An intelligent electrode patch transdermal physiotherapy system with self-adaptive adjustment of skin resistance

Abstract

The present invention relates to an intelligent electrode transdermal physiotherapy system with adaptive adjustment of skin resistance. The system includes: a collection module for performing real-time electrical parameter collection and electric field response measurement on the contact interface area of the intelligent electrode to obtain electrical parameter distribution data and electric field response distribution data; a feature analysis module for converting the electrical parameter distribution data into a first feature map and converting the electric field response distribution data into a second feature map to obtain interface characteristic data; a state classification module for performing interface state classification operations through a machine learning model with adaptive parameter adjustment to obtain an interface state classification result; an adjustment module for adjusting the output power of the intelligent electrode and controlling the substance release to obtain a predicted transmission current; and a parameter optimization module for outputting a working current that meets the conditions. The present invention reduces energy consumption, improves the substance transmission efficiency, and also improves the control accuracy of the substance migration depth.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent electrode sheet control, and particularly to an intelligent electrode sheet transdermal physiotherapy system with self - adaptive adjustment of skin resistance. Background Art

[0002] Traditional electric - field - driven mass transfer technologies usually adopt fixed parameter settings and lack the ability to specifically adjust for different interface characteristics, resulting in low mass transfer efficiency and energy waste problems. Although existing intelligent electrode sheet systems can achieve basic mass transfer functions, they generally have technical defects such as insufficient perception of changes in the electrical characteristics of the contact interface, simple feedback adjustment mechanisms, and inability to dynamically optimize output parameters according to the interface state. In addition, when traditional systems handle complex interface environments, it is difficult to control the migration depth and uniformity of substances, and manual intervention is often required to adjust parameters, making it difficult to achieve automated precise control.

[0003] Currently, the electrode sheet mass transfer systems on the market mainly rely on preset programs to run, lacking the ability to sense and analyze the real - time state of the contact interface, and unable to adaptively adjust output parameters according to different interface conditions. Common systems usually directly apply the electrode sheet to the target interface, unable to accurately identify the electrical characteristic differences of the contact interface, resulting in low mass transfer efficiency. Especially in a low - conductivity interface environment, the mass transfer resistance increases, the transmission depth is limited, and the system cannot intelligently adjust the current output parameters to adapt to this change. At the same time, the current control in existing technologies is mostly in a single mode, making it difficult to optimize the current waveform and release rate according to different interface states, resulting in uneven substance distribution and obvious edge effects, which limits the promotion of the system in high - precision application scenarios. Summary of the Invention

[0004] The main object of the present invention is to provide an intelligent electrode sheet transdermal physiotherapy system with self - adaptive adjustment of skin resistance. The present invention realizes the directional migration and efficient utilization of substances. At the same time, with the help of the interface adaptability evaluation model and high - precision electric - field analysis technology, the control accuracy of the substance migration depth is significantly improved, and the precise mass transfer control ability is achieved.

[0005] To achieve the above object, the present invention provides an intelligent electrode sheet transdermal physiotherapy system with self - adaptive adjustment of skin resistance, including:

[0006] An acquisition module, configured to perform real - time electrical parameter acquisition and electric - field response measurement on the contact interface area of the intelligent electrode sheet to obtain electrical parameter distribution data and electric - field response distribution data;

[0007] A feature analysis module, configured to convert the electrical parameter distribution data into a first feature map, convert the electric - field response distribution data into a second feature map, and perform feature analysis to obtain interface characteristic data;

[0008] A state classification module, configured to input the interface characteristic data into a machine learning model with adaptive parameter adjustment for interface state classification operation to obtain an interface state classification result;

[0009] An adjustment module, configured to determine a current output parameter and a mass transfer enhancement coefficient according to the interface state classification result, and perform output power adjustment and mass release control on the intelligent electrode sheet to obtain a predicted transmission current;

[0010] A parameter optimization module, configured to perform deviation calculation and compensation parameter optimization on the predicted transmission current, and output a working current that meets the conditions.

[0011] Optionally, in the first implementation manner of the first aspect of the present invention, the acquisition module is specifically configured to:

[0012] Perform micro-current detection operation on the interface between the intelligent electrode sheet and the contact surface to obtain a target electrical parameter map, and perform digital processing on the target electrical parameter map to obtain initial electrical parameter data;

[0013] Perform grid area statistical calculation on the initial electrical parameter data to obtain electrical parameter distribution data, and perform normalization processing and data smoothing operation on the electrical parameter distribution data to obtain interface electrical parameter distribution data;

[0014] Perform real-time sampling on the contact interface area by a multi-frequency electric field response sensor to obtain real-time electric field response data, and perform interface tomography calculation on the real-time electric field response data to obtain initial electric field response sampling data;

[0015] Perform spatial distribution statistical analysis on the initial electric field response sampling data to obtain electric field response statistical distribution data;

[0016] Perform coordinate transformation and numerical integration operation on the electric field response statistical distribution data to obtain electric field response correction data, and input the electric field response correction data into an electric field distribution reconstruction algorithm for spatial electric field distribution reconstruction to obtain electric field response distribution data.

[0017] Optionally, in the second implementation manner of the first aspect of the present invention, the feature analysis module is specifically configured to:

[0018] Perform normalization processing and statistical interval division on the interface electrical parameter distribution data to obtain an electrical parameter distribution statistical interval;

[0019] According to the electrical parameter distribution statistical interval, perform data mapping and frequency counting on the interface electrical parameter distribution data to obtain an electrical parameter frequency statistical result, and generate a first feature map according to the electrical parameter frequency statistical result;

[0020] Normalize the electric field response distribution data and divide the statistical interval to obtain the statistical interval of the electric field response distribution;

[0021] According to the statistical interval of the electric field response distribution, perform data mapping and frequency counting on the electric field response distribution data to obtain the statistical result of the electric field response frequency, and generate the second characteristic spectrum according to the statistical result of the electric field response frequency;

[0022] Perform principal component analysis on the first characteristic spectrum, extract the distribution feature vector of electrical parameters to obtain the first characteristic vector, and perform principal component analysis on the second characteristic spectrum, extract the distribution feature vector of the electric field response to obtain the second characteristic vector;

[0023] Calculate the feature weights of the first feature vector and the second feature vector respectively, and determine the weight coefficient matrix for feature fusion through covariance matrix operation;

[0024] Perform weighted fusion operation on the first feature vector and the second feature vector according to the weight coefficient matrix to obtain the interface characteristic data.

[0025] Optionally, in the third implementation manner of the first aspect of the present invention, the state classification module is specifically used for:

[0026] Separate the interface characteristic data according to the electrical parameter characteristics and the electric field response characteristics to obtain the electrical parameter characteristic data set and the electric field response characteristic data set;

[0027] Input the electrical parameter characteristic data set into the first-layer classification network of the machine learning model with adaptive parameter adjustment for electrical parameter classification operation. The first-layer classification network adopts a three-layer convolutional neural network structure to obtain the classification result of the interface electrical parameter state;

[0028] Perform feature enhancement processing on the classification result of the interface electrical parameter state, including adding a spatial attention mechanism and a channel attention mechanism. The spatial attention mechanism adopts a parallel structure of maximum pooling and average pooling, and the channel attention mechanism adopts a squeeze-and-excitation module to obtain the enhanced electrical parameter characteristics;

[0029] Fuse the electric field response characteristic data set and the enhanced electrical parameter characteristics to obtain the fused feature data;

[0030] Input the fused feature data into the second-layer classification network of the machine learning model with adaptive parameter adjustment for electric field response classification operation. The second-layer classification network adopts a residual network structure to obtain the classification result of the electric field response state;

[0031] According to the preset interface conductivity evaluation criteria, comprehensive scoring calculation and decision fusion are performed on the classified results of the interface electrical parameter status and the classified results of the electric field response status to obtain the comprehensive score of the double-layer classification. The interface conductivity evaluation criteria include interface humidity index, physical structure parameters, and ion distribution factor;

[0032] Threshold judgment is performed on the comprehensive score of the double-layer classification. When the comprehensive score is greater than the target value, it is determined as the high-conductivity interface state. When the comprehensive score is less than or equal to the target value, it is determined as the low-conductivity interface state, and the interface state classification result is obtained.

[0033] Optionally, in the fourth implementation manner of the first aspect of the present invention, the adjustment module further includes:

[0034] A setting unit, configured to judge the interface conduction state according to the interface state classification result, set low-intensity current parameters for the high-conductivity interface state, and set high-intensity current parameters for the low-conductivity interface state to obtain target current parameters;

[0035] A calculation unit, configured to calculate the frequency of the current control system of the intelligent electrode sheet based on the target current parameters, where the first channel is a constant current mode with an intensity of P milliamperes, and the second channel is a pulsed current mode with an intensity of Q milliamperes, to obtain current channel parameters;

[0036] A dynamic analysis unit, configured to perform dynamic analysis on the electric field distribution in the contact interface area, and calculate the material migration enhancement coefficient through the electric field enhancement effect to obtain migration enhancement parameters;

[0037] A modulation unit, configured to input the target current parameters into the current control system of the intelligent electrode sheet for current waveform modulation calculation to obtain current waveform trajectory data, and perform current output operation according to the current waveform trajectory data and the current channel parameters to obtain modulated current data;

[0038] A compensation unit, configured to perform interface compatibility analysis on the modulated current data to obtain interface reaction prediction data, and perform current waveform compensation calculation according to the migration enhancement parameters and the interface reaction prediction data to obtain transmitted current data;

[0039] A prediction unit, configured to calculate the functional substance release rate data based on the interface characteristic data, and perform functional substance release rate modulation operation on the transmitted current data to obtain predicted transmitted current.

