Method based on a magnetic resonance type wireless charging system

Abstract

This invention belongs to the field of wireless charging technology, specifically disclosing a method for a magnetic resonance wireless charging system. The method includes: collecting real-time temperature difference data using a high-temperature resistant miniature temperature sensor array arranged around the curved surface of a magnetic resonance coil, and integrating environmental noise signals as an initial multidimensional parameter set to obtain a preliminary temperature difference dataset; based on the preliminary temperature difference dataset, applying a support vector machine classification method to separate the difference signals between curved surface regions, and fusing preset coordinate points in the curved surface geometric parameter correlation model to determine the regional sensitivity distribution characteristics; and using a Kalman filter algorithm to perform temporal smoothing processing on the regional sensitivity distribution characteristics to calibrate the offset caused by the inaccurate matching between the temperature difference data and the curved surface geometry, thereby obtaining a calibrated sensitivity distribution. The purpose of this invention is to solve the problem that existing technologies cannot accurately identify and adapt to the impact of temperature distribution differences on frequency matching under dynamic temperature changes in complex curved surface coils.

Description

Technical Field

[0001] This invention relates to the field of wireless charging technology, and more specifically to a method based on a magnetic resonance wireless charging system. Background Technology

[0002] In the field of wireless charging technology, magnetic resonance wireless charging systems have attracted much attention due to their high efficiency and long transmission distance. This technology has demonstrated significant application value in various scenarios such as electric vehicles, portable devices, and medical devices, becoming a key direction for driving innovation in energy transmission. Through the principle of magnetic resonance, the system can achieve efficient energy transfer without direct contact, providing users with a convenient and safe charging experience.

[0003] However, current technologies still face some deep-seated challenges in practical applications. Many existing methods often struggle to adapt to dynamic changes in complex environments, especially when the surface shape of the charging device is complex or the operating conditions are unstable, making the system performance susceptible to interference. Particularly when uneven temperature distribution on the coil surface is involved, existing solutions often fail to accurately address the resulting efficiency degradation, leading to poor charging performance and even safety hazards. A deeper technical challenge lies in how to handle the impact of coil surface temperature differences on system performance. Temperature differences not only alter the physical properties of the coil material but also further interfere with the stability of energy transfer. Because the coil surface often exhibits a complex curved shape, the temperature distribution in different areas is extremely uneven, and this unevenness directly affects the frequency matching effect of energy transfer. If the frequency cannot be adjusted in time to adapt to these changes, the overall efficiency of the system will significantly decrease.

[0004] For example, in electric vehicle charging scenarios, the coil surface may generate localized high temperatures due to prolonged operation. If the temperature in one area is too high while the temperature in other areas is low, the system may not be able to accurately identify this difference, leading to frequency adjustment errors and ultimately affecting charging speed and equipment safety.

[0005] Therefore, accurately identifying and adapting to the impact of temperature distribution differences on frequency matching under dynamic temperature variations in complex curved coils has become a key problem that urgently needs to be solved in this research. Solving this problem will directly affect the stability and reliability of the magnetic resonance wireless charging system in practical applications. Summary of the Invention

[0006] This invention provides a method based on a magnetic resonance wireless charging system, aiming to solve the problem that existing technologies cannot accurately identify and adapt to the impact of temperature distribution differences on frequency matching under the dynamic temperature changes of complex curved coils.

[0007] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A method based on a magnetic resonance wireless charging system includes: collecting real-time temperature difference data by arranging an array of high-temperature resistant miniature temperature sensors around the curved surface of a magnetic resonance coil, and integrating environmental noise signals as an initial multidimensional parameter set to obtain a preliminary temperature difference dataset; based on the preliminary temperature difference dataset, applying a support vector machine classification method to separate the difference signals between curved surface regions, and fusing preset coordinate points in the curved surface geometric parameter correlation model to determine the regional sensitivity distribution characteristics; using a Kalman filter algorithm to perform time-series smoothing processing on the regional sensitivity distribution characteristics to calibrate the offset caused by the imprecise matching between the temperature difference data and the curved surface geometry, obtaining a calibrated sensitivity distribution; and extracting multidimensional parameters from the calibrated sensitivity distribution. The parameter fusion vector is injected into the acquisition circuit to determine the compensation coefficient. If the fusion vector exceeds a preset threshold, the reading calibration module is activated to obtain the deviation correction reading. The deviation correction reading is input into the model continuous update mechanism to track the temperature difference shift and generate an updated distribution model to determine the shift compensation parameters. For the shift compensation parameters, the changes in coil material characteristics during charging are monitored, and shape variations are matched. The final frequency band adjustment scheme is obtained through iterative fusion reading fusion process. It is determined whether the deviation calculation value in the final frequency band adjustment scheme is lower than the threshold. If it is lower than the threshold, the compensation process for complex curved surface coils is output. Otherwise, the multi-dimensional calibration steps are backtracked to reset the reading fusion and obtain the optimized compensation output.

[0008] In one aspect of this disclosure, the method of acquiring real-time temperature difference data by means of a high-temperature resistant miniature temperature sensor array arranged around the curved surface of a magnetic resonance coil, and integrating environmental noise signals as an initial multidimensional parameter set to obtain a preliminary temperature difference dataset, includes: By distributing a high-temperature resistant miniature temperature sensor array along the curved surface of a magnetic resonance coil, real-time temperature difference data is collected and fused with environmental noise signals to construct an initial multidimensional parameter set, thus obtaining a preliminary temperature difference dataset.

[0009] A pre-established sensor calibration model is used to standardize the collected temperature difference data, eliminate the deviation between devices, and determine the calibrated temperature difference data set.

[0010] Based on the calibrated temperature difference data set, the fluctuation characteristics of the environmental noise signal are extracted. If the fluctuation characteristics exceed the preset threshold, the noise signal is filtered to obtain the filtered noise data set.

[0011] The filtered noise data set and the calibrated temperature difference data set are fused to generate a comprehensive multi-dimensional parameter set, and the integrity of the fused data is then determined.

[0012] Obtain key temperature difference distribution features from a comprehensive multidimensional parameter set, perform hierarchical classification based on the distribution features, and determine the classified temperature difference feature set.

[0013] Based on the classified temperature difference feature set, the support vector machine algorithm is applied to perform pattern recognition on the data to obtain the pattern distribution results of the temperature difference data.

[0014] Based on the pattern distribution results, a structured storage format for the preliminary temperature difference dataset is constructed to obtain the final temperature difference dataset.

