Control method and system for stereoscopic dynamic interference electric therapeutic instrument

Abstract

The present application relates to the technical field of medical instrument electrical control, and particularly relates to a control method and system for a three-dimensional dynamic interference electric therapeutic instrument, comprising: injecting a current sequence into a target lesion area and measuring a boundary potential response matrix, reconstructing a time-varying dielectric tensor matrix representing anisotropic conduction characteristics of the tissue; defining an interference wave belly drift coordinate as a state vector, defining a compensation phase array of the three-dimensional dynamic interference electric signal and the intermediate frequency pulse treatment signal as a control vector, and constructing a nonlinear state evolution equation by using the time-varying dielectric tensor matrix; setting a target functional and introducing a co-state variable to construct a Hamilton function, and solving an optimal control vector in an allowable control set according to a maximum value principle; analytically solving the optimal control vector into a phase delay parameter to drive synchronous output of a multi-source signal, extracting an electro-physiological fluctuation residual error vector to return and drive iterative update of the time-varying dielectric tensor matrix. The present application overcomes the defects of interference wave belly wandering and scattering from the lesion target area and energy dissipation.

Description

Technical Field

[0001] This invention relates to the field of electrical control technology for medical devices, and in particular to a control method and system for a three-dimensional dynamic interference electrotherapy device. Background Technology

[0002] Three-dimensional dynamic interferential electrotherapy and intermediate-frequency pulse electrotherapy are commonly used in the clinical rehabilitation treatment of deep soft tissue lesions. Traditional interferential electrotherapy equipment adopts a static open-loop control architecture, setting a fixed output frequency and transmission phase at the beginning of treatment, and relying on the operator's experience for manual adjustment. Traditional interferential electrotherapy equipment is based on the physical assumption that human tissue is a uniform static pure resistor, without considering the cell membrane capacitance effect and the dependence of current conduction direction at different biological tissue levels.

[0003] Under long-term multi-source composite electrical stimulation therapy, with the increase of local microcirculation blood perfusion and tissue fluid exudation, the dielectric properties of the lesion area will undergo time-varying and anisotropic drift. Coupled with the physiological desensitization tolerance of local nerve motor endplates caused by long-term strong electric field stimulation, traditional interferential electrical therapy equipment lacks a dynamic sensing model and closed-loop feedback compensation mechanism for tissue physical impedance drift. As a result, the interference antinodes formed by the superposition of three-dimensional dynamic interference electrical signals and mid-frequency pulse therapy signals in the body space deviate from the preset lesion target point, resulting in spatial drift and energy scattering. Ultimately, this leads to the attenuation of the effective treatment efficacy in deep lesion target areas. The free and escaping high-voltage interference electric field forms abnormal energy accumulation in superficial healthy tissue and causes unexpected and severe stinging pain in patients. Summary of the Invention

[0004] To overcome the above shortcomings, this invention provides a control method and system for a three-dimensional dynamic interference electrotherapy device, aiming to improve the problem that most traditional interference electrotherapy devices adopt static open-loop control, which ignores the time-varying drift of human body impedance, resulting in interference wave antinodes leaving the target area and serious dissipation of treatment energy.

[0005] In a first aspect, the present invention provides the following technical solution: a control method for a three-dimensional dynamic interference electrotherapy device, comprising the following steps: S1. Inject a current sequence into the target lesion area and measure the boundary potential response matrix to reconstruct the time-varying dielectric tensor matrix characterizing the anisotropic conduction characteristics of the tissue. S2. Define the actual drift coordinates of the interference antinodes as the state vector, define the compensation phase array of the three-dimensional dynamic interference electrical signal and the intermediate frequency pulse therapy signal as the control vector, and construct a nonlinear state evolution equation using the time-varying dielectric tensor matrix. S3. Set the objective functional and introduce costate variables to construct a Hamiltonian function that combines the state vector, the control vector, the costate variables, and the nonlinear state evolution equation; S4. Set the allowable control set according to the hardware delay limit, and solve for the optimal control vector that makes the Hamiltonian function reach its maximum value within the allowable control set based on the maximum value principle. S5. The optimal control vector is parsed into a phase delay parameter, and the modulated stereo dynamic interference electrical signal and the intermediate frequency pulse therapy signal are output synchronously. The electrophysiological fluctuation residual vector is extracted and transmitted back to drive the iterative update of the time-varying dielectric tensor matrix.