[0040] Optionally, in the fifth implementation manner of the first aspect of the present invention, the prediction unit is specifically configured to:

[0041] Input the interface characteristic data into the functional substance release rate calculation module, and calculate according to the electric field migration characteristics of the preset functional molecule characteristics to obtain the target functional substance release rate map;

[0042] Perform release depth calibration on the target functional substance release rate map to obtain calibrated functional substance release rate data, and input the transmission current data into the substance migration model for electric field-driven calculation to obtain target functional substance distribution data;

[0043] Perform concentration gradient compensation calculation on the target functional substance distribution data according to the calibrated functional substance release rate data to obtain the compensated functional substance distribution, and input the compensated functional substance distribution into the intelligent electrode sheet for regionalized functional substance release regulation to obtain the initial transmission current;

[0044] Perform real-time feedback monitoring on the initial transmission current, calculate the current distribution uniformity and substance migration depth parameters through the interface electric field diagnosis system to obtain transmission quality monitoring data;

[0045] Perform optimization calculation on the calibrated functional substance release rate data according to the transmission quality monitoring data, and obtain optimized functional substance release rate data through iterative processing. Input the optimized functional substance release rate data into the intelligent electrode sheet for secondary current modulation to obtain the predicted transmission current.

[0046] Optionally, in the sixth implementation manner of the first aspect of the present invention, the parameter optimization module is specifically used for:

[0047] Perform target deviation analysis on the predicted transmission current, calculate the difference degree from the ideal transmission mode to obtain initial deviation data, and input the initial deviation data into the programmable current controller adaptive compensation unit for dynamic compensation calculation to obtain compensation amount parameters;

[0048] Perform iterative optimization of compensation parameters according to the compensation amount parameters, perform numerical calculation on the compensation parameters by the least square method to obtain optimized compensation parameters;

[0049] Input the optimized compensation parameters into the closed-loop control system to perform compensation operation on the predicted transmission current to obtain compensated current data;

[0050] Perform high-precision electric field analysis on the compensated current data to obtain interface electric field quality data, and input the interface electric field quality data into the interface electric field diagnosis system for electric field difference operation to obtain substance migration depth data through current density calculation;

[0051] An interface adaptability evaluation model is established based on the interface electric field quality data and the material migration depth data. When the fitness is lower than a preset adaptation threshold, it is determined that the current meets the output condition, and a current quality determination result is obtained. Based on the current quality determination result, a working current that meets the conditions is output.

[0052] In summary, the technical solution provided by the present invention can obtain the electrical parameters and electric field response data of the contact interface in real time through the acquisition module. Combining with the feature analysis module for in-depth data processing, the system can accurately identify different types of interface states. Based on the double-layer feature fusion technology, the interface characteristic data obtained by the system is more comprehensive and accurate, providing a reliable basis for parameter adjustment. The state classification module adopts an innovative double-layer classification network structure. Through the combined application of a three-layer convolutional neural network and a residual network, and cooperating with the spatial attention mechanism and the channel attention mechanism to enhance the feature expression ability, high-precision classification of the interface conductivity state is achieved. The adjustment module intelligently sets the current parameters according to the classification results. Different current intensities are used for high and low conductivity interfaces respectively, and the material migration enhancement coefficient is calculated through the electric field enhancement effect, improving the material transmission efficiency while reducing energy consumption. By adopting a dual-channel collaborative control strategy of constant current and pulsed current, the uniformity of material distribution is significantly improved, the edge effect is reduced, and the system adaptability is enhanced. The parameter optimization module constructs a complete closed-loop control system, monitors and adjusts the working current in real time, prevents inappropriate current parameters from having a negative impact on the interface, and greatly improves the long-term working stability of the system. In addition, the system accurately calculates the release rate of the functional material based on the interface characteristic data. Through the concentration gradient compensation and regionalized release control technology, the directional migration and efficient utilization of the material are realized. At the same time, with the help of the interface adaptability evaluation model and high-precision electric field analysis technology, the control accuracy of the material migration depth is significantly improved, and the precise material transmission control ability is achieved. Brief Description of the Drawings

[0053] Figure 1 It is a schematic diagram of an intelligent electrode patch transdermal physiotherapy system with skin resistance adaptive adjustment in an embodiment of the present invention.

[0054] The realization, functional characteristics and advantages of the object of the present invention will be further described in conjunction with the embodiments and with reference to the drawings. Detailed Embodiments

[0055] In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention and are not used to limit the present invention.

[0056] Refer to Figure 1 , this embodiment provides an intelligent electrode patch transdermal physiotherapy system with skin resistance adaptive adjustment, including:

[0057] The acquisition module 101 is used to perform real-time electrical parameter acquisition and electric field response measurement on the contact interface area of the intelligent electrode sheet, so as to obtain electrical parameter distribution data and electric field response distribution data;

[0058] Among them, the acquisition module performs refined electrical measurements on the interface between the intelligent electrode patch and the contact surface through microcurrent detection operations. During this process, microcurrent detection technology is used to apply a small alternating or direct current. By measuring the response voltage signal and calculating according to Ohm's law (U = IR) or impedance model, the target electrical parameter map of the interface is obtained. These parameters include characteristic indexes such as resistance, capacitance, and impedance spectrum, which can reflect information such as the conductivity, polarization characteristics, and charge transfer efficiency of the interface substances. The target electrical parameter map is digitally processed to convert the analog signal into discrete initial electrical parameter data. The digital processing process includes sampling, quantization, and encoding steps. A high-speed data acquisition card or analog-to-digital converter is used to achieve high-precision signal sampling, and appropriate filtering algorithms are used to remove noise interference to obtain a relatively pure initial data set. The initial electrical parameter data is stored in the form of a matrix or grid, and each grid cell corresponds to the electrical characteristics at a specific spatial position, forming a preliminary spatial distribution map of electrical parameters. Statistical calculations of the grid area are performed on the initial electrical parameter data. The interface area is divided into several small grids by the method of partition statistics, and the statistical quantities of electrical parameters, such as mean, variance, maximum value, and minimum value, are calculated within each grid to obtain the electrical parameter distribution data. Normalization processing and data smoothing operations are performed on the electrical parameter distribution data. During the normalization process, the value range of the electrical parameters is mapped to a unified standard range (for example, between 0 and 1) to avoid data bias caused by differences in numerical magnitudes. The data smoothing operation reduces the random noise of the data through algorithms such as moving average, weighted filtering, or Gaussian smoothing to improve the continuity and stability of the data, obtaining the interface electrical parameter distribution data. In terms of electric field response measurement, the acquisition module uses a multi-frequency electric field response sensor to sample the contact interface area in real time. The multi-frequency electric field response sensor can collect interface response data under the stimulation of electric fields at different frequencies and capture the response characteristics of materials at different electric field frequencies through frequency scanning techniques (such as impedance spectroscopy EIS, dielectric spectroscopy DES). This method is suitable for analyzing ion transport, dielectric response, and polarization phenomena in the interface. The response data collected by the sensor is subjected to signal conditioning and digital processing to obtain real-time electric field response data. These data are also organized in the form of a spatial grid, recording the electric field response conditions at different spatial positions and different frequency conditions. Interface tomography calculations are performed on the real-time electric field response data. Interface tomography techniques (such as electrical impedance tomography EIT, capacitance tomography CST) utilize the spatial distribution information of multiple sensors and calculate the internal physical field distribution of the interface through inversion algorithms to generate initial electric field response sampling data, including the distribution characteristics of the electric field in space, including electric field strength, direction, and field effect distribution. Spatial distribution statistical analysis is performed on the initial electric field response sampling data, and the data is sorted through statistical methods to obtain the electric field response statistical distribution data.Combined with methods such as spatial interpolation, cluster analysis, and statistical filtering, the discrete sampling point data is transformed into a continuous distribution map, and the reliability of the data is improved by removing outliers and filling in missing data. Coordinate transformation and numerical integration operations are performed on the statistical distribution data of the electric field response. Through coordinate transformation, the data is transformed from the sampling coordinate system to the actual space coordinate system to achieve the alignment of the data with the physical space, while the numerical integration operation is used to calculate the total response amount or response change rate of the electric field in a specific area. These processes make the data more in line with the physical meaning and actual application requirements, and the corrected data of the electric field response is obtained. The corrected data of the electric field response is input into the electric field distribution reconstruction algorithm for reconstructing the spatial electric field distribution. By solving the physical models of the electric field distribution (such as Poisson's equation, electric field strength distribution equation), the discrete corrected data is reconstructed into the spatially continuous electric field response distribution data.

[0059] The feature analysis module 102 is used to convert the electrical parameter distribution data into a first feature map, convert the electric field response distribution data into a second feature map, and perform feature analysis to obtain the interface characteristic data;

[0060] Specifically, the feature analysis module normalizes the interface electrical parameter distribution data to eliminate the imbalance problems caused by different dimensions or numerical range differences in the data. The normalization process uses methods such as linear normalization to map the electrical parameter data to a unified numerical range (such as between 0 and 1), so that the data maintains numerical consistency and comparability in subsequent statistical and feature extraction processes. The normalized electrical parameter data is divided into statistical intervals. The division method of the statistical intervals is based on equal-spacing segmentation or quantile segmentation, dividing the data space into several continuous intervals, and each interval corresponds to a specific statistical range, forming the electrical parameter distribution statistical intervals. According to the electrical parameter distribution statistical intervals, data mapping and frequency counting operations are performed on the normalized electrical parameter distribution data. In the data mapping process, each data point is mapped to the corresponding statistical interval according to its numerical size. By counting the data frequencies in each interval, the electrical parameter frequency statistical results are obtained. This result is represented in the form of a histogram, frequency distribution table or heat map, reflecting the spatial distribution characteristics and statistical features of different electrical parameters at the interface. According to the frequency statistical results, the first feature map is generated. Through specific image processing algorithms (such as gray-scale conversion, color mapping, feature enhancement), the frequency distribution information is converted into a high-dimensional data expression in the feature space, so that the system uses these maps as the input of the feature extraction algorithm. The electric field response distribution data is normalized to a unified numerical range. After the normalization process is completed, the electric field response distribution data is divided into statistical intervals to generate the electric field response distribution statistical intervals, and the same segmentation standard is used to ensure the comparability of the feature data during feature fusion and the consistency of calculations. According to the electric field response distribution statistical intervals, data mapping and frequency counting operations are performed on the normalized electric field response distribution data. The electric field response data points are mapped into their respective statistical intervals. By counting the response frequencies of each interval, the electric field response frequency statistical results are generated, and the second feature map is constructed based on these results. Principal component analysis is performed on the first feature map and the second feature map respectively. Principal component analysis is a dimensionality reduction technique. By calculating the covariance matrix of the feature map, the main feature components in the map data are extracted, so that the high-dimensional data is projected into a low-dimensional space while maximizing the variance retention degree of the data. After principal component analysis of the first feature map, the electrical parameter distribution feature vector is obtained, which contains the main information in the first feature map, such as the main change trends and distribution characteristics of the electrical parameters in different statistical intervals. Similarly, principal component analysis is performed on the second feature map to extract the electric field response distribution feature vector, reflecting the main change patterns and response characteristics of the electric field response in the interface space. The feature analysis module calculates the feature weights of the first feature vector and the second feature vector respectively, and determines the weight coefficient matrix for feature fusion through covariance matrix operations.In this process, the covariance matrix reflects the linear relationship and correlation between two eigenvectors in different dimensions. By calculating the eigenvalues and eigenvectors of the covariance matrix, the allocation strategy of feature weights is obtained. The first eigenvector and the second eigenvector are subjected to weighted fusion operation according to the weight coefficient matrix. Using the method of weighted linear combination, the two eigenvectors are linearly weighted and superimposed according to the weight coefficient matrix to obtain the fused interface characteristic data.

