caused by a single feature and improves the accuracy and completeness of magnetic field temporal distortion feature extraction.

[0029] Preferably, the expression for the multi-source electromagnetic interference separation and decoupling algorithm is: ,in The separated single-source interference signal matrix, For interference separation weight matrix, The collected mixed interference signal matrix; The decoupling optimization formula is calculated as follows: ,in The optimized interference separation weight matrix, For the first Information entropy of the path-separated signal For the number of interference sources, The regularization coefficient is . Let L be the L2 norm of the weight matrix.

[0030] Specifically, the multi-source electromagnetic interference separation and decoupling algorithm is based on the principle of linear separation of mixed interference signals. It assumes that the mixed interference signal is formed by the linear superposition of multiple single-source interference signals. An interference separation weight matrix is ​​introduced to achieve a linear transformation between the mixed signal and the single-source signal, establishing an initial separation expression. To improve separation accuracy, considering the independence of the separated single-source signals, information entropy is introduced as an independence evaluation index. Simultaneously, a regularization term is added to avoid overfitting of the weight matrix, constructing an optimization objective function that includes minimizing information entropy and regularization constraints, forming a complete decoupling algorithm system. The algorithm parameters have been optimized through multiple simulations. The number of interference sources is determined to be 6 based on spectral analysis results, the regularization coefficient is set between 0.001 and 0.01, and the initial value of the weight matrix is ​​set to an identity matrix. During implementation, spectral analysis is first performed on the mixed interference signal to determine the number of interference sources. After initializing the weight matrix, the objective function is iteratively solved using a gradient descent optimization algorithm. During the iteration process, the weight matrix parameters are continuously adjusted until the information entropy reaches its minimum value and the weight matrix tends to stabilize. Finally, the separated single-source interference signal is obtained through matrix operations. By leveraging the linear superposition characteristics and independence requirements of multi-source electromagnetic interference signals, this method overcomes the limitations of traditional interference processing that cannot distinguish between single-source interference. Through precise separation and decoupling, it eliminates the superposition effects of multi-source interference for subsequent error calibration, providing a clean signal environment for equipment frequency offset calibration.

[0031] Preferably, the expression for the device frequency offset intelligent convergence calibration algorithm is: ,in This is the frequency offset calibration value. To calibrate the step size factor, Let the frequency offset error be the objective function. For the error model parameter vector, Let be the partial derivative of the objective function with respect to frequency. The historical calibration weighting coefficient. This is the frequency offset calibration value from the previous moment. The convergence decay coefficient is... To calibrate the iteration time.

[0032] Specifically, the intelligent convergence calibration algorithm for frequency offset is based on the principles of error feedback and dynamic calibration. It comprehensively considers the gradient information of the frequency offset error objective function and the attenuation effect of historical calibration values. First, it calculates the partial derivative of the frequency offset error objective function with respect to frequency to obtain the current adjustment direction of the frequency offset, and introduces a calibration step size coefficient to control the adjustment amplitude. Simultaneously, to avoid oscillations during the calibration process, it introduces a historical calibration value weighting coefficient and a convergence attenuation coefficient. An exponential function describes the attenuation trend of historical calibration values, balancing the contributions of current gradient information and historical calibration experience to establish a complete expression for calculating the frequency offset calibration value. The algorithm's parameter values ​​have been optimized through actual equipment testing. The calibration step size coefficient is set between 0.1 and 0.3, the historical calibration value weighting coefficient is set between 0.4 and 0.6, the convergence attenuation coefficient is set between 0.01 and 0.05, and the calibration iteration time is set to 2 milliseconds per calibration cycle. During implementation, the output frequency data of the equipment is first collected to calculate the objective function of the frequency deviation error. Then, the partial derivative of the objective function with respect to frequency is solved. Combining the frequency deviation calibration amount and various parameters from the previous moment, the current frequency deviation calibration amount is calculated using the formula. Subsequently, the calibration amount is transmitted to the equipment control system to adjust the frequency parameters. This formula achieves intelligent convergence of frequency deviation calibration by taking into account the dynamic changes in frequency deviation error and the convergence requirements of the calibration process, avoiding the problem of untimely adjustment caused by static calibration. At the same time, the historical calibration amount attenuation mechanism improves calibration stability, ensuring that the equipment frequency deviation can quickly and accurately return to the rated range.