[0015] In one aspect of this disclosure, the step of separating the difference signals between surface regions using a support vector machine classification method based on the preliminary temperature difference dataset, and fusing preset coordinate points in the surface geometric parameter correlation model to determine the regional sensitivity distribution characteristics, includes: The initial temperature difference data is obtained, and the data is standardized through preprocessing to obtain a normalized temperature difference dataset.

[0016] For a normalized temperature difference dataset, a support vector machine classification method is used to separate the signal differences between curved regions and determine the set of signal differences after classification.

[0017] Based on the classified signal difference set, combined with the geometric parameters of the curved surface region, the corresponding preset coordinate points are extracted from the preset association model to obtain the region-related coordinate point set.

[0018] By analyzing the correspondence between the surface region and the region sensitivity through the set of region-related coordinate points, we can determine the preliminary characteristics of the region sensitivity distribution.

[0019] If the initial features are not uniformly distributed, the geometric parameters of the curved surface structure are processed in layers to obtain the layered geometric parameter data and determine the refined features of the regional sensitivity distribution.

[0020] Based on the refined characteristics and the correspondence between the signal difference set and the temperature difference data, the overall consistency of the distribution characteristics is analyzed to obtain the final regional sensitivity distribution results.

[0021] In one aspect of this disclosure, the step of employing a Kalman filter algorithm to perform time-series smoothing processing on the regional sensitivity distribution characteristics to calibrate the offset caused by the imprecise matching between the temperature difference data and the surface geometry, thereby obtaining a calibrated sensitivity distribution, includes: Initial regional sensitivity distribution data is obtained, and the raw temperature difference data and surface geometry information extracted from the acquisition device are preliminarily processed to obtain uncalibrated sensitivity distribution records.

[0022] The Kalman filter algorithm is used to perform time-series smoothing on uncalibrated sensitivity distribution records. Recursive estimation is performed on data fluctuations at consecutive time points to determine the smoothed distribution sequence.

[0023] By analyzing the matching offset between temperature difference data and surface geometry information through the smoothed distribution sequence, the time periods with large offsets are marked to obtain an offset mark set.

[0024] Based on the offset marker set, a calibration process is performed on the data within the marked time period to adjust the correspondence between the temperature difference data and the surface geometry information, resulting in a calibrated matching dataset.

[0025] The updated values ​​of the regional sensitivity distribution are extracted from the calibrated matching dataset. The updated values ​​are compared with the original records to determine the changing trend of the distribution characteristics.

[0026] Based on the changing trends of the distribution characteristics, the final sensitivity distribution result is generated, and outliers in the changing trends are smoothed and corrected to obtain calibrated sensitivity distribution data.

[0027] In one aspect of this disclosure, the step of extracting a multidimensional parameter fusion vector through the calibrated sensitivity distribution and injecting a compensation coefficient into the acquisition circuit for judgment, and activating the reading calibration module if the fusion vector exceeds a preset threshold to obtain a deviation correction reading, includes: By using sensitivity distribution data, the multidimensional parameter values ​​are processed using parameter extraction methods to obtain a preliminary set of multidimensional parameters.

[0028] Based on the initial set of multidimensional parameters, a fusion vector group is generated using vector calculation formulas, and the specific numerical range of the fusion vector is determined.

[0029] For the data of the fused vector group, a comparison is made with a preset threshold line. If the value of the fused vector group exceeds the preset threshold line, the reading calibration method is triggered to obtain the calibrated reading data.

[0030] After the data is processed by the reading calibration method, the calibration module group is activated to calculate the deviation correction value and obtain the corrected reading result.

[0031] Based on the deviation correction value, adjust the configuration of the compensation coefficient item to generate new compensation coefficient data.

[0032] For the new compensation coefficient data, the circuit injection method is used to load it into the acquisition circuit board, thereby updating the parameters of the acquisition circuit board.

[0033] By collecting the updated parameters of the circuit board, the sensitivity distribution data is re-acquired, forming a closed-loop processing flow.

[0034] In one aspect of this disclosure, the mechanism for continuously updating the model by inputting the deviation correction reading to track temperature difference shifts and generating an updated distribution model, and determining shift compensation parameters, includes: By collecting temperature difference offset data in the environment and using sensors to record real-time changes, an initial offset dataset is obtained.

[0035] Based on the initial offset dataset, a pre-built support vector machine model is used to classify the offset variation information contained therein, thereby determining the category distribution of the offset variation.

[0036] Obtain the distribution of deviation changes after classification, build corresponding tracking mechanisms for different categories, and if the deviation change of a certain category exceeds a preset threshold, trigger the continuous update process to obtain the updated deviation tracking results.

[0037] Based on the updated deviation tracking results, and considering the temperature difference shift trend they reflect, a corrected distribution logic is constructed, and a corresponding distribution model framework is generated.

[0038] Based on the generated distribution model framework and combined with continuously updated data input, the internal parameters of the model are adjusted to determine a corrected distribution model suitable for the current temperature difference shift.

[0039] Obtain the output of the modified distribution model, calculate the corresponding offset compensation parameters based on the offset characteristics reflected therein, and obtain the final compensation parameter values.

[0040] By using the final compensation parameter values, parameter calibration is performed on the input model to complete the dynamic compensation process for temperature difference offset.

[0041] In one aspect of this disclosure, the step of monitoring changes in coil material properties during charging and matching shape variations to obtain a final frequency band adjustment scheme through an iterative fusion reading fusion process for the offset compensation parameters includes: By collecting data in real time on the changes in the properties of the coil material during the charging process, initial monitoring data is obtained. Preliminary screening is performed using a preset threshold range to obtain the original dataset that meets the criteria.

[0042] Based on the filtered raw dataset, a comparative analysis is performed on the correlation between changes in coil material properties and shape variations. If a property change is detected to exceed a preset range, the corresponding shape variation data is recorded to determine the distribution of anomalies.

[0043] By further processing the distribution of outliers and combining it with offset compensation parameters, the compensated corrected data is obtained. The corrected data and the reading data are then integrated using an iterative fusion method to obtain a unified feature dataset.

[0044] Based on a unified feature dataset, the key factors affecting frequency band adjustment during the charging process are analyzed. If the correlation between the key factors and shape variation is higher than a preset threshold, the compensation parameters are adjusted to determine the initial direction of frequency band adjustment.

[0045] By refining the initial frequency band adjustment direction, an adjustment range matching the changes in coil material characteristics is obtained. The adjustment range is then optimized using a support vector machine algorithm to obtain the final frequency band adjustment scheme.

[0046] For the final frequency band adjustment scheme, the execution parameters of the adjustment scheme are obtained by combining the feature dataset after iterative fusion. If the execution parameters deviate significantly from the preset target, the data of the parameter monitoring stage is backtracked to determine whether there are any unidentified influencing factors.