[0061] The state classification module 103 is configured to input the interface characteristic data into a machine learning model with adaptive parameter adjustment for interface state classification operation to obtain an interface state classification result;

[0062] It should be noted that, based on the electrical parameter characteristics and the electric field response characteristics, the input interface characteristic data is subjected to data separation processing to obtain an electrical parameter characteristic data set and an electric field response characteristic data set. Through the feature decoupling algorithm, the multi-dimensional feature data is segmented according to specific feature channels or feature types, so that the electrical parameters and the electric field response characteristics respectively form independent data sets. The electrical parameter characteristic data set is input into the first-layer classification network of the machine learning model with adaptive parameter adjustment for electrical parameter classification operations. The first-layer classification network adopts a three-layer convolutional neural network structure. This network structure realizes the layer-by-layer extraction and high-dimensional feature expression of the electrical parameter characteristic data through the combination of convolutional operations, activation functions, and pooling layers. In the first-layer convolution, the network extracts local features in the data through small-sized convolutional kernels (such as 3×3 or 5×5). Through the channel expansion and feature enhancement techniques in the feature map, the spatial distribution characteristics of the electrical parameters are expressed in a finer granularity. The second-layer convolution further aggregates features in a larger range and combines pooling operations (such as max pooling or average pooling) to reduce the dimension of the feature data, reducing the computational complexity while retaining key information. The third-layer convolution combines a fully connected layer to achieve a high-level expression of the features. Through the Softmax activation function or other classification activation functions, the classification results of the interface electrical parameter states are output, including the classification probability distribution of the electrical parameter characteristics in different interface states, revealing the possibility that the interface is in a high conductivity state, a low conductivity state, or an intermediate state. Feature enhancement processing is performed on the classification results of the interface electrical parameter states. The feature enhancement process includes introducing a spatial attention mechanism and a channel attention mechanism to achieve the focusing on key features and the suppression of redundant features. In the spatial attention mechanism, a parallel structure of max pooling and average pooling is adopted. By simultaneously performing these two pooling operations in the spatial dimension of the feature map, the significant regions in the feature map are expressed with higher weights. This two-channel pooling processing ensures the comprehensiveness of feature extraction and enhances the sensitivity to local high-response regions. At the same time, in the channel attention mechanism, a squeeze-and-excitation module is adopted. This module obtains the global feature response of each channel through global average pooling operations, then realizes the correlation modeling between channels through a fully connected layer, and generates the weight coefficient of each channel through the Sigmoid activation function to achieve the weighted enhancement of useful channels and the suppression of useless channels in the feature map. After these enhancement operations, the enhanced electrical parameter characteristics are obtained. The enhanced electrical parameter characteristics are fused with the electric field response characteristic data set to obtain fused feature data. In the feature fusion process, multi-modal fusion technology is adopted. Through methods such as feature splicing, feature weighting, or feature crossing, the electrical parameters and the electric field response characteristics are seamlessly docked in the feature space to form a higher-dimensional and higher-information feature representation. The fused feature data is input into the second-layer classification network of the machine learning model with adaptive parameter adjustment for electric field response classification operations. This classification network adopts a residual network structure.Residual networks effectively solve the common problems of vanishing gradients and exploding gradients in deep learning networks by introducing skip connections, enabling the model to maintain good training performance even as the depth increases. In the residual module of the residual network, the input features are directly added to the output features to achieve an identity mapping of the features, avoiding information loss during the transfer of features between layers. This design allows the network to improve the stability and accuracy of classification operations while maintaining the depth of the model, resulting in a classification result for the electric field response state. After completing the double-layer classification, according to the preset interface conductivity evaluation criteria, a comprehensive score calculation and decision fusion are performed on the classification results of the interface electrical parameter state and the electric field response state. This evaluation criteria includes key evaluation dimensions such as the interface humidity index, physical structure parameters, and ion distribution factor. Through a multi-dimensional weighted scoring method, the two classification results are integrated into a unified score value. During the comprehensive score calculation process, according to the weight coefficients of each evaluation index, methods such as weighted average, decision tree, or fuzzy logic are used to fuse the multi-faceted information of the interface state into a numerical conductivity evaluation score. The higher the score, the better the interface conductivity. A threshold judgment is made on the comprehensive score of the double-layer classification. When the comprehensive score is greater than the target value, the current interface state is determined to be a high conductivity state, meaning the interface is wet, the ion concentration is high, and the physical structure is conducive to current conduction. When the comprehensive score is less than or equal to the target value, it is determined to be a low conductivity interface state, indicating that the interface is dry, the ion concentration is low, or the interface structure hinders current transmission.

[0063] The adjustment module 104 is used to determine the current output parameters and the mass transfer enhancement coefficient according to the interface state classification result, and perform output power adjustment and mass release control on the intelligent electrode sheet to obtain the predicted transmission current;

[0064] Specifically, the setting unit in the adjustment module determines the conductive state of the current interface based on the interface state classification result. After the system obtains the comprehensive score of the interface electrical parameters and the electric field response through the state classification module, the setting unit compares the comprehensive score with the preset interface conductivity evaluation standard. When the comprehensive score is greater than the target value, it is judged as a high-conductivity interface state, and when the comprehensive score is less than or equal to the target value, it is judged as a low-conductivity interface state. Based on this judgment result, the setting unit adopts different current parameter setting strategies. For the high-conductivity interface state, it selects to set low-intensity current parameters to avoid excessive stimulation of the interface or unnecessary energy consumption caused by too high current; for the low-conductivity interface state, it sets high-intensity current parameters to improve the interface mass transfer efficiency by increasing the current intensity and obtains the target current parameters. The calculation unit calculates the frequency of the current control system of the intelligent electrode sheet based on the target current parameters, and realizes the refined control of the current output through the mode selection and frequency adjustment of the current channel. The current control system of the system is designed in a dual-channel mode. The first channel operates in a constant current mode with an output intensity of P milliamperes. This mode is suitable for maintaining a stable current flow and is suitable for use when the contact state between the electrode sheet and the interface is good and the conductivity is high to maintain a stable substance release rate. The second channel adopts a pulsed current mode with an output intensity of Q milliamperes. This mode generates a dynamic change in the electric field intensity by periodically applying pulsed current, thereby enhancing the electric field response effect of the interface. Especially in the low-conductivity interface state, through the adjustment of the pulse frequency, pulse width, and duty cycle, it realizes the effective stimulation of ion migration and mass transfer at the interface. The calculation unit dynamically adjusts the current output parameters of the two channels by analyzing the target current parameters to obtain the current channel parameters. The dynamic analysis unit conducts a dynamic analysis of the electric field distribution in the contact interface area. Through the electric field sensor installed on the intelligent electrode sheet, it real-time collects the electric field distribution data in the interface area and calculates the mass transfer enhancement coefficient using the electric field enhancement effect. The electric field enhancement effect is based on the local enhancement phenomenon of the electric field in the non-uniform area of the interface. For example, at the interface microstructures such as capillaries, micropores, or sharp edges, the electric field intensity will increase significantly, thereby accelerating the directional migration of ions and the release of functional substances. The dynamic analysis unit calculates the mass transfer enhancement coefficient under different electric field distribution conditions through numerical simulation and finite element analysis methods to obtain the migration enhancement parameters. The target current parameters are input into the current control system of the intelligent electrode sheet for current waveform modulation calculation. The modulation unit generates specific current waveform trajectory data through current control algorithms (such as pulse width modulation PWM, frequency modulation FM, or amplitude modulation AM). These data include the amplitude change of the current and also include the phase, frequency, and period characteristics of the waveform.After generating the current waveform trajectory data, the modulation unit performs current output operations according to the calculated current channel parameters (P and Q values), maintains the basic electric field environment through the constant current of the first channel, combines the pulsed current of the second channel to generate a dynamic electric field effect, and obtains the modulated current data. The compensation unit conducts an interface compatibility analysis on the modulated current data. By analyzing the electrical response characteristics and chemical reaction characteristics of the interface, it predicts the interface reaction behavior and obtains the interface reaction prediction data. During the interface compatibility analysis process, the electrochemical characteristics of the interface material, the electrode polarization effect, and the ion concentration change trend are comprehensively considered. Through electrochemical impedance spectroscopy (EIS) or surface potential scanning technology, the reaction process occurring at the interface under a specific current waveform is predicted, including phenomena such as ion exchange, material degradation, or electrochemical deposition. After obtaining the interface reaction prediction data, the compensation unit combines the migration enhancement parameters and finely adjusts the current waveform through a waveform compensation algorithm to calculate the transmitted current data. This data not only ensures the stability of the current output at the physical level but also optimizes the mass transfer efficiency by dynamically adjusting the waveform characteristics (such as fine-tuning the pulse frequency and duty cycle). The prediction unit calculates the release rate of the functional substance based on the interface characteristic data. Through a comprehensive analysis of the electric field distribution, electrical parameters, and interface state, a prediction model for the release rate of the functional substance is established. This model is implemented through a machine learning regression model or a mass transfer equation based on a physical model (such as the Nernst-Planck equation). After obtaining the functional substance release rate data, the prediction unit performs a functional substance release rate modulation operation on the transmitted current data. By calculating the non-linear relationship between the current and the substance release rate (such as the ion concentration gradient and the change in the diffusion coefficient under the electric field drive), the current output parameters are adjusted, and finally, the predicted transmitted current is obtained.