[0033] Preferably, the physiotherapy electromagnetic wave error tracing and analysis platform includes an FPGA processing module, a multi-channel high-speed data acquisition module, a millimeter-wave magnetic field sensing array, a real-time signal processing module, and a calibration parameter storage module. The FPGA processing module uses a 40nm Cyclone V series chip and integrates a hardware acceleration engine for parallel algorithm computation. The multi-channel high-speed data acquisition module has a sampling rate of 1.25GSps, a quantization bit depth of 16 bits, and supports 8-channel synchronous acquisition. The millimeter-wave magnetic field sensing array consists of 16 sensing units arranged in a tetrahedral pattern, with a sensing unit detection bandwidth covering 30GHz-300GHz. The real-time signal processing module has a built-in embedded Linux system with a custom signal processing kernel, and the data processing latency is controlled within 50μs. The calibration parameter storage module uses an NVMe interface solid-state drive with a storage capacity of not less than 1TB, supporting real-time writing and fast reading of calibration parameters.

[0034] Preferred, such as Figure 2As shown, step S2 includes the following sub-steps: S21, performing time-series segmentation processing on the collected magnetic field time-series data, dividing the continuous time-series data into several data segments according to a fixed time interval, with each data segment including the same number of sampling points, and using sliding window technology to achieve non-overlapping coverage of the data segments; S22, calculating the time-series correlation coefficient of each data segment, and initially screening out data segments with distortion characteristics by comparing the amplitude change trend and phase synchronization of adjacent data segments; S23, locating distortion extrema points, processing the screened data segments using a difference algorithm, determining the location of extrema points by the sign change of the difference between adjacent sampling points, and statistically analyzing the number and distribution density of extrema points within each data segment; S24, constructing a distortion feature vector, extracting the extrema point amplitude, extrema point spacing, data segment mean, and data segment standard deviation of each data segment, and arranging the parameters in a set order to form a magnetic field time-series distortion feature set.

[0035] Specifically, step S2 involves extracting the temporal distortion features of the magnetic field. S21 Divide the continuous magnetic field time series data into 40 data segments at fixed 5-millisecond time intervals. Each data segment includes 62,500 sampling points. A sliding window technique is used to ensure that the data segments do not overlap, avoiding data omission or duplicate processing. S22 Calculate the temporal correlation coefficient of each data segment and set 0.85 as the correlation threshold. Select suspected distorted data segments with coefficients below the threshold. By comparing the amplitude change trend and phase synchronization of adjacent data segments, the region where distortion may exist is initially identified. S23 Process the selected data segments using a differential algorithm. Locate the distortion extrema by the sign change of the difference between adjacent sampling points. Set an extrema density threshold of 1,000 per millisecond. Statistically count the number and distribution density of extrema within each data segment to further accurately select 15 data segments with obvious distortion. S24 Extract four core parameters from each valid data segment: extrema amplitude, extrema interval, data segment mean, and data segment standard deviation. Arrange them in descending order of extrema amplitude to form a core feature set of magnetic field time series distortion. The parameters set have been verified through multiple experiments. The values ​​of time interval, number of sampling points, and threshold can balance processing efficiency and feature extraction accuracy. A complete processing link is formed from data segmentation to feature vector construction. Through refined step-by-step operations, the temporal distortion characteristics of magnetic field are captured comprehensively and accurately, providing reliable feature input for interference separation and error calibration, and avoiding calibration errors caused by incomplete feature extraction.