[0006] By adopting the above technical solution, combining time-varying dielectric tensor and the principle of maximum value closed-loop optimization multi-source compensation phase, the precise dynamic locking of deep interference focus can be achieved. This improves the problem that most traditional interference electrotherapy devices use static open-loop control, which ignores the time-varying drift of human body impedance, resulting in interference wave antinodes leaving the target area and serious dissipation of treatment energy.

[0007] Optionally, in S1, injecting a current sequence into the target lesion region and measuring the boundary potential response matrix includes: Acquire preset scanning frequency band parameters and activate the multi-channel excitation probe array; The multi-channel excitation probe array is controlled to alternately output weak excitation currents with different frequency components to the target lesion area according to the preset scanning frequency band parameters; The voltage distribution characteristics of the surface of the target lesion area are simultaneously acquired using multiplexed electrodes in a multi-dimensional space. The voltage distribution characteristics are subjected to analog-to-digital conversion and filtering noise reduction to generate the corresponding boundary potential response matrix.

[0008] Optionally, in S1, the reconstructing of the time-varying dielectric tensor matrix characterizing the anisotropic conduction features of the tissue includes: Obtain the anatomical geometric boundary model of the target lesion region, and perform mesh discretization on the anatomical geometric boundary model to generate a three-dimensional spatial mesh model; Extract the spatial gradient parameters and temporal evolution parameters of the boundary potential response matrix on the nodes of the three-dimensional spatial mesh model; The spatial gradient parameter and the time evolution parameter are input into the Laplace potential distribution model to perform iterative conductivity calculation, thereby obtaining the anisotropic conductivity data at each grid node location. The time-varying dielectric tensor matrix is ​​generated based on the combination of the anisotropic conductivity data.

[0009] Optionally, in S2, defining the actual drift coordinates of the interference antinodes as a state vector and defining the compensated phase array of the three-dimensional dynamic interference electrical signal and the intermediate frequency pulse therapy signal as a control vector includes: Extract the three-dimensional carrier frequency corresponding to the three-dimensional dynamic interference electrical signal and the reference frequency corresponding to the intermediate frequency pulse therapy signal; The spatial interference energy distribution matrix of the three-dimensional carrier frequency and the reference frequency in the preset three-dimensional coordinate system is calculated by combining the time-varying dielectric tensor matrix. Extract the three-dimensional coordinate parameters of the highest interference energy point in the spatial interference energy distribution matrix, and designate the three-dimensional coordinate parameters as the state vector; Extract the initial phase offset set of each output channel within the reference clock cycle of the digital signal processor, and designate the initial phase offset set as the control vector.

[0010] Optionally, in S2, constructing the nonlinear state evolution equation using the time-varying dielectric tensor matrix includes: Extract the positional drift of the state vector in the continuous evolution time series, and calculate the first derivative matrix of the state vector with respect to evolution time; Extract the phase modulation weights of the control vector on the position drift amount; Combining the physical space attenuation condition defined by the time-varying dielectric tensor matrix, the first derivative matrix is ​​expressed as a mapping function containing the state vector, the control vector, and the phase modulation weight, thereby generating the nonlinear state evolution equation.

[0011] Optionally, in S3, the step of setting the objective functional and introducing costate variables to construct a Hamiltonian function that combines the state vector, the control vector, the costate variables, and the nonlinear state evolution equation includes: Obtain the coordinates of the preset ideal lesion target point, and calculate the negative square of the Euclidean distance between the actual drift coordinates of the interference antinode and the coordinates of the preset ideal lesion target point at the current evolution moment; The negative square is set as the operating benefit function, and the operating benefit function is integrated within a preset treatment time interval and the final state penalty function at the time of treatment termination is added to generate the target functional. Calculate the transpose tensor of the costate variable, and perform an inner product operation between the transpose tensor and the nonlinear state evolution equation to generate differential constraint terms; The operating benefit function is arithmetically added to the differential constraint term to generate the Hamiltonian function that combines the state vector, the control vector, the costate variables, and the nonlinear state evolution equation.