[0065] The prediction unit inputs the interface characteristic data into the functional substance release rate calculation module. The interface characteristic data includes electrical parameters, spatial electric field distribution, and interface state classification results. The functional substance release rate calculation module calculates based on the electric field migration characteristics of the preset functional molecule characteristics. Through a model-based calculation method (such as the Nernst-Planck equation, diffusion model driven by an electric field), it simulates the migration behavior of functional molecules under the action of an electric field, and combines the electric field strength, charge density, and electrode interface characteristics to obtain the target functional substance release rate map, which shows the release rate distribution of the functional substance at different spatial positions and reflects the dynamic change trend of substance release under specific electric field conditions. The release depth calibration is performed on the target functional substance release rate map. By adjusting the spatial depth of the data in the map, the calculated release rate is made to better conform to the depth distribution characteristics in the actual substance migration process. The calibration method uses multi-level spatial transformation techniques, including spatial reconstruction based on depth sensors, depth correction based on the electric field strength gradient, and dynamic calibration based on the molecular diffusion model. This process adjusts the spatial distribution parameters of the release rate by comparing the actually measured substance distribution data with the release rate map in the calculation model, so that the calibrated functional substance release rate data truly reflects the actual migration and release state of functional molecules under the drive of an electric field. At the same time, the prediction unit inputs the transmitted current data into the substance migration model for electric field-driven calculation. The substance migration model uses methods such as the electric field-ion exchange coupling model and the electric field-driven hydrodynamics model to convert the current data into the specific distribution form of the substance in space, obtaining the target functional substance distribution data, which represents the concentration and migration direction of functional molecules at different positions in three-dimensional space coordinates. In the concentration gradient compensation calculation, the prediction unit adjusts the target functional substance distribution data according to the calibrated functional substance release rate data. The concentration gradient compensation mainly balances the areas with too high or too low local concentration in the distribution data by calculating the change in the concentration gradient of functional molecules in space. A gradient compensation algorithm (such as the compensation model based on Fick's diffusion law, the ion balance model guided by an electric field) is used to achieve uniform processing of the substance distribution by dynamically adjusting the electric field strength and substance release rate, obtaining the compensated functional substance distribution. The compensated functional substance distribution data is input into the intelligent electrode sheet. The intelligent electrode sheet realizes regionalized functional substance release regulation through current modulation and field strength control according to the substance demand in different regions. This regulation method uses multi-channel current control technology on the electrode sheet to apply electric fields with different intensities or frequencies in different spatial regions, thereby precisely controlling the release concentration and migration direction of functional molecules in the target region to obtain the initial transmitted current.The prediction unit conducts real-time feedback monitoring on the initial transmission current. The feedback monitoring is achieved through the interface electric field diagnostic system. This system collects current distribution data and electric field response data in real time through a sensor array, and obtains transmission quality monitoring data by calculating the current distribution uniformity and the material migration depth parameter. In the analysis of current distribution uniformity, by calculating the current density distribution of the current in different regions, it is judged whether there is excessive current concentration or current blank area, so as to evaluate the stability of current output. The material migration depth parameter is calculated through the coupling relationship between the electric field distribution and the change of material concentration, reflecting the actual penetration depth of functional molecules under the guidance of the electric field. These monitoring data provide real-time feedback information for the system, enabling the system to quickly identify problems occurring in the material transmission process, such as insufficient migration depth, too fast or too slow release rate, etc. According to the transmission quality monitoring data, the calibrated functional substance release rate data is optimized. The optimization process is achieved through an iterative processing method, using gradient descent algorithm, genetic algorithm or dynamic optimization algorithm based on reinforcement learning to adjust the functional substance release rate in multiple rounds. In each iteration, the error between the predicted substance release rate and the actual transmission effect is calculated, and the parameters in the release rate calculation model are adjusted through the error backpropagation mechanism to obtain the optimized functional substance release rate data. The optimized functional substance release rate data is input into the intelligent electrode sheet for secondary current modulation. By readjusting the current waveform, frequency and intensity, the refined control of the predicted transmission current is realized, so that the finally output predicted transmission current better matches the actual interface requirements, while ensuring the material release efficiency and effectively controlling the safety and stability of current output.

[0066] The parameter optimization module 105 is used to calculate the deviation of the predicted transmission current and optimize the compensation parameters, and output the working current that meets the conditions.

[0067] Among them, the parameter optimization module conducts target deviation analysis on the predicted transmission current, evaluates the difference degree between the predicted current and the ideal transmission mode. The difference degree is not only a simple deviation in current intensity, but also includes inconsistencies in current waveform, frequency response, phase difference, and spatial current distribution. The system compares the predicted transmission current with the preset ideal current model, uses the difference calculation method to obtain the initial deviation data, quantifies the deviation degree between the current output and the target mode, and reveals the systematic errors existing in the current control process (such as current fluctuations caused by changes in interface impedance). The initial deviation data is input into the adaptive compensation unit of the programmable current controller. The compensation unit calculates the real-time required compensation parameter through dynamic compensation calculation methods (such as adaptive filters, PID control algorithms, or dynamic error compensation models). After obtaining the compensation parameter, the parameter optimization module continues the iterative optimization of the compensation parameter. The iterative optimization process numerically calculates the compensation parameter by the least squares method. The least squares method is an optimization method based on the principle of error minimization. By constructing an objective function, it minimizes the sum of the squares of the errors between the actual output and the expected output to obtain the optimal compensation parameter. The system recalculates the error function in each iteration, adjusts the value of the compensation parameter, so that the error gradually converges to the minimum value. This iterative calculation method can not only improve the accuracy of the compensation parameter, but also further accelerate the convergence process through adaptive algorithms (such as gradient descent method, Newton iteration method) to obtain the final optimized compensation parameter. The optimized compensation parameter is input into the closed-loop control system to perform compensation operation on the predicted transmission current. The core of the closed-loop control system lies in real-time feedback and dynamic adjustment. By real-time monitoring the current output data, it compares the current state with the ideal current model, and adjusts the control signal according to the feedback data to achieve refined compensation of the current. The compensation operation process is based on classical control theories (such as proportional-integral-derivative PID control, model predictive control MPC). The system adjusts the output of the control signal according to the optimized compensation parameter to achieve synchronous compensation operation of the current amplitude, frequency, and waveform, and obtains the compensated current data. High-precision electric field analysis is performed on the compensated current data to evaluate the electric field quality generated by the compensated current in the interface region. The electric field distribution data in the interface region is collected by a high-precision electric field sensor, and the interface electric field quality data is calculated using numerical analysis methods (such as finite element analysis FEM, electric field distribution simulation technology). The interface electric field quality data is input into the interface electric field diagnosis system for electric field difference operation. By comparing the current electric field distribution with the ideal electric field model, the electric field difference degree is calculated, so as to evaluate the influence of the current output on the interface mass transfer effect. At the same time, through the current density calculation model, the mass transfer depth data is deduced using the current distribution data. These data calculate the actual penetration depth of functional molecules under the action of the electric field by combining the electric field strength and the mass concentration distribution characteristics.The parameter optimization module establishes an interface adaptability evaluation model based on the interface electric field quality data and the material migration depth data. This model is based on multi-dimensional adaptability evaluation criteria, including the uniformity of electric field distribution, the stability of current density, and the rationality of material migration depth. Through multi-factor analysis methods (such as fuzzy logic analysis and weighted scoring method), the interface fitness value is calculated. When the fitness is lower than the preset adaptation threshold, it is determined that the current output meets the actual requirements of material transmission, and the current quality determination result is output. The parameter optimization module outputs the working current that meets the conditions based on the current quality determination result.

[0068] Optionally, the acquisition module 101 is specifically configured to:

[0069] Perform a micro-current detection operation on the interface between the intelligent electrode sheet and the contact surface to obtain a target electrical parameter map, and perform digital processing on the target electrical parameter map to obtain initial electrical parameter data;

[0070] Perform grid area statistical calculation on the initial electrical parameter data to obtain electrical parameter distribution data, and perform normalization processing and data smoothing operation on the electrical parameter distribution data to obtain interface electrical parameter distribution data;

[0071] Perform real-time sampling on the contact interface area with a multi-frequency electric field response sensor to obtain real-time electric field response data, and perform interface tomography calculation on the real-time electric field response data to obtain initial electric field response sampling data;

[0072] Perform spatial distribution statistical analysis on the initial electric field response sampling data to obtain electric field response statistical distribution data;

[0073] Perform coordinate transformation and numerical integration operation on the electric field response statistical distribution data to obtain electric field response correction data, and input the electric field response correction data into the electric field distribution reconstruction algorithm for spatial electric field distribution reconstruction to obtain electric field response distribution data.