[0047] By conducting secondary analysis of the retrospective data, potential influencing factors are identified. Based on the data analysis results, the frequency band adjustment scheme is fine-tuned to obtain an optimized execution scheme.

[0048] In one aspect of this disclosure, determining whether the deviation calculation value in the final frequency band adjustment scheme is lower than a threshold, and if it is lower than the threshold, outputting a compensation process for the complex curved surface coil; otherwise, backtracking the multi-dimensional calibration steps to reset the reading fusion and obtain the optimized compensation output, includes: The initial frequency band adjustment data is obtained, and preliminary processing is performed on the complex curved surface coil design. By comparing with the preset threshold range, the preliminary results of the deviation calculation are judged.

[0049] If the deviation calculation result is lower than the preset threshold, the compensation process module is used to correct the parameters for the specific structure of the complex curved surface coil and obtain the adjusted compensation data.

[0050] If the deviation calculation result is higher than the preset threshold, the backtracking step is triggered, and the multidimensional calibration module is called to recalibrate the original readings and determine the calibrated data set.

[0051] By using the calibrated dataset, a reading fusion operation is performed, integrating data sources from multiple dimensions to obtain a fused, optimized dataset.

[0052] Based on the fused optimized dataset, the support vector machine algorithm is applied to predict the compensation parameters of the complex curved surface coil, and the predicted compensation scheme is obtained.

[0053] Based on the predicted compensation scheme and the requirements of the final scheme, parameter mapping and adjustment are performed to determine the final frequency band adjustment output result.

[0054] By adjusting the final frequency band output, a complete compensation process document for complex curved surface coils is generated, thus achieving the business objectives.

[0055] Compared with the prior art, the present invention has the following beneficial effects: This invention employs a high-temperature resistant miniature temperature sensor array arranged around the curved surface of a coil to collect temperature difference data in real time and integrate noise signals to form a preliminary dataset. Then, a support vector machine is applied to classify and separate regional difference signals, and the surface geometric parameters are fused to determine the sensitivity distribution characteristics. Next, a Kalman filter algorithm is used to perform temporal smoothing calibration on the features, extracting multi-dimensional fusion vectors and injecting compensation coefficients. If the deviation exceeds a threshold, a reading calibration module is activated to obtain a deviation correction reading, and a continuous update mechanism tracks the offset to generate an updated distribution model. To address coil shape variations during charging, this invention iteratively fuses the readings, monitors the offset compensation parameters, and finally outputs a frequency band adjustment scheme. If the deviation is below a threshold, compensation is completed; otherwise, backtracking optimization is performed. This invention effectively improves temperature management and charging stability of complex curved surface coils, ensuring efficient and safe wireless power transmission. Attached Figure Description

[0056] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained from these drawings without creative effort.

[0057] Figure 1 This is one of the flowcharts for the method of the magnetic resonance wireless charging system of the present invention.

[0058] Figure 2 This is the second flowchart of the method based on the magnetic resonance wireless charging system of the present invention.

[0059] Figure 3 This is the third flowchart of the method based on the magnetic resonance wireless charging system of the present invention. Detailed Implementation

[0060] The present invention will be further described below with reference to embodiments. These embodiments are merely some, not all, of the embodiments described. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the protection scope of the present invention.

[0061] Please see Figure 1-3 As shown, this embodiment discloses a method based on a magnetic resonance wireless charging system, the specific steps of which include: S1. Real-time temperature difference data is collected by arranging an array of high-temperature resistant miniature temperature sensors around the curved surface of the magnetic resonance coil, and environmental noise signals are integrated as an initial multidimensional parameter set to obtain a preliminary temperature difference dataset. S1-1. A high-temperature resistant miniature temperature sensor array is distributed along the curved surface of the magnetic resonance coil to collect real-time temperature difference data. The data is then fused with environmental noise signals to construct an initial multidimensional parameter set, thus obtaining a preliminary temperature difference dataset.

[0062] S1-2. Using a pre-established sensor calibration model, the collected temperature difference data is standardized to eliminate deviations between devices and determine the calibrated temperature difference data set.

[0063] S1-3. Based on the calibrated temperature difference data set, extract the fluctuation characteristics of the environmental noise signal. If the fluctuation characteristics exceed the preset threshold, filter the noise signal to obtain the filtered noise data set.

[0064] S1-4. The filtered noise data set and the calibrated temperature difference data set are fused to generate a comprehensive multi-dimensional parameter set, and the integrity of the fused data is judged.

[0065] S1-5. Obtain the key temperature difference distribution characteristics in the comprehensive multidimensional parameter set, perform hierarchical classification based on the distribution characteristics, and determine the temperature difference feature set after classification.

[0066] S1-6. Based on the classified temperature difference feature set, the support vector machine algorithm is applied to perform pattern recognition on the data to obtain the pattern distribution results of the temperature difference data.

[0067] S1-7. Based on the pattern distribution results, construct a structured storage format for the preliminary temperature difference dataset to obtain the final temperature difference dataset.

[0068] As an optional implementation, in this embodiment, for example, a high-temperature resistant miniature temperature sensor array is constructed and arranged around the curved surface of the magnetic resonance coil. Each sensor has a diameter of 2.5 mm, a temperature resistance range of -40℃ to 200℃, and an accuracy of ±0.1℃. The array contains a total of 64 sensors, which are evenly distributed in an 8×8 grid on the curved surface. Real-time temperature difference data is collected at a frequency of 10 times per second, and the data is transmitted to the central processing unit via a wireless transmission module using the Bluetooth 5.0 protocol.

[0069] Next, the environmental noise signal is integrated. A high-sensitivity microphone array with a sensitivity of -26 dB and a sampling rate of 44.1 kHz is used to collect noise data in the environment. The time-domain signal is converted into a frequency-domain signal through a fast Fourier transform algorithm. The main noise components in the frequency range of 20-2000 Hz are extracted, and their average power spectral density is calculated as the initial noise parameter.

[0070] Based on this, the temperature difference data and noise parameters are combined to form an initial multidimensional parameter set. The principal component analysis algorithm is used to reduce the dimensionality of the temperature difference data of 64 sensors, and the first three principal components are extracted with contribution rates of 60%, 25% and 10%, respectively. The noise power spectral density is used as the fourth dimension parameter to form a four-dimensional dataset. Subsequently, for the construction of the preliminary temperature difference dataset, the Kalman filter algorithm was used to denoise the temperature difference data. In the filtering parameters, the state transition matrix was set to the identity matrix, the measurement noise covariance was set to 0.01, and the process noise covariance was set to 0.001, resulting in smoothed temperature difference data. Combined with the noise parameters, the correlation between temperature difference and noise was analyzed through a linear regression model. The correlation coefficient was calculated to be 0.75, indicating that noise has a significant impact on temperature difference. Finally, a structured dataset containing timestamps, temperature difference values, and noise power spectral density was generated, providing a foundation for subsequent analysis.