[0036] Preferred, such as Figure 3As shown, step S3 includes the following sub-steps: S31, performing spectral analysis on the mixed interference signal, obtaining the frequency distribution characteristics of the signal through fast Fourier transform, identifying the signal energy peaks in different frequency ranges, and initially determining the possible frequency range of the interference source; S32, constructing an interference signal propagation path model, analyzing the propagation path of the interference signal from the source to the acquisition end based on the structural parameters and working environment layout of the radio therapy equipment, and determining the path attenuation coefficient and phase offset; S33, establishing a multi-source interference separation objective function, aiming to maximize the independence of the separated signals, and constructing the objective function by combining the spectral analysis results and path model parameters; S34, solving the interference separation weight matrix, using the gradient descent optimization algorithm to iteratively solve the objective function, obtaining the optimal weight matrix, and performing matrix operations to separate and decouple the mixed interference signal.

[0037] Specifically, step S3 includes: S31 performing spectral analysis on the mixed interference signal, using Fast Fourier Transform to obtain the signal frequency distribution characteristics, analyzing the frequency range from 30 Hz to 300 GHz, identifying the signal energy peaks in different frequency ranges, and initially determining the frequency ranges of the six main interference sources; S32 constructing a model of six interference signal propagation paths based on the structural parameters and working environment layout of the radio therapy equipment, calculating the attenuation coefficient and phase offset of each path, with the attenuation coefficient controlled between 0.1 and 0.9, and the phase offset between 0 and 180 degrees, accurately describing the propagation characteristics of the interference signal; S33 constructing a multi-source interference separation objective function based on the spectral analysis results and path model parameters, aiming to maximize the independence of the separated signal, and clarifying the direction of separation optimization; S34 using the gradient descent optimization algorithm to iteratively solve the objective function, setting 1000 iterations and a learning rate of 0.001, obtaining the optimal interference separation weight matrix, and achieving the separation and decoupling of the mixed interference signal through matrix operations. The parameter settings take into account both the interference identification range and the separation accuracy. The four steps from signal analysis to model building and optimization solution form a systematic interference processing flow, breaking the traditional extensive mode of interference processing. Through step-by-step refined operation, it achieves accurate separation of multi-source interference, eliminates the superposition of interference for subsequent equipment frequency offset calibration, and improves the accuracy of calibration results.

[0038] Preferred, such as Figure 4As shown, step S4 includes the following sub-steps: S41, collecting frequency data of the radio waves output by the radio therapy device, obtaining the output frequency values ​​at different times through a high-precision frequency counter, and forming a frequency time sequence; S42, calculating the frequency offset, comparing the collected frequency time sequence with the rated operating frequency of the device, obtaining the frequency offset data at each time, and statistically analyzing the maximum, minimum, and fluctuation range of the frequency offset; S43, constructing a frequency offset error objective function, combining the magnetic field time sequence distortion feature set and the decoupled interference signal, and establishing a correlation model between the frequency offset and the distortion feature parameters and the amplitude of the interference signal; S44, solving the objective function through the device's intelligent frequency offset convergence calibration algorithm, obtaining the frequency offset calibration amount, adjusting the device's frequency control parameters according to the calibration amount, and performing dynamic calibration of the frequency offset.

[0039] Specifically, step S4 performs intelligent convergence calibration of the device's frequency offset. S41 uses a high-precision frequency counter to collect frequency data of the device's output radio waves, setting a collection frequency of 1000 times per second for 100 seconds to form a frequency time sequence of 100,000 data points, ensuring data continuity and integrity. S42 compares the collected frequency time sequence with the device's rated operating frequency, calculates the frequency offset data at each moment, and statistically obtains the fluctuation range of the maximum frequency offset of 50 Hz, the minimum of 5 Hz, and 45 Hz, comprehensively understanding the frequency offset situation. S43 combines the core feature set of magnetic field time-series distortion and the decoupled single-source interference signal to establish a correlation model between the frequency offset, distortion feature parameters, and interference signal amplitude, setting 80,000 training samples and 20,000 test samples to ensure the model's generalization ability. S44 solves the objective function using the device's intelligent convergence calibration algorithm for frequency offset, setting a dynamic calibration step size of 0.01 Hz to 1 Hz, and adjusting the device's frequency control parameters in real time according to the calibration amount to achieve dynamic calibration of the frequency offset. The parameter values ​​have been optimized through actual equipment testing. The settings of acquisition frequency, sample quantity and calibration step size can balance data acquisition efficiency and calibration accuracy. From data acquisition to model building and then to dynamic calibration, a closed-loop processing flow is formed to realize intelligent and dynamic frequency offset calibration, accurately compensate for the frequency offset caused by magnetic field interference, and improve the frequency stability of the output radio waves of the equipment.