[0074] In this embodiment, a weak alternating or direct current is applied between the intelligent electrode sheet and the contact interface through a high-precision micro-current detector. Suppose the applied micro-current is , and the micro-current detector measures the response voltage , and uses Ohm's law to calculate the resistance at the interface. At the same time, through the alternating current impedance spectroscopy technology, the impedance of the interface under different frequency currents is measured, where is the frequency, and the calculation formula is:

[0075] ; where, and are respectively at the frequency The voltage and current values below. After analyzing these measurement data in the time domain and frequency domain, a target electrical parameter map is constructed. This map uses spatial coordinates and frequency as dimensions to show the changes in the electrical properties of the interface region at different positions and different frequencies. Digital processing is performed on the target electrical parameter map. The continuous electrical signal is discretized into a digital signal through an analog-to-digital converter. Assuming the sampling frequency is , the discretized data is represented as , where i, j represent spatial coordinates, k = 1, 2,..., N is the serial number of the sampling points, and N represents the total number of samplings. These discrete initial electrical parameter data are constructed into a multi-dimensional array, representing the electrical response of the interface region at different spatial positions and different frequencies. Statistical calculations are performed on the initial electrical parameter data for the grid region. The entire interface region is divided into grids of equal size. Assuming the size of the grid is , by calculating statistics such as the average value, variance, and extreme value of the electrical parameters within each grid, electrical parameter distribution data is obtained. The statistical calculation process is implemented through the following formula:

[0076] ;

[0077] where, represents the average value of the electrical parameters in the th grid, is the number of data points within the grid, represents the value of the th data point in the grid. These distribution data are normalized to map the parameter values to a standard range (such as [0, 1]). The normalization formula is:

[0078] ;

[0079] where, and are the minimum and maximum values of the electrical parameters in all grids respectively. To eliminate noise and spikes in the data, data smoothing operations are performed on the electrical parameter distribution data. For example, the moving average smoothing formula is:

[0080]

[0081] where, is the size of the smoothing window. By performing weighted averaging on the data in adjacent grids, the data is smoothed to obtain the interface electrical parameter distribution data. At the same time, real-time sampling is performed on the multi-frequency electric field response sensor for the contact interface region. The electric field response measured by the sensor at different frequencies is , obtain real-time electric field response data. Input these data into the interface tomography calculation module. Through electrical impedance tomography or capacitance tomography technology, based on the boundary electric field data obtained by multi-channel sensors, use the inversion algorithm to reconstruct the electric field distribution inside the interface and obtain the initial electric field response sampling data. Based on the initial electric field response sampling data, perform spatial distribution statistical analysis to calculate the statistical characteristics of the electric field response in different regions, including mean value, variance, frequency response characteristics, etc. The spatial distribution statistical analysis uses clustering analysis or spatial interpolation methods. For example, calculate the average value of the electric field intensity in a certain region:

[0082]

[0083] Among them, represents the average electric field intensity in the region , is the area of the region, is the intensity of the electric field at the position . These statistical analysis results are constructed into the electric field response statistical distribution data. Perform coordinate transformation and numerical integration operations on the electric field response statistical distribution data. Through coordinate transformation, transform the data from the sampling coordinate system to the actual space coordinate system. For example, through polar coordinate transformation , , to achieve spatial remapping of the data. In the numerical integration operation, calculate the total response or field effect of the electric field in a specific region. For example, calculate the electric field flux through surface integral:

[0084] ;

[0085] Among them, represents the electric field flux, is the integration region, is the infinitesimal area element. Through these calculations, obtain the electric field response correction data, and input these data into the electric field distribution reconstruction algorithm. Use the electric field distribution reconstruction algorithm (such as the finite element method FEM, boundary element method BEM) to reconstruct the spatial electric field distribution of the electric field response correction data. The reconstruction algorithm is based on Poisson's equation:

[0086] ;

[0087] Among them, represents the conductivity distribution of the interface region, is the electric potential distribution, is the charge density. By solving this equation, obtain the spatial electric field response distribution data of the entire interface region, and present these data in the form of a three-dimensional space atlas, showing the intensity and direction changes of the electric field at different positions.

[0088] Optionally, the feature analysis module 102 is specifically used for:

[0089] Normalize the data of the interface electrical parameter distribution and divide the statistical intervals to obtain the statistical intervals of the electrical parameter distribution;

[0090] According to the statistical intervals of the electrical parameter distribution, perform data mapping and frequency counting on the data of the interface electrical parameter distribution to obtain the statistical results of the electrical parameter frequencies, and generate the first characteristic spectrum based on the statistical results of the electrical parameter frequencies;

[0091] Normalize the data of the electric field response distribution and divide the statistical intervals to obtain the statistical intervals of the electric field response distribution;

[0092] According to the statistical intervals of the electric field response distribution, perform data mapping and frequency counting on the data of the electric field response distribution to obtain the statistical results of the electric field response frequencies, and generate the second characteristic spectrum based on the statistical results of the electric field response frequencies;

[0093] Perform principal component analysis on the first characteristic spectrum, extract the feature vectors of the electrical parameter distribution to obtain the first feature vector, and perform principal component analysis on the second characteristic spectrum, extract the feature vectors of the electric field response distribution to obtain the second feature vector;

[0094] Calculate the feature weights of the first feature vector and the second feature vector respectively, and determine the weight coefficient matrix for feature fusion through covariance matrix operations;

[0095] Perform weighted fusion operations on the first feature vector and the second feature vector according to the weight coefficient matrix to obtain the interface characteristic data.

[0096] In this embodiment, the data of the interface electrical parameter distribution is normalized to eliminate the differences caused by different physical dimensions and numerical ranges of the data. Assume that the data of the interface electrical parameter distribution is , where represents the position coordinates of the data in the two-dimensional space. Through the normalization formula:

[0097]

[0098] where, is the normalized electrical parameter data, and are the minimum and maximum values in the electrical parameter data respectively. The normalization process compresses the original data into the standard range of [0, 1] to ensure the numerical stability of subsequent statistical analysis. To analyze the distribution characteristics of the data, the normalized data is divided into statistical intervals. Assume that the data is divided into statistical intervals, and the width of each interval is:

[0099]

[0100] where, is the step size of the statistical interval. By mapping each data point to the corresponding statistical interval, the frequency of data occurrences in each interval is counted to obtain the frequency statistics result of the electrical parameter. The formula for the frequency statistics result is:

[0101]

[0102] where represents the frequency of the th statistical interval, is the indicator function. When falls into the th interval, , otherwise it is 0. This frequency counting process presents the distribution characteristics of the electrical parameter data in the form of a histogram. By visualizing the frequency statistics result, the first characteristic spectrum is generated. A similar processing method is applied to the electric field response distribution data , and normalization is performed. The normalization formula is:

[0103]

[0104] where and are the minimum and maximum values in the electric field response data respectively, is the normalized electric field response data. After normalization, the data is also divided into statistical intervals to obtain the electric field response distribution statistical intervals, and the electric field response frequency statistics result is calculated by the frequency counting method. The calculation method of the electric field response frequency statistics result is similar to that of the electrical parameter frequency statistics result. By counting the frequency of data occurrences in different intervals, the distribution characteristics of the electric field response data are transformed into a discrete frequency distribution form, and the second characteristic spectrum is generated accordingly. The main eigenvectors in the first characteristic spectrum and the second characteristic spectrum are extracted by the principal component analysis method. The principal component analysis extracts the main change direction of the data through the characteristic covariance matrix . Assume that the data matrix of the first characteristic spectrum is , with a size of , representing samples and features. Then the covariance matrix is calculated as:

[0105]

[0106] where is the transpose of the matrix. By performing eigenvalue decomposition on the covariance matrix, the eigenvalues and the corresponding eigenvectors 。The eigenvector represents the projection of the data on a specific dimension, and the eigenvalue reflects the variance of the data on that dimension. Select the eigenvectors with the largest first few eigenvalues as the first eigenvector, denoted as 。For the second eigen-spectrum, use the same method to calculate the covariance matrix and extract the eigenvector to obtain the second eigenvector 。To achieve feature fusion, calculate the feature weights for the first eigenvector and the second eigenvector respectively. The weight coefficient matrix is calculated through the covariance matrix of the eigenvectors as follows:

[0107]

[0108] Calculate the weight coefficient matrix through singular value decomposition or maximum correlation analysis , and this matrix represents the similarity and correlation degree of the two eigenvectors in each dimension. Perform weighted fusion operations on the first eigenvector and the second eigenvector according to the weight coefficient matrix. The calculation formula for weighted fusion is:

[0109]

[0110] where is the interface characteristic data after fusion, is the identity matrix. Through weighted fusion, effective integration of data in different feature dimensions is achieved.

[0111] Optionally, the status classification module 103 is specifically used for:

[0112] Separate the interface characteristic data according to the electrical parameter characteristics and the electric field response characteristics to obtain an electrical parameter characteristic data set and an electric field response characteristic data set;

[0113] Input the electrical parameter characteristic data set into the first-layer classification network of the machine learning model with adaptive parameter adjustment for electrical parameter classification operations. The first-layer classification network adopts a three-layer convolutional neural network structure to obtain the interface electrical parameter status classification result;

[0114] Perform feature enhancement processing on the interface electrical parameter status classification result, including adding a spatial attention mechanism and a channel attention mechanism. The spatial attention mechanism adopts a parallel structure of max pooling and average pooling, and the channel attention mechanism adopts a squeeze-and-excitation module to obtain the enhanced electrical parameter characteristics;

[0115] Fuse the electric field response characteristic data set and the enhanced electrical parameter characteristics to obtain fused feature data;

[0116] Input the fused feature data into the second-layer classification network of the machine learning model with adaptive parameter adjustment for electric field response classification operations. The second-layer classification network adopts a residual network structure to obtain the classification result of the electric field response state;

[0117] According to the preset interface conductivity evaluation criteria, perform comprehensive scoring calculation and decision fusion on the classification results of the interface electrical parameter state and the electric field response state to obtain the comprehensive score of the double-layer classification. The interface conductivity evaluation criteria include the interface humidity index, physical structure parameters, and ion distribution factor;

[0118] Perform threshold judgment on the comprehensive score of the double-layer classification. When the comprehensive score is greater than the target value, it is determined as the high-conductivity interface state. When the comprehensive score is less than or equal to the target value, it is determined as the low-conductivity interface state to obtain the interface state classification result.