[0071] The above process is run automatically by the embedded system. The acquisition, processing and integration of sensor data and noise signals are all completed by the preset program, ensuring the real-time performance and accuracy of data processing.

[0072] S2. Based on the preliminary temperature difference dataset, the support vector machine classification method is applied to separate the difference signals between surface regions, and the preset coordinate points in the surface geometric parameter association model are fused to determine the regional sensitivity distribution characteristics. S2-1. Obtain the initial temperature difference data, and standardize the data through preprocessing methods to obtain a standardized temperature difference dataset.

[0073] S2-2. For the normalized temperature difference dataset, the support vector machine classification method is used to separate the signal differences between the curved regions and determine the set of signal differences after classification.

[0074] S2-3. Based on the classified signal difference set and combined with the geometric parameters of the curved surface region, extract the corresponding preset coordinate points from the preset association model to obtain the set of region-related coordinate points.

[0075] S2-4. By analyzing the correspondence between the surface region and the regional sensitivity through the set of coordinate points related to the region, we can determine the preliminary characteristics of the regional sensitivity distribution.

[0076] S2-5. If the distribution of the initial features is not uniform, the geometric parameters of the curved surface structure are processed in layers to obtain the layered geometric parameter data and determine the refined features of the regional sensitivity distribution.

[0077] S2-6. Based on the refined characteristics and the correspondence between the signal difference set and the temperature difference data, analyze the overall consistency of the distribution characteristics to obtain the final regional sensitivity distribution results.

[0078] As an optional implementation method, in this embodiment, for example: applying the support vector machine classification method to separate the difference signals between surface regions and fusing the preset coordinate points in the surface geometric parameter association model to determine the regional sensitivity distribution characteristics for the preliminary temperature difference dataset, the specific implementation method is as follows: First, assume that we have a preliminary dataset containing 1000 temperature difference data points, each data point contains a temperature value, such as 25.5℃ and the corresponding surface position coordinates (x, y, z), with the data range being 0 to 10 on the x-axis, 0 to 10 on the y-axis, and 0 to 5 on the z-axis.

[0079] Then, the temperature difference data was classified using the support vector machine algorithm. The radial basis function kernel function was used, with the penalty parameter C set to 1.0 and the kernel parameter gamma set to 0.1. The data was divided into high difference regions with temperature differences greater than 2.0℃ and low difference regions with temperature differences less than 2.0℃. The classification results showed that the high difference regions accounted for about 30%.

[0080] Next, the differential signals after classification were extracted, and combined with the surface geometric parameter model, assuming that the surface is defined by the quadratic equation z=0.2x²+0.3y², the curvature value of each data point was calculated. For example, the curvature at the point (2,3,1.7) is about 0.4. The curvature was then correlated with the temperature difference signal to construct a regression model. It was found that the temperature difference signal was significantly enhanced in the region with a curvature greater than 0.3, with a correlation coefficient of 0.85.

[0081] Subsequently, based on 10 key points such as (1,1,0.2) and (5,5,1.3), the sensitivity distribution of the entire surface is calculated using an interpolation algorithm, such as Kriging interpolation. The sensitivity values ​​range from 0.1 to 0.9, with the highest sensitivity of 0.88 near coordinates (5,5,1.3).

[0082] Finally, by combining the sensitivity distribution characteristics with the surface geometric parameters, the analysis showed that the high sensitivity areas are mainly concentrated in the central part of the surface with large curvature, forming a complete logical chain from data classification to feature distribution. This verified the strong correlation between temperature difference and geometry, providing data support for subsequent optimization design.

[0083] S3. Using the Kalman filter algorithm, the regional sensitivity distribution characteristics are smoothed over time to calibrate the offset caused by the inaccurate matching between the temperature difference data and the surface geometry, and then the calibrated sensitivity distribution is obtained. S3-1. Obtain initial regional sensitivity distribution data. Perform preliminary processing on the raw temperature difference data and surface geometry information extracted from the acquisition device to obtain uncalibrated sensitivity distribution records.

[0084] S3-2. The Kalman filter algorithm is used to perform time-series smoothing on the uncalibrated sensitivity distribution records. The data fluctuations at continuous time points are recursively estimated to determine the smoothed distribution sequence.

[0085] S3-3. Analyze the matching offset between temperature difference data and surface geometry information through the smoothed distribution sequence, and highlight the time periods with large offsets to obtain the offset mark set.

[0086] S3-4. Based on the offset marker set, perform calibration processing on the data within the marked time period to adjust the correspondence between the temperature difference data and the surface geometry information, and obtain the calibrated matching dataset.

[0087] S3-5. Extract the updated values ​​of the regional sensitivity distribution from the calibrated matching dataset, compare and analyze the updated values ​​with the original records, and determine the changing trend of the distribution characteristics.

[0088] S3-6. Based on the changing trend of the distribution characteristics, generate the final sensitivity distribution result, and perform smoothing correction on outliers in the changing trend to obtain calibrated sensitivity distribution data.

[0089] As an optional implementation, in this embodiment, for example, when performing time-series smoothing on the regional sensitivity distribution characteristics to calibrate the offset caused by the inaccurate matching between temperature difference data and surface geometry, accurate calibration can be achieved through the Kalman filter algorithm.

[0090] First, suppose we have sensitivity distribution data for a region, with an initial sensitivity value of 10.5, a measurement noise variance of 0.2, a process noise variance of 0.1, a temperature difference data collected by sensors of 12.3, and a predicted value of 11.8 by the surface geometry model, indicating a certain offset.

[0091] The first step in Kalman filtering is to initialize the state estimate and covariance matrix. The initial state estimate is set to 10.5 and the initial covariance is set to 1.0 to describe the uncertainty.

[0092] Next, in the prediction phase, using the state transition matrix, assumed to be an identity matrix of 1.0 and a process noise covariance of 0.1, the predicted state estimate for the next time step is 10.5, and the prediction covariance is 1.1.

[0093] Then, in the update phase, the Kalman gain is calculated, and based on the measurement noise variance of 0.2 and the prediction covariance of 1.1, the gain value is 0.846.

[0094] Based on the actual measured value of 12.3, the updated state estimate is 11.7, and the updated covariance is 0.154, completing one filtering iteration.