[0040] Preferred, such as Figure 5As shown, step S5 includes the following sub-steps: S51, extract common parameters: select common parameters that are not affected by equipment model and working environment from the magnetic field time-series distortion feature set obtained in S2 to form a common parameter set; S52, perform localized parameter adaptation: combine the specific model, rated power, and working frequency range of the radio-controlled physiotherapy equipment with the magnetic field strength and interference source type of the environment to adaptively adjust the common parameters; S53, integrate feature data and interference signals: fuse the adapted localized parameters with the single-source interference signal decoupled from S3 and the frequency offset calibration model parameters obtained in S4 to form a multi-dimensional calibration parameter matrix; S54, optimize the calibration parameter set: use principal component analysis to reduce the dimensionality of the multi-dimensional calibration parameter matrix, retain key calibration parameters, remove redundant information, and generate the final equipment error calibration parameter set.

[0041] Specifically, step S5 involves extracting common parameters and adapting localized parameters to optimize the calibration parameter set. S51 selects 15 common parameters from the core features of magnetic field temporal distortion that are unaffected by equipment model or operating environment, including the mean distance between extreme points and the mean standard deviation of data segments, forming a common parameter set to ensure parameter universality. S52 combines equipment parameters such as the specific model, rated power, and operating frequency range of the radio-controlled physiotherapy equipment, as well as environmental parameters such as magnetic field strength and interference source type, to adaptively adjust the common parameters. The adjustment coefficient is set between 0.8 and 1.2 to adapt the parameters to specific scenarios. S53 fuses the adapted 15 localized parameters with the decoupled 6-channel single-source interference signal characteristic parameters and the 8 core parameters of the frequency offset calibration model to form a multi-dimensional calibration parameter matrix containing 29 parameters, enriching the parameter dimensions. S54 uses principal component analysis to reduce the dimensionality of the multi-dimensional calibration parameter matrix, retaining 12 key calibration parameters with a cumulative variance contribution rate of 95%, removing redundant information, and generating the final equipment error calibration parameter set. The parameter settings take into account both universality and specificity. From parameter selection, adaptation to integration and optimization, a complete parameter processing chain is formed. Through step-by-step operation, the calibration parameters are accurately optimized, ensuring that the parameter set is not only compatible with different devices and environments, but also has high simplification and effectiveness. This provides reliable support for device parameter adjustment and improves the specificity and efficiency of calibration operations.

[0042] A method for calibrating errors in radio-controlled physiotherapy equipment under complex magnetic field interference is disclosed. This method is implemented through a system for calibrating errors in radio-controlled physiotherapy equipment under complex magnetic field interference, comprising: a magnetic field time-series data acquisition and preprocessing unit, a multi-source electromagnetic interference separation and decoupling unit, a magnetic field time-series distortion feature extraction unit, a device frequency offset intelligent convergence calibration unit, a physiotherapy radio wave error tracing and analysis unit, and a calibration parameter adaptation and execution unit. The magnetic field time-series data acquisition and preprocessing unit is connected to the multi-source electromagnetic interference separation and decoupling unit and the magnetic field time-series distortion feature extraction unit via a high-speed data bus, used to acquire magnetic field time-series data and mixed interference signals and transmit them to the corresponding processing units. The multi-source electromagnetic interference separation and decoupling unit is connected to the device frequency offset intelligent convergence calibration unit via a signal interface. The separated single-source interference signal is transmitted to the device frequency offset intelligent convergence calibration unit; the magnetic field timing distortion feature extraction unit is connected to the device frequency offset intelligent convergence calibration unit through a data link, providing it with a magnetic field timing distortion feature set; the device frequency offset intelligent convergence calibration unit communicates bidirectionally with the physiotherapy wave error tracing analysis unit, receiving the error tracing analysis results and feeding back calibration progress data; the physiotherapy wave error tracing analysis unit is connected to the calibration parameter adaptation and execution unit through a control bus, transmitting the error tracing data to the calibration parameter adaptation and execution unit; the calibration parameter adaptation and execution unit is connected to the control system of the radio-controlled physiotherapy equipment, dynamically adjusting the equipment operating parameters based on the calibration parameter set, and all units work together to calibrate the equipment error under complex magnetic field interference.