[0119] In this embodiment, the interface characteristic data is separated according to the electrical parameter characteristics and the electric field response characteristics to obtain the electrical parameter characteristic data set and the electric field response characteristic data set. Assume that the interface characteristic data matrix is , where represents the coordinates of the data in the spatial dimension, and each element in the data matrix consists of the electrical parameter characteristics and the electric field response characteristics . Then the data separation process is expressed as:

[0120]

[0121] where and are the mask matrices for feature separation respectively. The part belonging to the electrical parameters in the data set is extracted as through matrix dot product operation, and the part belonging to the electric field response is extracted as . These separated data respectively constitute the electrical parameter characteristic data set and the electric field response characteristic data set. After completing the data separation, input the electrical parameter characteristic data set into the first-layer classification network of the machine learning model with adaptive parameter adjustment for electrical parameter classification operations. This classification network adopts a three-layer convolutional neural network structure, and through multi-level convolution and feature extraction, realizes the deep learning of electrical parameter data. In the convolutional network, the first-layer convolution operation extracts local spatial features through a small-size convolution kernel (for example ). Assume that the input feature map is , and the convolution kernel is . The convolution operation calculation formula is:

[0122]

[0123] where It represents the first-layer feature map obtained after convolution. Through the first-layer convolution, edge features and local electrical characteristics in the electrical parameter data are extracted. In the second-layer convolution, a larger receptive field (e.g., convolution kernel) is used to aggregate features in a larger range. Through pooling operations (such as max pooling and average pooling), the dimension of the feature map is further reduced, reducing the computational complexity while retaining key information. The third-layer convolution combines with the fully connected layer, and a non-linear activation function (such as the ReLU activation function) is used for feature enhancement, and the classification result of the interface electrical parameter state is output, denoted as , where is the classification result of the electrical parameter state, and are the weight and bias parameters of the convolutional layer respectively, is the activation function. After obtaining the classification result, feature enhancement processing is performed on the classification result of the interface electrical parameter state. The enhancement process includes adding a spatial attention mechanism and a channel attention mechanism. In the spatial attention mechanism, a parallel structure of max pooling and average pooling is adopted. Through the max pooling operation and the average pooling operation , the important regions in the feature map are calculated simultaneously, and the calculation formula is , where is the enhanced feature map, is the Sigmoid activation function, which is used to compress the feature values into the range of (0,1). In the channel attention mechanism, a squeeze-and-excitation module (SE module) is adopted. Through global average pooling, the global feature response of each channel is obtained, and then through the fully connected layer, the correlation modeling between channels is realized, and the enhanced electrical parameter feature is obtained. The enhanced electrical parameter feature is fused with the electric field response feature dataset . The feature fusion process adopts a weighted fusion method, and the fused feature data is calculated through the weight coefficient matrix :

[0124]

[0125] Among them, is the feature fusion weight matrix, and the fusion ratio of the two features is dynamically adjusted through an optimal weighting method (such as linear weighting or covariance-based weighting algorithm) to ensure that the fused feature data fully expresses the comprehensive characteristics of the electrical parameters and the electric field response. The fused feature data The input is subjected to an electric field response classification operation in the second-layer classification network of the machine learning model with adaptive parameter adjustment. The second-layer classification network adopts a residual network structure. By introducing skip connections, the residual network effectively solves the common problem of gradient disappearance in deep learning networks, ensuring that the model can still maintain good classification performance when the depth increases. The residual module in the residual network is calculated by the formula where, represents the output of the residual module, is the convolutional kernel weight in the residual module, is the convolution operation. By directly adding the input features to the output features, the identity mapping of features is realized. Through the stacking of multiple residual modules, the classification result of the electric field response state is obtained. To realize the evaluation of interface conductivity, according to the preset interface conductivity evaluation criteria, a comprehensive score calculation and decision fusion are performed on the classification results of electrical parameter states and the classification results of electric field response states . The interface conductivity evaluation criteria include the interface humidity index , physical structure parameters , and ion distribution factor . The comprehensive score calculation formula is:

[0126]

[0127] where, is the comprehensive score, is the weight coefficient of each evaluation index. The comprehensive score calculation is realized through the weighted average method. The system performs a threshold judgment on the comprehensive score of the double-layer classification. When the comprehensive score is greater than the target value , it is determined as the high-conductivity interface state, otherwise it is determined as the low-conductivity interface state, and the interface state classification result is obtained:

[0128]

[0129] Optionally, the adjustment module 104 further includes:

[0130] A setting unit for judging the interface conduction state according to the interface state classification result, setting low-intensity current parameters for the high-conductivity interface state and high-intensity current parameters for the low-conductivity interface state to obtain target current parameters;

[0131] A calculation unit for calculating the frequency of the current control system of the intelligent electrode sheet based on the target current parameters, where the first channel is a constant current mode with an intensity of P milliamperes, and the second channel is a pulsed current mode with an intensity of Q milliamperes, to obtain current channel parameters;

[0132] A dynamic analysis unit for dynamically analyzing the electric field distribution in the contact interface area, calculating and determining the material migration enhancement coefficient through the electric field enhancement effect, and obtaining the migration enhancement parameter;

[0133] A modulation unit for inputting the target current parameter into the current control system of the intelligent electrode sheet for current waveform modulation calculation, obtaining the current waveform trajectory data, and performing current output operation according to the current waveform trajectory data and the current channel parameter to obtain the modulated current data;

[0134] A compensation unit for performing interface compatibility analysis on the modulated current data to obtain the interface reaction prediction data, and performing current waveform compensation calculation according to the migration enhancement parameter and the interface reaction prediction data to obtain the transmission current data;

[0135] A prediction unit for calculating the functional substance release rate data based on the interface characteristic data and performing functional substance release rate modulation operation on the transmission current data to obtain the predicted transmission current.

[0136] In this embodiment, the setting unit determines the conductive state of the interface according to the interface state classification result. When the classification model of the system outputs that the interface state is highly conductive, a low-intensity current parameter is selected and set to avoid excessive interface electrolysis or ion imbalance caused by too high a current. Suppose the set low-intensity current parameter is mA; when the interface state is low conductive, the system sets a high-intensity current parameter mA to enhance the electric field intensity through a higher current output, improve the material migration efficiency and ion flux, and thus obtain the target current parameter , and the calculation method of this parameter is expressed as:

[0137]

[0138] After obtaining the target current parameter, it enters the calculation unit for specific calculation of the current control system frequency. The current control system of the intelligent electrode sheet is designed in a dual-channel mode. The first channel is a constant current mode, and the set current intensity is mA, which is mainly used to maintain the basic electric field environment to ensure a stable current path is formed between the electrode sheet and the contact interface, and is suitable for use when the interface conductivity is good to maintain the uniformity and stability of material release; the second channel is a pulsed current mode, and the set current intensity is mA. By applying high-frequency or low-frequency pulsed current, a periodically changing electric field effect is generated, and the pulsed current is used to enhance the ion migration rate and the functional substance release rate in the interface state with low ion concentration or poor conductivity. The calculation formula of the current channel parameter is:

[0139]

[0140] Among them, is the output current of the current channel, represents the timing function of the pulsed current (such as a square wave or triangular wave function). By adjusting and ratio, as well as frequency, the current output control under different modes is realized. The dynamic analysis unit conducts a dynamic analysis of the electric field distribution in the contact interface area. By collecting the electric field data of the interface area in real time, the material migration enhancement coefficient is calculated and determined using the electric field enhancement effect. This coefficient reflects the enhancement multiple of the material migration speed relative to the standard state under the current electric field distribution conditions. Assuming the electric field strength is , within a certain area , the electric field enhancement effect is expressed by the formula:

[0141]

[0142] Among them, is the migration enhancement coefficient, is the reference electric field strength, is the area of the region . By this formula, the actual influence of the electric field in different spatial regions on material migration is quantified, and the migration enhancement parameter is obtained. The target current parameters are input into the current control system of the intelligent electrode patch for current waveform modulation calculation. The modulation unit generates specific current waveform trajectory data through current control algorithms (such as pulse width modulation PWM, frequency modulation FM, or amplitude modulation AM). These data include the amplitude change of the current, and also include the phase, frequency, and period characteristics of the waveform. The calculation formula of the waveform trajectory data is expressed as:

[0143]

[0144] Among them, is the current waveform at time , is the modulation frequency, is the phase shift. In this way, complex current waveforms are generated to achieve refined control of the electric field distribution. After obtaining the current waveform trajectory data, combined with the current channel parameters, the modulated current data is obtained through current output operation to achieve dynamic current control of the intelligent electrode patch. On this basis, the compensation unit conducts an interface compatibility analysis on the modulated current data. By analyzing the electrochemical characteristics of the interface material, the electrode polarization effect, and the ion concentration change trend, the interface reaction behavior is predicted, and the interface reaction prediction data is obtained. The system combines the migration enhancement parameter and finely adjusts the current waveform through the waveform compensation algorithm to achieve dynamic compensation of the current waveform. The compensation calculation formula is:

[0145]

[0146] Among them, is the compensated current data, is the interface reaction prediction data. By comparing the predicted interface response with the current current waveform, the output current is dynamically adjusted to obtain the transmission current data , so that the current output better adapts to the actual mass transfer requirements. The prediction unit calculates the release rate of the functional substance based on the interface characteristic data. By analyzing the electric field distribution, electrical parameters and interface state, a prediction model for the release rate of the functional substance is established. Assuming the release rate of the functional substance is , the prediction model is expressed as:

[0147]

[0148] Among them, is the substance release coefficient, is the substance concentration distribution function in the interface region. The system calculates the release rate data of the functional substance through this model, and performs a functional substance release rate modulation operation on the transmission current data. By calculating the non-linear relationship between the current and the substance release rate, the current output parameters are adjusted, and finally the predicted transmission current is obtained.

[0149] Optionally, the prediction unit is specifically used for:

[0150] Input the interface characteristic data into the functional substance release rate calculation module, and calculate according to the electric field migration characteristics of the preset functional molecule characteristics to obtain the target functional substance release rate map;

[0151] Perform release depth calibration on the target functional substance release rate map to obtain the calibrated functional substance release rate data, and input the transmission current data into the substance migration model for electric field-driven calculation to obtain the target functional substance distribution data;

[0152] Perform concentration gradient compensation calculation on the target functional substance distribution data according to the calibrated functional substance release rate data to obtain the compensated functional substance distribution, and input the compensated functional substance distribution into the intelligent electrode sheet for regionalized functional substance release regulation to obtain the initial transmission current;

[0153] Perform real-time feedback monitoring on the initial transmission current, and calculate the current distribution uniformity and substance migration depth parameters through the interface electric field diagnosis system to obtain the transmission quality monitoring data;

[0154] Optimize the calibrated functional substance release rate data based on the transmission quality monitoring data, obtain the optimized functional substance release rate data through iterative processing, and input the optimized functional substance release rate data into the intelligent electrode sheet for secondary current modulation to obtain the predicted transmission current.