[0095] Through multiple iterations, such as 10 consecutive time steps, assuming the measured value fluctuates between 12.0 and 12.5, the final smoothed sensitivity distribution value is 11.9, which is closer to the true temperature difference data than the initial value and reduces the impact of noise.

[0096] To address the offset caused by inaccurate matching of surface geometry, assuming the difference between the predicted value 11.8 and the filtered value 11.9 is 0.1, this offset is evenly distributed to each grid point on the surface using a linear interpolation algorithm. For example, if the total number of grid points is 100, the calibration amount for each grid point is 0.001, generating a calibrated sensitivity distribution matrix. The final matrix value ranges from 11.89 to 11.91, thus completing the calibration.

[0097] This process, through automatic calculation using algorithms and combined with temperature difference data and geometric model deviation analysis, ensures the temporal smoothness of distribution characteristics and calibration accuracy, while also being linked to business needs, such as being used to optimize the area detection accuracy of thermal imaging systems.

[0098] S4. Extract the multi-dimensional parameter fusion vector through the calibrated sensitivity distribution, and inject the compensation coefficient into the acquisition circuit for judgment. If the fusion vector exceeds the preset threshold, activate the reading calibration module to obtain the deviation correction reading. S4-1. Using the sensitivity distribution data, the parameter extraction method is used to process the multidimensional parameter values ​​to obtain a preliminary multidimensional parameter set.

[0099] S4-2. Based on the preliminary multidimensional parameter set, apply vector calculation formula to generate a fusion vector group and determine the specific numerical range of the fusion vector.

[0100] S4-3. For the data of the fused vector group, compare it with the preset threshold line. If the value of the fused vector group exceeds the preset threshold line, trigger the reading calibration method and obtain the calibrated reading data.

[0101] S4-4. After processing the data using the reading calibration method, activate the calibration module group, calculate the deviation correction value, and obtain the corrected reading result.

[0102] S4-5. Based on the deviation correction value, adjust the configuration of the compensation coefficient item to generate new compensation coefficient data.

[0103] S4-6. For the new compensation coefficient data, the circuit injection method is used to load it into the acquisition circuit board to complete the parameter update of the acquisition circuit board.

[0104] S4-7. By collecting the updated parameters of the circuit board, the sensitivity distribution data is re-acquired to form a closed-loop processing flow.

[0105] As an optional implementation, in this embodiment, for example, in the process of extracting the multidimensional parameter fusion vector from the sensitivity distribution after calibration, environmental data, such as temperature, humidity, and pressure, are first collected through a sensor array. Assuming the collected raw data are a temperature of 25.3°C, humidity of 60.2%, and pressure of 101.2 kPa, these data are normalized using a standardization algorithm, and the calculation formula is X. norm =(XX min ) / (X max -X min The normalized values ​​obtained were 0.51, 0.60 and 0.55, respectively. Then, the eigenvectors were extracted by principal component analysis, and the weight matrix of each parameter was calculated to obtain the fusion vector V=[0.51,0.60,0.55], with a Euclidean norm of 0.98, which was used for subsequent analysis.

[0106] Next, a compensation coefficient is injected into the acquisition circuit. Assuming the preset compensation coefficient is 0.95, the signal offset after injection is calculated using circuit simulation software, with the formula being S. comp =S raw • 0.95, the original signal value is 10.5 volts, after compensation it is 9.975 volts, and the signal stability is monitored in real time to ensure that the error is controlled within 0.1 volts.

[0107] Then, it is determined whether the fusion vector exceeds the preset threshold. Assuming the threshold is 0.90, the calculated result 0.98 is greater than 0.90, and the system automatically triggers the calibration mechanism to proceed to the next step.

[0108] If the threshold is not exceeded, the current state is recorded and monitored cyclically.

[0109] Finally, the reading calibration module is activated. Through historical data regression analysis, assuming the deviation model is D=0.02·V+0.01, the deviation value of 0.0296 is calculated. The corrected reading is the original reading minus the deviation value. For example, if the original reading is 100.5, the corrected reading is 100.4704. At the same time, the correction result is stored in the database to form a closed-loop feedback to ensure continuous optimization of system accuracy.

[0110] The above process is executed automatically by the embedded system, logically forming a complete chain from data acquisition to feature extraction, compensation adjustment, threshold judgment and deviation correction, ensuring the rigor and reliability of the technology implementation.

[0111] S5. Input the deviation correction reading into the model continuous update mechanism to track the temperature difference shift, generate the updated distribution model, and determine the shift compensation parameters. S5-1. By collecting temperature difference offset data in the environment and using sensors to record real-time changes, an initial offset dataset is obtained.

[0112] S5-2. Based on the initial offset dataset, classify the offset change information contained therein using a pre-established support vector machine model to determine the category distribution of offset changes.

[0113] S5-3. Obtain the distribution of deviation changes after classification, build corresponding tracking mechanisms for different categories, and determine if the deviation change of a certain category exceeds the preset threshold. Then, trigger the continuous update process to obtain the updated deviation tracking results.

[0114] S5-4. Based on the updated deviation tracking results, construct the corrected distribution logic and generate the corresponding distribution model framework according to the temperature difference shift trend reflected by the results.

[0115] S5-5. Based on the generated distribution model framework and combined with continuously updated data input, adjust the internal parameters of the model to determine the corrected distribution model suitable for the current temperature difference shift.

[0116] S5-6. Obtain the output results of the modified distribution model, calculate the corresponding offset compensation parameters based on the offset characteristics reflected therein, and obtain the final compensation parameter values.

[0117] S5-7. Using the final compensation parameter values, perform parameter calibration on the input model to complete the dynamic compensation processing for temperature difference offset.

[0118] As an optional implementation, in this embodiment, for example, the following specific methods can be used to technically process the continuous updating mechanism of the deviation correction reading input model to track temperature difference shifts and generate an updated distribution model to determine the shift compensation parameters.

[0119] First, let's assume we collect ambient temperature data through sensors, recording it once per hour. The initial data is 25.5℃, while the standard operating temperature of the equipment is 25.0℃, so the initial deviation is calculated to be 0.5℃.

[0120] The system automatically inputs the deviation value into a time series-based deviation tracking algorithm, such as using the sliding window averaging method with a window size of 24 hours to calculate the average deviation value over the past 24 hours. Assuming the average deviation is 0.45℃, it indicates that the temperature difference shift has slight fluctuations.

[0121] Next, the system uses the average deviation to update the input model and uses the Kalman filter algorithm to dynamically correct the temperature reading. The process noise covariance of the filter is set to 0.01 and the measurement noise covariance is set to 0.05. Through iterative calculation, the corrected reading at the current moment is obtained as 25.1℃, which is closer to the standard value than the original reading.