[0043] A method for calibrating errors in radio-controlled physiotherapy equipment under complex magnetic field interference is proposed. This method constructs a systematic complex magnetic field interference processing and dynamic calibration system. Through deep integration of algorithms and a dedicated analysis platform, it achieves precise and intelligent interference processing and error calibration. Magnetic field temporal distortion feature extraction technology comprehensively captures the distortion patterns and core characteristics of magnetic field temporal data. Combined with multi-source electromagnetic interference separation and decoupling technology, it can accurately separate interference signals from different sources, fundamentally solving the problem of distinguishing complex interference. Simultaneously, the intelligent frequency offset convergence calibration technology and the physiotherapy radio wave error tracing analysis platform work together, combining common parameter extraction and localized parameter adaptation mechanisms to achieve dynamic optimization and precise matching of calibration parameters, improving the pertinence and reliability of calibration operations.

[0044] This method addresses the problem of insufficient adaptability of existing technologies to complex magnetic field interference. It abandons the crude processing mode of simple filtering or shielding and achieves refined processing of complex magnetic field interference through precise distortion feature extraction and multi-source interference separation and decoupling. It can completely isolate the influence of different interference sources on the equipment and improve calibration accuracy. To address the lack of dynamic adaptation capability in existing technologies, it adopts an intelligent convergent calibration algorithm and parameter adaptation mechanism, which can respond in real time to the dynamic changes of magnetic field interference and the individual differences of different equipment. By dynamically adjusting the calibration strategy and parameters, it establishes a complete error tracing and compensation process, which changes the limitations of the traditional static calibration mode and enhances the stability and versatility of the calibration effect.

[0045] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," "link," and "fix" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal communication between two components. Those skilled in the art will understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0046] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various equivalent 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 method for calibrating errors in radio-guided physiotherapy equipment under complex magnetic field interference, characterized in that, Includes the following steps: S1 collects magnetic field timing data and multi-source electromagnetic interference signals during the operation of radio therapy equipment through the physiotherapy wave error tracing and analysis platform, and simultaneously records the frequency, phase and amplitude correlation data of the equipment output waves. S2, a magnetic field time series distortion feature extraction algorithm is used to identify distortion features in the collected magnetic field time series data. The magnetic field time series distortion feature set is obtained through time series correlation analysis, distortion extreme point location and feature vector construction. S3 utilizes a multi-source electromagnetic interference separation and decoupling algorithm to separate multi-source electromagnetic interference signals, and decouples and splits signals from different interference sources based on the spectral characteristics and propagation path differences of the interference signals. S4. The frequency offset of the output radio waves of the radio therapy equipment is dynamically calculated by using the intelligent convergence calibration algorithm of the equipment frequency offset. The frequency offset calibration model is established by combining the magnetic field time sequence distortion feature set and the decoupled interference signal. S5 integrates the feature data, decoupled signals, and frequency offset calibration model parameters obtained from S2-S4 through a common parameter extraction and localized parameter adaptation mechanism to generate a set of equipment error calibration parameters. S6, based on the physiotherapy radio wave error tracing and analysis platform, imports the calibration parameter set into the radio-controlled physiotherapy equipment control system to complete the dynamic adjustment of equipment operating parameters and error calibration.