[0012] Optionally, in S4, the step of setting an allowable control set based on hardware delay limits and solving for the optimal control vector within the allowable control set that maximizes the Hamiltonian function according to the maximum principle includes: Extract the inherent upper limit of conduction delay time, lower limit of conduction delay time, and maximum phase resolution parameters of the multi-source phase servo hardware system; The permissible control set, which includes boundary constraints, is defined based on the upper limit of the conduction delay time, the lower limit of the conduction delay time, and the maximum phase resolution parameter. Perform a maximum search within the parameter boundaries defined by the permissible control set to obtain a combination of phase parameters that allows the Hamiltonian function to reach its maximum upper limit at any time of treatment evolution, and designate the combination of phase parameters as the optimal control vector; The negative values ​​of the partial derivative matrix of the Hamiltonian function with respect to the state vector are calculated to update the costate variables in real time during the process of solving the optimal control vector.

[0013] Optionally, in S5, resolving the optimal control vector into a phase delay parameter and synchronously outputting the modulated stereo dynamic interference electrical signal and the intermediate frequency pulse therapy signal includes: Extract the multi-channel phase control components from the optimal control vector and map the multi-channel phase control components to the hardware clock register to generate microsecond-level delay trigger instructions; Adjust the pulse emission start time of each output channel signal generator according to the microsecond-level delay trigger command; The multi-channel transformer-coupled output circuit is driven to synchronously output the stereoscopic dynamic interference electrical signal and the intermediate frequency pulse therapy signal, which are modulated by the microsecond-level delay trigger command, to the electrode array attached to the patient's body surface.

[0014] Optionally, in S5, the step of extracting and backpropagating the electrophysiological fluctuation residual vector and driving the iterative update of the time-varying dielectric tensor matrix includes: Extract the wave node times at which the three-dimensional dynamic interference electrical signal and the intermediate frequency pulse therapy signal interfere and cancel each other out; Within the same frequency microsecond-level time window corresponding to the wave node time, the electrophysiological acquisition circuit is activated to acquire the composite potential fluctuation signal on the surface of the patient's target area; The high-voltage stimulation artifact component in the composite potential fluctuation signal is removed by blind source separation algorithm to extract the pure induced surface electromyographic response signal. The electrophysiological fluctuation residual vector is constructed based on the induced surface electromyography response signal, and the electrophysiological fluctuation residual vector is transmitted to the input of the dielectric tensor inversion module to drive the iterative update of the time-varying dielectric tensor matrix in the next control cycle.

[0015] Secondly, the present invention provides the following technical solution for a control system of a three-dimensional dynamic interference electrotherapy device, comprising the following modules: The tissue dielectric tensor inversion module is used to inject current sequences into the target lesion region and measure the boundary potential response matrix to reconstruct the time-varying dielectric tensor matrix characterizing the anisotropic conduction characteristics of the tissue. The spatial interferometric state observation module is used to define the actual drift coordinates of the interferometric antinodes as the state vector, define the compensation phase array of the three-dimensional dynamic interference electrical signal and the intermediate frequency pulse therapy signal as the control vector, and construct a nonlinear state evolution equation using the time-varying dielectric tensor matrix. The Hamiltonian function construction module is used to set the objective functional and introduce costate variables to construct a Hamiltonian function that combines the state vector, the control vector, the costate variables, and the nonlinear state evolution equation. The maximum value optimization module is used to set an allowable control set according to the hardware delay limit, and to find the optimal control vector within the allowable control set that makes the Hamiltonian function reach a maximum value according to the maximum value principle. The multi-source phase servo compensation module is used to resolve the optimal control vector into a phase delay parameter, synchronously output the modulated stereo dynamic interference electrical signal and the intermediate frequency pulse therapy signal, extract the electrophysiological fluctuation residual vector and transmit it back, and drive the iterative update of the time-varying dielectric tensor matrix.

[0016] The present invention has the following beneficial effects: 1. In this invention, by combining time-varying dielectric tensor and the principle of maximum value closed-loop optimization multi-source compensation phase, the precise dynamic locking of deep interference focus is achieved, thereby improving the problem that most traditional interference electrotherapy devices adopt static open-loop control, which ignores the time-varying drift of human body impedance, resulting in interference wave antinodes leaving the target area and serious dissipation of treatment energy.