[0122] Subsequently, the system generates an updated distribution model based on the corrected readings. Assuming that the temperature data follows a normal distribution, the system calculates the distribution parameters through maximum likelihood estimation, resulting in a new distribution model with a mean of 25.1℃ and a standard deviation of 0.2℃.

[0123] Finally, the system determines the offset compensation parameters based on the distribution model, sets the compensation formula to the difference between the mean and the standard value, i.e., 25.1 minus 25.0 equals 0.1℃, and automatically applies this parameter to subsequent temperature readings to ensure that the equipment operates within a stable range.

[0124] The above process forms a closed-loop logic through sensor data acquisition, algorithm processing, and parameter updates. It is also linked to the equipment operation status monitoring business. If the temperature still exceeds the tolerance range of ±0.3℃ after compensation, the system will automatically trigger an alarm signal to notify the background maintenance system for further analysis, ensuring the accuracy and reliability of deviation correction.

[0125] S6. For the offset compensation parameters, monitor the changes in coil material properties during charging, and match the shape variation to obtain the final frequency band adjustment scheme through an iterative fusion reading fusion process; S6-1. By collecting data in real time on the changes in the properties of the coil material during the charging process, initial monitoring data is obtained. Preliminary screening is performed using a preset threshold range to obtain the original dataset that meets the conditions.

[0126] S6-2. Based on the filtered original dataset, conduct a comparative analysis on the correlation between the characteristic changes and shape variations of the coil material. If the characteristic changes are detected to exceed the preset range, record the corresponding shape variation data and determine the distribution of anomalies.

[0127] S6-3. By further processing the distribution of outliers and combining it with offset compensation parameters, the corrected data after compensation is obtained. The corrected data and the reading data are integrated by iterative fusion to obtain a unified feature dataset.

[0128] S6-4. Based on the unified feature dataset, analyze the key factors affecting frequency band adjustment during the charging process. If the correlation between the key factors and shape variation is higher than the preset threshold, adjust the compensation parameters to determine the initial direction of frequency band adjustment.

[0129] S6-5. By refining the initial frequency band adjustment direction, an adjustment range matching the changes in coil material characteristics is obtained. The adjustment range is then optimized using a support vector machine algorithm to obtain the final frequency band adjustment scheme.

[0130] S6-6. For the final frequency band adjustment scheme, combine the feature dataset after iterative fusion to obtain the execution parameters of the adjustment scheme. If the execution parameters deviate significantly from the preset target, perform data backtracking on the parameter monitoring stage to determine whether there are any unidentified influencing factors.

[0131] S6-7. By conducting secondary analysis of the backtracking data, potential influencing factors are obtained. Based on the data analysis results, the frequency band adjustment scheme is fine-tuned to obtain an optimized execution scheme.

[0132] As an optional implementation method, in this embodiment, for example: the specific implementation method of monitoring the changes in coil material properties and matching shape variations during the charging process for offset compensation parameters, and obtaining the final frequency band adjustment scheme through iterative fusion reading fusion process is as follows: First, the system collects electromagnetic property data of the coil material in real time through built-in sensors. For example, during the charging process, the inductance value of the coil is monitored. Assuming the initial inductance value is 10.5 microhenries, as the charging power increases from 50W to 100W, the inductance value may change to 10.8 microhenries due to the thermal effect of the material. The system uses this change of 0.3 microhenries as the offset parameter input to the analysis model, and uses a preset linear regression algorithm to calculate the relationship between offset and material stress, and obtains a stress change coefficient of 0.02, thereby judging the trend of material property changes.

[0133] Secondly, regarding the coil shape variation, the system acquires three-dimensional data of the coil geometry using a high-precision laser scanner. Assuming the initial diameter is 5.0 cm and the diameter becomes 5.1 cm after deformation, the system simulates the impact of shape variation on electromagnetic field distribution based on the geometric difference of 0.1 cm and the finite element analysis algorithm, calculating the field strength deviation to be 3.5%, and mapping this deviation to the frequency band adjustment requirements.

[0134] Then, the system enters the iterative fusion reading process, which weights and fuses the field strength effects of the inductance offset of 0.3 μH and the shape deviation of 3.5%. Using the Kalman filter algorithm, the inductance weight is set to 0.6 and the shape weight is set to 0.4. After 10 iterations, the comprehensive influence factor is obtained as 2.1%. The charging frequency band is adjusted according to this factor. Assuming the initial frequency band is 100 kHz, it is finally adjusted to 102.1 kHz to ensure a 2% improvement in charging efficiency.

[0135] Finally, the system feeds back the adjusted frequency band parameters to the control module, automatically updates the charging strategy, and records the data of each iteration to form a closed-loop optimization logic, ensuring the stability of the subsequent charging process.

[0136] Using the above methods, the system achieves fully automated processing from data acquisition to parameter adjustment, with rigorous logic and traceability.

[0137] S7. Determine whether the deviation calculation value in the final frequency band adjustment scheme is lower than the threshold. If it is lower than the threshold, output the compensation process for the complex curved surface coil. Otherwise, backtrack the multi-dimensional calibration steps to reset the reading fusion and obtain the optimized compensation output.

[0138] S7-1. Obtain initial frequency band adjustment data, perform preliminary processing for complex curved surface coil designs, and determine the preliminary results of deviation calculation by comparing with the preset threshold range.

[0139] S7-2. If the deviation calculation result is lower than the preset threshold, the compensation process module is used to correct the parameters for the specific structure of the complex curved surface coil and obtain the adjusted compensation data.

[0140] S7-3. If the deviation calculation result is higher than the preset threshold, the backtracking step is triggered, and the multi-dimensional calibration module is called to recalibrate the original readings and determine the calibrated data set.

[0141] S7-4. Using the calibrated dataset, perform a reading fusion operation, integrating data sources from multiple dimensions to obtain a fused optimized dataset.

[0142] S7-5. Based on the fused optimized dataset, the support vector machine algorithm is applied to predict the compensation parameters of the complex curved surface coil, and the predicted compensation scheme is obtained.

[0143] S7-6. Based on the predicted compensation scheme and the requirements of the final scheme, perform parameter mapping and adjustment to determine the final frequency band adjustment output result.

[0144] S7-7. By adjusting the final frequency band output, generate a complete compensation process document for complex curved surface coils to achieve the business objectives.