2. The method for calibrating errors in radio therapy equipment under complex magnetic field interference as described in claim 1, characterized in that, The expression for the magnetic field temporal distortion feature extraction algorithm is as follows: ,in The time-series correlation weighting coefficient. This is the original magnetic field time series data. This is distorted reference time series data. The length of the feature extraction time window. For time delay variables, This is the current observation time; The formula for constructing the distorted feature vector is calculated as follows: ,in, This represents the maximum value of the distortion characteristic function. This represents the minimum value of the distortion characteristic function. The integral value of the distortion characteristic function. The variance of the distortion characteristic function. This is the distorted feature vector.

3. The method for calibrating errors in radio therapy equipment under complex magnetic field interference as described in claim 1, characterized in that, The expression for the multi-source electromagnetic interference separation and decoupling algorithm is as follows: ,in The separated single-source interference signal matrix, For interference separation weight matrix, The collected mixed interference signal matrix; The decoupling optimization formula is calculated as follows: ,in The optimized interference separation weight matrix, For the first Information entropy of the path-separated signal For the number of interference sources, The regularization coefficient is . Let L be the L2 norm of the weight matrix.

4. The method for calibrating errors in radio therapy equipment under complex magnetic field interference as described in claim 1, characterized in that, The expression for the intelligent convergence calibration algorithm for frequency offset of the device is: ,in This is the frequency offset calibration value. To calibrate the step size factor, Let the frequency offset error be the objective function. For the error model parameter vector, Let be the partial derivative of the objective function with respect to frequency. The historical calibration weighting coefficient. This is the frequency offset calibration value from the previous moment. The convergence decay coefficient is... To calibrate the iteration time.

5. The method for calibrating errors in radio therapy equipment under complex magnetic field interference according to claim 1, characterized in that, The physiotherapy electromagnetic wave error tracing and analysis platform includes an FPGA processing module, a multi-channel high-speed data acquisition module, a millimeter-wave magnetic field sensing array, a real-time signal processing module, and a calibration parameter storage module. The FPGA processing module uses a 40nm Cyclone V series chip and integrates a hardware acceleration engine for parallel algorithm computation. The multi-channel high-speed data acquisition module has a sampling rate of 1.25GSps, a quantization bit depth of 16 bits, and supports 8-channel synchronous acquisition. The millimeter-wave magnetic field sensing array consists of 16 sensing units arranged in a tetrahedral pattern, with a sensing unit detection bandwidth covering 30GHz-300GHz. The real-time signal processing module has a built-in embedded Linux system with a custom signal processing kernel, and the data processing latency is controlled within 50μs. The calibration parameter storage module uses an NVMe interface solid-state drive with a storage capacity of no less than 1TB, supporting real-time writing and fast reading of calibration parameters.

6. The method for calibrating errors in radio therapy equipment under complex magnetic field interference according to claim 1, characterized in that, S2 includes: The collected magnetic field time series data is processed by time series segmentation. The continuous time series data is divided into several data segments according to a fixed time interval. Each data segment includes the same number of sampling points. The data segments are covered without overlap by sliding window technology. The time series correlation coefficient of each data segment is calculated. By comparing the amplitude change trend and phase synchronization of adjacent data segments, data segments with distortion characteristics are initially screened. The extreme points of distortion are located, and the filtered data segments are processed using a difference algorithm. The location of the extreme points is determined by the sign change of the difference between adjacent sampling points. The number and distribution density of extreme points in each data segment are counted. A distortion feature vector is constructed, and the extreme point amplitude, extreme point spacing, data segment mean, and data segment standard deviation of each data segment are extracted. The parameters are arranged in a set order to form a magnetic field temporal distortion feature set.

7. The method for calibrating errors in radio therapy equipment under complex magnetic field interference according to claim 1, characterized in that, S3 includes: Spectral analysis was performed on the mixed interference signal. The frequency distribution characteristics of the signal were obtained through Fast Fourier Transform, and the signal energy peaks in different frequency ranges were identified to preliminarily determine the possible frequency range of the interference source. An interference signal propagation path model was constructed. Based on the structural parameters and working environment layout of the radio-controlled physiotherapy equipment, the propagation path of the interference signal from the source to the acquisition end was analyzed, and the path attenuation coefficient and phase offset were determined. A multi-source interference separation objective function was established, with the goal of maximizing the independence of the separated signals. The objective function was constructed by combining the spectral analysis results and path model parameters. The interference separation weight matrix is ​​solved by using a gradient descent optimization algorithm to iteratively solve the objective function and obtain the optimal weight matrix. Matrix operations are then used to separate and decouple the mixed interference signals.