[0155] In this embodiment, input the interface characteristic data into the functional substance release rate calculation module. These interface characteristic data include the distribution of electrical parameters , the electric field response characteristics , and the interface state classification result . Among them represents the spatial coordinates. By calculating these data with the electric field migration characteristics of the preset functional molecule characteristics, the target functional substance release rate map is obtained. Calculate the release rate of the functional substance using the migration model of the functional molecule under the action of the electric field (such as the Nernst-Planck equation). Assume that the concentration of the functional molecule is , the electric field strength is , and the mobility is . Then the calculation formula for the target functional substance release rate is:

[0156]

[0157] Among them, represents the substance release rate at the spatial coordinates . The negative sign indicates the migration direction of the substance from the high-concentration region to the low-concentration region. Through this formula, the interface characteristic data is combined with the migration characteristics of the functional substance to generate the target functional substance release rate map, which shows the release rate changes in different regions in the form of a two-dimensional spatial distribution. Calibrate the release depth of the target functional substance release rate map to ensure that the release rate data accurately reflects the penetration effect of the functional substance at different depths. The calibration process calculates the actual penetration depth of the substance inside the interface through the electric field strength and the substance diffusion coefficient . The calibration formula is:

[0158]

[0159] Among them, is the diffusion coefficient, is the electric field strength. Through this calibration, the calibrated functional substance release rate data is obtained. At the same time, input the transmission current data The input material migration model performs electric field-driven calculations to obtain the target functional material distribution data. The material migration model is based on the theory of ion migration induced by an electric field and calculates through the relationship between current and the change in material concentration. Assuming the current density is , the charge number is , and the Faraday constant is , then the calculation formula for the target functional material distribution data is:

[0160]

[0161] Among them, is the initial concentration, ( is the electrode area). This formula directly correlates the current data with the change in material concentration, enabling the system to calculate the material distribution in different regions in real time. According to the calibrated functional material release rate data, concentration gradient compensation calculations are performed on the target functional material distribution data to eliminate the uneven transmission phenomenon caused by excessive or too low local concentration. The concentration gradient compensation performs compensation operations by calculating the concentration gradient and the diffusion coefficient . The calculation formula for the compensated functional material distribution is:

[0162]

[0163] Among them, is the spatial gradient of the material concentration, and the negative sign indicates that the concentration difference is balanced through the diffusion process, making the distribution of the material more uniform in space. The compensated functional material distribution data is input into the intelligent electrode sheet, and through regionalized functional material release regulation, fine control of the material release in different spatial regions is achieved, obtaining the initial transmission current . Real-time feedback monitoring of the initial transmission current is carried out, and the current distribution uniformity and the material migration depth parameter are calculated through the interface electric field diagnosis system. The calculation formula for the transmission quality monitoring data is:

[0164]

[0165] Among them, is the monitoring area, represents the degree of uniformity of the current distribution within the region , is the actual depth of material migration. Based on these monitoring data, the matching degree between the current output and the material transmission effect is evaluated in real time. The calibrated functional material release rate data is optimized according to the transmission quality monitoring data. By iterative processing methods (such as gradient descent method or adaptive learning algorithm), the key parameters in the release rate model are continuously adjusted, and the optimized functional material release rate data The calculation formula is:

[0166]

[0167] Wherein, and are adjustment coefficients, and are the target current uniformity and the target migration depth respectively. By the difference between the feedback data and the target value, the release rate is automatically adjusted to achieve the dynamic optimization of the material transmission process. The optimized functional material release rate data is input into the intelligent electrode sheet for secondary current modulation. By readjusting the current waveform, frequency and intensity, the predicted transmission current is precisely controlled. The calculation formula of the predicted transmission current is:

[0168]

[0169] Wherein, is the modulation coefficient. Through this calculation, the system dynamically adjusts the current output to cope with different interface state changes on the premise of maintaining the stability of the material release rate, realizes the adaptive regulation and efficient management of the material transmission process, ensures that the intelligent electrode sheet system can maintain the best working performance in various complex environments, and achieves high-precision material transmission and release effects.

[0170] Optionally, the parameter optimization module 105 is specifically used for:

[0171] Conduct target deviation analysis on the predicted transmission current, obtain the initial deviation data by calculating the difference degree from the ideal transmission mode, and input the initial deviation data into the adaptive compensation unit of the programmable current controller for dynamic compensation calculation to obtain the compensation amount parameter;

[0172] Perform iterative optimization of the compensation parameters according to the compensation amount parameter, numerically calculate the compensation parameters by the least squares method to obtain the optimized compensation parameters;

[0173] Input the optimized compensation parameters into the closed-loop control system to perform compensation operation on the predicted transmission current to obtain the compensated current data;

[0174] Perform high-precision electric field analysis on the compensated current data to obtain interface electric field quality data, and input the interface electric field quality data into the interface electric field diagnosis system for electric field difference calculation, and obtain the substance migration depth data through current density calculation;

[0175] Establish an interface adaptability evaluation model based on the interface electric field quality data and the substance migration depth data. When the fitness is lower than the preset adaptation threshold, it is determined that the current meets the output conditions, obtain the current quality determination result, and output the working current that meets the conditions based on the current quality determination result.

[0176] In this embodiment, perform target deviation analysis on the predicted transmission current to obtain the initial deviation data by calculating the degree of difference from the ideal transmission mode . By calculating the error between the predicted current and the ideal current, the method of mean square error is used for difference calculation. The specific formula is: where

[0177]

[0178] is the total number of sampling points, is the number of the th sampling time, and and are the values of the predicted current and the ideal current at this time respectively. Through this formula, the deviation degree between the current output and the target value is quantified, and the initial deviation data is input into the adaptive compensation unit of the programmable current controller for dynamic compensation calculation to obtain the compensation parameter . The adaptive compensation unit adjusts the current output in real time through a dynamic compensation algorithm (such as proportional-integral-differential PID control or adaptive filtering algorithm) based on the initial deviation data . The calculation formula of the compensation parameter is:

[0179]

[0180] where , and are the proportional, integral and differential control coefficients respectively. By adjusting these coefficients, the balance between fast response and output stability is achieved, and the fine compensation of the current deviation is realized. Iteratively optimize the compensation parameters, and use the least squares method to perform numerical calculation on the compensation parameters to obtain the optimized compensation parameter . The least squares method minimizes the sum of squared errors by constructing an error objective function to achieve the best estimate of the compensation parameters. The objective function is defined as:

[0181]

[0182] By minimizing the objective function and using the gradient descent method or Newton's method to calculate the optimal value of the compensation parameter, the optimized compensation parameter is obtained . The optimization formula is as follows:

[0183]

[0184] where is the learning rate. Through multiple iterations, the system converges to the minimum error point, achieving precise optimization of the compensation parameter. The optimized compensation parameter is input into the closed-loop control system to perform compensation operations on the predicted transmission current, obtaining the compensated current data . The closed-loop control system makes the current output continuously approach the ideal current value through real-time feedback adjustment signals. The calculation formula for the compensation operation is:

[0185]

[0186] Through this calculation, the system can dynamically correct the error in the predicted transmission current during the current output process, making the compensated current data more in line with the target requirements of current control. High-precision electric field analysis is performed on the compensated current data to obtain the interface electric field quality data . The electric field distribution data under the compensated current is collected by the electric field sensor, and the electric field strength and distribution characteristics in the interface region are reconstructed using numerical analysis methods (such as finite element analysis FEM). The calculation formula for the electric field quality data is:

[0187]

[0188] where is the conductivity distribution function in the interface region. Through this calculation, the electric field strength distribution at different spatial positions is obtained. At the same time, the interface electric field quality data is input into the interface electric field diagnosis system for electric field difference calculation. By comparing with the ideal electric field distribution , the electric field difference degree is calculated as follows:

[0189]

[0190] The uniformity of the current density distribution and the material migration depth parameter are analyzed through the electric field difference degree. The material migration depth is calculated through Faraday's law and the current density , and the calculation formula is:

[0191]

[0192] where is the current density, is the electrode surface area, is the number of charges, is the Faraday constant, is the current application time. This formula quantifies the actual migration depth of the functional material under the action of current. An interface adaptability evaluation model is established based on the electric field quality data and the material migration depth data. The model calculates the interface fitness through a multi-factor weighted scoring method , and the calculation formula is:

[0193]

[0194] Among them, , and are the adaptability evaluation weight coefficients. By adjusting these coefficients, the impacts of the electric field quality, migration depth, and electric field difference degree on the adaptability are balanced. When the fitness is lower than the preset adaptation threshold , the system determines that the current meets the output conditions and outputs the current quality determination result :

[0195]

[0196] Outputs the working current that meets the conditions based on the current quality determination result . When it is determined to meet the conditions, the system directly outputs the compensated current data as the final working current. Otherwise, the system adjusts the compensation parameters through a feedback mechanism, and performs iterative calculations and compensation operations again until the current output reaches the preset quality standard, thereby realizing the closed-loop optimization and precise control of the current output process, ensuring that the intelligent electrode sheet always maintains an efficient and stable material transmission ability under different interface states, and providing an ideal current output effect.