[0145] As an optional implementation method, in this embodiment, for example, in the process of deviation calculation and optimization compensation of frequency band adjustment scheme, firstly, regarding the judgment of deviation calculation value, assuming the preset threshold is 0.5, the system collects the difference between the actual value after frequency band adjustment and the target value. For example, if the actual frequency band value is 10.2MHz and the target value is 10.0MHz, the calculated deviation is 0.2, which is lower than the threshold of 0.5. The system automatically triggers the compensation process for complex curved surface coils.

[0146] The compensation process uses a gradient descent-based algorithm with an initial learning rate of 0.01. It iteratively calculates the non-uniformity of the electromagnetic field distribution on the coil surface, assuming an initial non-uniformity of 3.5% and a target of 1.0%. The coil current distribution parameters are adjusted in each iteration, and the change in non-uniformity after each iteration is recorded. For example, after the first iteration, it drops to 2.8%, and after the second iteration, it drops to 2.2%, until it approaches the target value. The system then outputs the compensated current distribution matrix.

[0147] If the calculated deviation is higher than the threshold, for example, if the deviation is 0.7, which is higher than 0.5, the system will automatically backtrack the multidimensional calibration steps, call historical calibration data, reset the reading fusion model, and use a weighted average algorithm to fuse the readings of multiple sensors, such as sensor A reading of 10.1MHz and sensor B reading of 10.3MHz, with weights of 0.6 and 0.4 respectively, into a new reference value of 10.18MHz. Then the deviation is recalculated. Assuming the new deviation is 0.3, which is lower than the threshold, the compensation process begins.

[0148] The entire process involves real-time monitoring of deviation changes through the system's built-in analysis module, generating optimized compensation outputs. For example, the final output adjusted frequency band value is 10.05MHz, with a deviation of only 0.05. Simultaneously, the electromagnetic field distribution optimization log is recorded during the compensation process to ensure traceability for subsequent analysis.

[0149] The above process is automated through algorithms and data-driven approaches, forming a complete logical chain from deviation judgment to compensation optimization, ensuring that the performance of complex curved surface coils meets design requirements.

[0150] 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 method based on a magnetic resonance wireless charging system, characterized in that, include: Real-time temperature difference data is collected by arranging an array of high-temperature resistant miniature temperature sensors around the curved surface of a magnetic resonance coil, and environmental noise signals are integrated as an initial multidimensional parameter set to obtain a preliminary temperature difference dataset. Based on the preliminary temperature difference dataset, the support vector machine classification method is applied to separate the difference signals between surface regions, and the preset coordinate points in the surface geometric parameter association model are fused to determine the regional sensitivity distribution characteristics. The Kalman filter algorithm is used to perform time-series smoothing on the regional sensitivity distribution characteristics to calibrate the offset caused by the imprecise matching between the temperature difference data and the surface geometry, and then the calibrated sensitivity distribution is obtained. The multidimensional parameter fusion vector is extracted from the calibrated sensitivity distribution, and a compensation coefficient is injected into the acquisition circuit for judgment. If the fusion vector exceeds the preset threshold, the reading calibration module is activated to obtain the deviation correction reading. The deviation correction readings are input into the model continuous update mechanism to track temperature difference shifts, and an updated distribution model is generated to determine the shift compensation parameters. For the offset compensation parameters, the changes in coil material properties during charging are monitored, and the shape variation is matched to obtain the final frequency band adjustment scheme through an iterative fusion reading fusion process; If the deviation calculation value in the final frequency band adjustment scheme is lower than the threshold, the compensation process for the complex curved surface coil is output; otherwise, the multi-dimensional calibration steps are traced back to reset the reading fusion and obtain the optimized compensation output.

2. The method based on a magnetic resonance wireless charging system according to claim 1, characterized in that: The process involves collecting real-time temperature difference data using a high-temperature resistant miniature temperature sensor array arranged around the curved surface of a magnetic resonance coil, and integrating environmental noise signals as an initial multidimensional parameter set to obtain a preliminary temperature difference dataset, including: By distributing a high-temperature resistant miniature temperature sensor array along the curved surface of a magnetic resonance coil, real-time temperature difference data is collected and fused with environmental noise signals to construct an initial multidimensional parameter set, thus obtaining a preliminary temperature difference dataset. The collected temperature difference data is standardized using a pre-established sensor calibration model to eliminate deviations between devices and determine the calibrated temperature difference data set. Based on the calibrated temperature difference data set, the fluctuation characteristics of the environmental noise signal are extracted. If the fluctuation characteristics exceed the preset threshold, the noise signal is filtered to obtain the filtered noise data set. By fusing the filtered noise data set with the calibrated temperature difference data set, a comprehensive multi-dimensional parameter set is generated, and the integrity of the fused data is judged. Obtain key temperature difference distribution features from a comprehensive multidimensional parameter set, perform hierarchical classification based on the distribution features, and determine the classified temperature difference feature set. Based on the classified temperature difference feature set, the support vector machine algorithm is applied to perform pattern recognition on the data to obtain the pattern distribution results of the temperature difference data. Based on the pattern distribution results, a structured storage format for the preliminary temperature difference dataset is constructed to obtain the final temperature difference dataset.

3. The method based on a magnetic resonance wireless charging system according to claim 1, characterized in that: The step of separating the difference signals between surface regions using a support vector machine classification method based on the preliminary temperature difference dataset, and fusing preset coordinate points in the surface geometric parameter correlation model to determine the regional sensitivity distribution characteristics, includes: The initial temperature difference data is obtained, and the data is standardized through preprocessing methods to obtain a normalized temperature difference dataset. For a normalized temperature difference dataset, a support vector machine classification method is used to separate the signal differences between curved regions and determine the set of classified signal differences. Based on the classified signal difference set, combined with the geometric parameters of the curved surface region, the corresponding preset coordinate points are extracted from the preset association model to obtain the set of region-related coordinate points; By analyzing the correspondence between the curved surface region and the regional sensitivity through the set of region-related coordinate points, the preliminary characteristics of the regional sensitivity distribution can be determined. If the initial features are not uniformly distributed, the geometric parameters of the curved surface structure are processed in layers to obtain the layered geometric parameter data and determine the refined features of the regional sensitivity distribution. Based on the refined characteristics and the correspondence between the signal difference set and the temperature difference data, the overall consistency of the distribution characteristics is analyzed to obtain the final regional sensitivity distribution results.