8. The method for calibrating errors in radio therapy equipment under complex magnetic field interference according to claim 1, characterized in that, S4 includes the following steps: The frequency data of the radio waves output by the radio therapy equipment is collected, and the output frequency values ​​at different times are obtained through a high-precision frequency counter to form a frequency time sequence. The frequency offset is calculated by comparing the collected frequency time series with the rated operating frequency of the equipment to obtain the frequency offset data at each moment, and statistically analyzing the maximum, minimum, and fluctuation range of the frequency offset. A frequency offset error objective function is constructed, and a correlation model is established between the frequency offset, distortion characteristic parameters, and interference signal amplitude by combining the magnetic field time series distortion feature set and the decoupled interference signal. The objective function is solved by the equipment frequency offset intelligent convergence calibration algorithm to obtain the frequency offset calibration amount. The frequency control parameters of the equipment are adjusted according to the calibration amount to perform dynamic calibration of the frequency offset.

9. The method for calibrating errors in radio therapy equipment under complex magnetic field interference according to claim 1, characterized in that, S5 includes: Common parameters are extracted. Common parameters that are not affected by equipment model and working environment are selected from the magnetic field time-series distortion feature set obtained from S2 to form a common parameter set. Localized parameter adaptation is performed by combining the specific model, rated power, and operating frequency range of the radio therapy equipment with the environmental parameters such as the magnetic field strength and interference source type of the usage environment. The common parameters are then adaptively adjusted. Feature data and interference signals are integrated, and the adapted localized parameters are fused with the single-source interference signal decoupled from S3 and the frequency offset calibration model parameters obtained from S4 to form a multi-dimensional calibration parameter matrix. The calibration parameter set is optimized by using principal component analysis to reduce the dimensionality of the multi-dimensional calibration parameter matrix, retaining key calibration parameters, removing redundant information, and generating the final equipment error calibration parameter set.

10. A method for calibrating errors in radio therapy equipment under complex magnetic field interference according to any one of claims 1-9, characterized in that, This method is achieved through a radio therapy equipment error calibration system under complex magnetic field interference, which includes: a magnetic field time series data acquisition and preprocessing unit, a multi-source electromagnetic interference separation and decoupling unit, a magnetic field time series distortion feature extraction unit, an equipment frequency offset intelligent convergence calibration unit, a therapy radio wave error tracing and analysis unit, and a calibration parameter adaptation and execution unit. The magnetic field timing data acquisition and preprocessing unit is connected to the multi-source electromagnetic interference separation and decoupling unit and the magnetic field timing distortion feature extraction unit via a high-speed data bus. This unit is used to acquire magnetic field timing data and mixed interference signals and transmit them to the corresponding processing units. The multi-source electromagnetic interference separation and decoupling unit is connected to the equipment frequency offset intelligent convergence calibration unit via a signal interface, transmitting the separated single-source interference signal to the equipment frequency offset intelligent convergence calibration unit. The magnetic field timing distortion feature extraction unit is connected to the equipment frequency offset intelligent convergence calibration unit via a data link, providing it with a magnetic field timing distortion feature set. The equipment frequency offset intelligent convergence calibration unit communicates bidirectionally with the physiotherapy wave error tracing analysis unit, receiving error tracing analysis results and feeding back calibration progress data. The physiotherapy wave error tracing analysis unit is connected to the calibration parameter adaptation and execution unit via a control bus, transmitting error tracing data to the calibration parameter adaptation and execution unit. The calibration parameter adaptation and execution unit is connected to the control system of the radio-controlled physiotherapy equipment, dynamically adjusting the equipment operating parameters based on the calibration parameter set. All units work collaboratively to calibrate equipment errors under complex magnetic field interference.

Images