[0197] In the embodiments of the present application, the acquisition module 101 is used to obtain the electrical parameters and electric field response data of the contact interface in real time. Combined with the feature analysis module 102 for in-depth data processing, the system can accurately identify different types of interface states. Based on the double-layer feature fusion technology, the interface characteristic data obtained by the system is more comprehensive and accurate, providing a reliable basis for parameter adjustment. The state classification module 103 adopts an innovative double-layer classification network structure. Through the combined application of a three-layer convolutional neural network and a residual network, and cooperating with the spatial attention mechanism and the channel attention mechanism to enhance the feature expression ability, high-precision classification of the interface conductivity state is achieved. The adjustment module 104 intelligently sets the current parameters according to the classification results. Different current intensities are used for high- and low-conductivity interfaces respectively, and the material migration enhancement coefficient is calculated through the electric field enhancement effect, improving the material transmission efficiency while reducing energy consumption. By adopting the constant current and pulsed current dual-channel collaborative control strategy, the uniformity of material distribution is significantly improved, the edge effect is reduced, and the system adaptability is enhanced. The parameter optimization module 105 constructs a complete closed-loop control system, monitors and adjusts the working current in real time, prevents inappropriate current parameters from having a negative impact on the interface, and greatly improves the long-term working stability of the system. In addition, the system accurately calculates the release rate of the functional material based on the interface characteristic data. Through the concentration gradient compensation and regionalized release control technologies, the directional migration and efficient utilization of the material are realized. At the same time, with the help of the interface adaptability evaluation model and the high-precision electric field analysis technology, the control accuracy of the material migration depth is significantly improved, and the precise material transmission control ability is achieved.

[0198] It should be noted that in this article, the term "including", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, device, article or system including a series of elements not only includes those elements but also includes other elements not expressly listed, or further includes elements inherent to such process, device, article or system. Without further limitation, an element defined by the statement "including a..." does not exclude the existence of additional identical elements in the process, device, article or system including that element.

[0199] The above are only the preferred embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. Any equivalent structure or equivalent process transformation made by using the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall be equally included in the patent protection scope of the present invention.

Claims

1. An intelligent electrode transdermal physiotherapy system with self - adaptive adjustment of skin resistance, characterized in that, Including: A collection module for performing real-time electrical parameter collection and electric field response measurement on the contact interface area of the intelligent electrode sheet to obtain electrical parameter distribution data and electric field response distribution data; A feature analysis module for converting the electrical parameter distribution data into a first feature map, converting the electric field response distribution data into a second feature map, and performing feature analysis to obtain interface characteristic data; A state classification module for inputting the interface characteristic data into a machine learning model with adaptive parameter adjustment for interface state classification operation to obtain an interface state classification result; An adjustment module for determining current output parameters and a mass transfer enhancement coefficient according to the interface state classification result, and performing output power adjustment and mass release control on the intelligent electrode sheet to obtain a predicted transmission current; The adjustment module further includes: a setting unit for judging the interface conduction state according to the interface state classification result, setting low-intensity current parameters for a high-conductivity interface state, and setting high-intensity current parameters for a low-conductivity interface state to obtain target current parameters; a calculation unit for calculating the frequency of the current control system of the intelligent electrode sheet based on the target current parameters, where the first channel is in a constant current mode with an intensity of P milliamperes, and the second channel is in a pulsed current mode with an intensity of Q milliamperes to obtain current channel parameters; a dynamic analysis unit for performing dynamic analysis of the electric field distribution in the contact interface area and calculating the mass transfer enhancement coefficient through the electric field enhancement effect to obtain a migration enhancement parameter; a modulation unit for inputting the target current parameters into the current control system of the intelligent electrode sheet for current waveform modulation calculation to obtain current waveform trajectory data, and performing current output operation according to the current waveform trajectory data and the current channel parameters to obtain modulated current data; a compensation unit for performing interface compatibility analysis on the modulated current data to obtain interface reaction prediction data, and performing current waveform compensation calculation according to the migration enhancement parameter and the interface reaction prediction data to obtain transmission current data; a prediction unit for calculating functional substance release rate data based on the interface characteristic data and performing functional substance release rate modulation operation on the transmission current data to obtain a predicted transmission current; A parameter optimization module for performing deviation calculation and compensation parameter optimization on the predicted transmission current and outputting a working current that meets the conditions.

2. The intelligent electrode sheet transdermal physiotherapy system with self-adaptive adjustment of skin resistance according to claim 1, wherein The collection module is specifically used for: Performing micro-current detection operation on the interface between the intelligent electrode sheet and the contact surface to obtain a target electrical parameter map, and performing digital processing on the target electrical parameter map to obtain initial electrical parameter data; Performing grid area statistical calculation on the initial electrical parameter data to obtain electrical parameter distribution data, and performing normalization processing and data smoothing operation on the electrical parameter distribution data to obtain interface electrical parameter distribution data; Performing real-time sampling on the contact interface area by a multi-frequency electric field response sensor to obtain real-time electric field response data, and performing interface tomography calculation on the real-time electric field response data to obtain initial electric field response sampling data; Perform spatial distribution statistical analysis on the sampled data of the initial electric field response to obtain electric field response statistical distribution data; Perform coordinate transformation and numerical integration operations on the electric field response statistical distribution data to obtain electric field response correction data, and input the electric field response correction data into an electric field distribution reconstruction algorithm for spatial electric field distribution reconstruction to obtain electric field response distribution data.

3. The intelligent electrode patch transdermal physiotherapy system with self - adaptive adjustment of skin resistance according to claim 2, characterized in that, The feature analysis module is specifically used for: Perform normalization processing and statistical interval division on the interface electrical parameter distribution data to obtain electrical parameter distribution statistical intervals; According to the electrical parameter distribution statistical intervals, perform data mapping and frequency counting on the interface electrical parameter distribution data to obtain electrical parameter frequency statistical results, and generate a first feature map according to the electrical parameter frequency statistical results; Perform normalization processing and statistical interval division on the electric field response distribution data to obtain electric field response distribution statistical intervals; According to the electric field response distribution statistical intervals, perform data mapping and frequency counting on the electric field response distribution data to obtain electric field response frequency statistical results, and generate a second feature map according to the electric field response frequency statistical results; Perform principal component analysis on the first feature map, extract the electrical parameter distribution feature vector to obtain a first feature vector, and perform principal component analysis on the second feature map, extract the electric field response distribution feature vector to obtain a second feature vector; Calculate the feature weights of the first feature vector and the second feature vector respectively, and determine the weight coefficient matrix for feature fusion through covariance matrix operations; Perform weighted fusion operations on the first feature vector and the second feature vector according to the weight coefficient matrix to obtain interface characteristic data.

4. The intelligent electrode patch transdermal physiotherapy system with self - adaptive adjustment of skin resistance according to claim 1, characterized in that, The state classification module is specifically used for: Separate the interface characteristic data according to the electrical parameter characteristics and electric field response characteristics to obtain an electrical parameter characteristic data set and an electric field response characteristic data set; Input the electrical parameter characteristic data set into the first-layer classification network of a machine learning model with adaptive parameter adjustment for electrical parameter classification operations. The first-layer classification network adopts a three-layer convolutional neural network structure to obtain the interface electrical parameter state classification result; Perform feature enhancement processing on the interface electrical parameter state classification result, including adding a spatial attention mechanism and a channel attention mechanism. The spatial attention mechanism adopts a parallel structure of max pooling and average pooling, and the channel attention mechanism adopts a squeeze-and-excitation module to obtain enhanced electrical parameter features; Fuse the electric field response characteristic data set and the enhanced electrical parameter features to obtain fused feature data; Input the fused feature data into the second-layer classification network of the machine learning model with adaptive parameter adjustment for electric field response classification operations. The second-layer classification network adopts a residual network structure to obtain the electric field response state classification result; According to the preset interface conductivity evaluation criteria, comprehensive scoring calculation and decision fusion are performed on the classification results of the interface electrical parameter state and the classification results of the electric field response state to obtain the comprehensive score of the double-layer classification. The interface conductivity evaluation criteria include interface humidity index, physical structure parameters, and ion distribution factor; Threshold judgment is performed on the comprehensive score of the double-layer classification. When the comprehensive score is greater than the target value, it is determined as the high-conductivity interface state. When the comprehensive score is less than or equal to the target value, it is determined as the low-conductivity interface state to obtain the interface state classification result.

5. The intelligent electrode patch transdermal physiotherapy system with self - adaptive adjustment of skin resistance according to claim 1, characterized in that, The prediction unit is specifically used for: Input the interface characteristic data into the functional substance release rate calculation module, and calculate according to the electric field migration characteristics of the preset functional molecule characteristics to obtain the target functional substance release rate map; Perform release depth calibration on the target functional substance release rate map to obtain the calibrated functional substance release rate data, and input the transmission current data into the substance migration model for electric field-driven calculation to obtain the target functional substance distribution data; Perform concentration gradient compensation calculation on the target functional substance distribution data according to the calibrated functional substance release rate data to obtain the compensated functional substance distribution, and input the compensated functional substance distribution into the intelligent electrode sheet for regionalized functional substance release regulation to obtain the initial transmission current; Perform real-time feedback monitoring on the initial transmission current, calculate the current distribution uniformity and the substance migration depth parameter through the interface electric field diagnosis system to obtain the transmission quality monitoring data; Perform optimization calculation on the calibrated functional substance release rate data according to the transmission quality monitoring data, and obtain the optimized functional substance release rate data through iterative processing. Then input the optimized functional substance release rate data into the intelligent electrode sheet for secondary current modulation to obtain the predicted transmission current.

6. The intelligent electrode patch transdermal physiotherapy system with self-adaptive adjustment of skin resistance according to claim 1, characterized in that, The parameter optimization module is specifically used for: Perform target deviation analysis on the predicted transmission current, calculate the difference degree from the ideal transmission mode to obtain the initial deviation data, and input the initial deviation data into the programmable current controller adaptive compensation unit for dynamic compensation calculation to obtain the compensation amount parameter; Perform iterative optimization of the compensation parameter according to the compensation amount parameter, perform numerical calculation on the compensation parameter by the least square method to obtain the optimized compensation parameter; Input the optimized compensation parameter into the closed-loop control system to perform compensation operation on the predicted transmission current to obtain the compensated current data; Perform high-precision electric field analysis on the compensated current data to obtain the interface electric field quality data, and input the interface electric field quality data into the interface electric field diagnosis system for electric field difference operation, and obtain the substance migration depth data through current density calculation; Establish an interface adaptability evaluation model according to the interface electric field quality data and the substance migration depth data. When the fitness is lower than the preset adaptation threshold, it is determined that the current meets the output condition to obtain the current quality determination result, and output the working current that meets the condition based on the current quality determination result.

Images