4. The method based on a magnetic resonance wireless charging system according to claim 1, characterized in that: The process employs a Kalman filter algorithm to perform time-series smoothing on the regional sensitivity distribution characteristics to calibrate the offset caused by the imprecise matching between the temperature difference data and the surface geometry, thereby obtaining the calibrated sensitivity distribution, including: Initial regional sensitivity distribution data is obtained, and the raw temperature difference data and surface geometry information extracted from the acquisition device are preliminarily processed to obtain uncalibrated sensitivity distribution records. The Kalman filter algorithm is used to perform time-series smoothing on the uncalibrated sensitivity distribution records. The data fluctuations at consecutive time points are recursively estimated to determine the smoothed distribution sequence. By analyzing the matching offset between temperature difference data and surface geometry information through the smoothed distribution sequence, the time periods with large offsets are marked to obtain an offset mark set. Based on the offset marker set, a calibration process is performed on the data within the marked time period to adjust the correspondence between the temperature difference data and the surface geometry information, resulting in a calibrated matching dataset. The updated values ​​of the regional sensitivity distribution are extracted from the calibrated matching dataset, and the updated values ​​are compared with the original records to determine the changing trend of the distribution characteristics. Based on the changing trends of the distribution characteristics, the final sensitivity distribution result is generated, and outliers in the changing trends are smoothed and corrected to obtain calibrated sensitivity distribution data.

5. The method based on a magnetic resonance wireless charging system according to claim 1, characterized in that: The process involves extracting a multidimensional parameter fusion vector from the calibrated sensitivity distribution and injecting a compensation coefficient into the acquisition circuit. If the fusion vector exceeds a preset threshold, the reading calibration module is activated to obtain a deviation-corrected reading. This includes: By using sensitivity distribution data, the multidimensional parameter values ​​are processed using parameter extraction methods to obtain a preliminary set of multidimensional parameters; Based on the initial set of multidimensional parameters, a fusion vector group is generated by applying vector calculation formulas, and the specific numerical range of the fusion vector is determined. For the data of the fused vector group, a comparison is made with a preset threshold line. If the value of the fused vector group exceeds the preset threshold line, the reading calibration method is triggered to obtain the calibrated reading data. After the data is processed by the reading calibration method, the calibration module group is activated to calculate the deviation correction value and obtain the corrected reading result. Based on the deviation correction value, adjust the configuration of the compensation coefficient item to generate new compensation coefficient data; For the new compensation coefficient data, the circuit injection method is used to load it into the acquisition circuit board to complete the parameter update of the acquisition circuit board; By collecting the updated parameters of the circuit board, the sensitivity distribution data is re-acquired, forming a closed-loop processing flow.

6. The method based on a magnetic resonance wireless charging system according to claim 1, characterized in that: The mechanism for continuously updating the model by inputting the deviation correction readings to track temperature difference shifts and generating an updated distribution model, and determining the shift compensation parameters, includes: By collecting temperature difference offset data in the environment and using sensors to record real-time changes, an initial offset dataset is obtained. Based on the initial offset dataset, the pre-built support vector machine model is used to classify the offset change information contained therein to determine the category distribution of the offset change. Obtain the distribution of deviation changes after classification, build corresponding tracking mechanisms for different categories, and if the deviation change of a certain category exceeds the preset threshold, trigger the continuous update process to obtain the updated deviation tracking results. Based on the updated deviation tracking results, and considering the temperature difference shift trend they reflect, a corrected distribution logic is constructed, and a corresponding distribution model framework is generated. Based on the generated distribution model framework and combined with continuously updated data input, the internal parameters of the model are adjusted to determine a corrected distribution model suitable for the current temperature difference shift. Obtain the output of the modified distribution model, calculate the corresponding offset compensation parameters based on the offset characteristics reflected therein, and obtain the final compensation parameter values. By using the final compensation parameter values, parameter calibration is performed on the input model to complete the dynamic compensation process for temperature difference offset.

7. The method based on a magnetic resonance wireless charging system according to claim 1, characterized in that: The process of monitoring changes in coil material properties during charging and matching shape variations to obtain the final frequency band adjustment scheme through an iterative fusion reading fusion process, for the offset compensation parameters, includes: By collecting data in real time on the changes in the properties of the coil material during the charging process, initial monitoring data is obtained. Preliminary screening is performed using a preset threshold range to obtain a raw dataset that meets the criteria. Based on the filtered raw dataset, a comparative analysis is performed on the correlation between changes in coil material properties and shape variations. If a property change is detected to exceed a preset range, the corresponding shape variation data is recorded to determine the distribution of anomalies. By further processing the distribution of outliers and combining it with offset compensation parameters, the compensated corrected data is obtained. The corrected data and reading data are then integrated using an iterative fusion method to obtain a unified feature dataset. Based on a unified feature dataset, the key factors affecting frequency band adjustment during the charging process are analyzed. If the correlation between the key factors and shape variation is higher than a preset threshold, the compensation parameters are adjusted to determine the initial direction of frequency band adjustment. By refining the initial frequency band adjustment direction, an adjustment range matching the changes in coil material properties is obtained. The adjustment range is then optimized using a support vector machine algorithm to obtain the final frequency band adjustment scheme. For the final frequency band adjustment scheme, the execution parameters of the adjustment scheme are obtained by combining the feature dataset after iterative fusion. If the execution parameters deviate significantly from the preset target, the data of the parameter monitoring link is backtracked to determine whether there are any unidentified influencing factors. By conducting secondary analysis of the retrospective data, potential influencing factors are identified. Based on the data analysis results, the frequency band adjustment scheme is fine-tuned to obtain an optimized execution scheme.

8. The method based on a magnetic resonance wireless charging system according to claim 1, characterized in that: The determination of whether the deviation calculation value in the final frequency band adjustment scheme is lower than a threshold is made. If it is lower than the threshold, the compensation process for the complex curved surface coil is output; otherwise, the multi-dimensional calibration steps are backtracked to reset the reading fusion and obtain the optimized compensation output, including: Acquire initial frequency band adjustment data, perform preliminary processing for complex curved surface coil designs, and determine the preliminary results of deviation calculation by comparing with preset threshold ranges; If the deviation calculation result is lower than the preset threshold, the compensation process module is used to correct the parameters for the specific structure of the complex curved surface coil and obtain the adjusted compensation data. If the deviation calculation result is higher than the preset threshold, the backtracking step is triggered, and the multidimensional calibration module is called to recalibrate the original readings and determine the calibrated data set. By using the calibrated dataset, a reading fusion operation is performed, integrating data sources from multiple dimensions to obtain a fused, optimized dataset. Based on the fused optimized dataset, the support vector machine algorithm is applied to predict the compensation parameters of the complex curved surface coil, and the predicted compensation scheme is obtained. Based on the predicted compensation scheme and the requirements of the final scheme, parameter mapping and adjustment are performed to determine the final frequency band adjustment output result; By adjusting the final frequency band output, a complete compensation process document for complex curved surface coils is generated, thus achieving the business objectives.

Images