[0017] 2. In this invention, the boundary potential-related parameters are input into the Laplace model to solve the anisotropic conductivity, thereby accurately reconstructing the time-varying dielectric tensor matrix. This improves the problem that traditional physical field modeling mostly adopts the homogeneous assumption, which does not consider the dependence of the direction of current conduction in the organization, resulting in low accuracy of internal impedance distribution reconstruction.

[0018] 3. In this invention, by defining the allowable control set based on the hardware turn-on delay time and phase resolution, the extreme value optimization result is ensured to fall within the safe transmission boundary. This improves the problem that traditional optimization control mostly uses unconstrained theory to solve, and the signal generator output waveform is distorted because the instruction exceeds the device response limit.

[0019] 4. In this invention, by using a blind source separation algorithm at wave node time to remove high voltage artifacts and extract surface electromyographic response, a pure electrophysiological residual vector is constructed. This improves the problem that traditional electromyographic acquisition mostly uses continuous listening, and strong electrical stimulation can easily lead to circuit saturation and blockage, making it difficult to extract weak physiological feedback signals. Attached Figure Description

[0020] Figure 1 This is a flowchart of the control method for a three-dimensional dynamic interference electrotherapy device proposed in this invention; Figure 2 This is a flowchart of the closed-loop feedback and phase servo microprocessing of the control method for a three-dimensional dynamic interference electrotherapy device proposed in this invention. Figure 3 This is a diagram illustrating the architecture of the control system for the three-dimensional dynamic interference electrotherapy device proposed in this invention. Detailed Implementation

[0021] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Example 1: In a first embodiment of the present invention, the present invention provides a control method for a three-dimensional dynamic interference electrotherapy device, such as... Figures 1-2 As shown, it includes the following steps: S1. Inject a current sequence into the target lesion area and measure the boundary potential response matrix to reconstruct the time-varying dielectric tensor matrix characterizing the anisotropic conduction characteristics of the tissue. Furthermore, in S1, injecting a current sequence into the target lesion region and measuring the boundary potential response matrix includes: Acquire preset scanning frequency band parameters and activate the multi-channel excitation probe array; The multi-channel excitation probe array is controlled to alternately output weak excitation currents with different frequency components to the target lesion area according to preset scanning frequency band parameters; The voltage distribution characteristics of the surface of the target lesion area are simultaneously acquired using multiplexed electrodes in a multi-dimensional space. The voltage distribution characteristics are subjected to analog-to-digital conversion and filtering noise reduction to generate the corresponding boundary potential response matrix.

[0023] In S1, the time-varying dielectric tensor matrix characterizing the anisotropic conduction properties of the tissue is reconstructed, including: Obtain the anatomical geometric boundary model of the target lesion region, and perform mesh discretization on the anatomical geometric boundary model to generate a three-dimensional spatial mesh model; Extract the spatial gradient parameters and temporal evolution parameters of the boundary potential response matrix at the nodes of the three-dimensional spatial mesh model; The spatial gradient parameter and the time evolution parameter are input into the Laplace potential distribution model to perform iterative conductivity calculation and obtain the anisotropic conductivity data at each grid node location. The time-varying dielectric tensor matrix is ​​generated by combining anisotropic conductivity data.

[0024] Specifically, the control module first sends preset scanning frequency band parameters as the initial input data for the system. The multi-channel excitation probe array receives these parameters and outputs a weak, frequency-sweeping excitation current to the target lesion area. At this time, multiplexed electrodes synchronously acquire the analog voltage signal fed back from the body surface. After analog-to-digital conversion and filtering for noise reduction, this analog voltage signal generates a boundary potential response matrix.

[0025] During the model solution phase, the system extracts a pre-defined anatomical geometric boundary model of the target lesion region as input, and generates a three-dimensional spatial mesh model through discretization. The system extracts the spatial gradient parameters and time evolution parameters of the boundary potential response matrix at the mesh nodes as core solution data. The Laplace potential distribution model receives the above parameters and performs inverse physics field iterative solution, finally outputting the anisotropic conductivity data at each mesh node location, and combining them to output a time-varying dielectric tensor matrix.

[0026] Reducing the two-dimensional electric potential of the body surface to a three-dimensional internal impedance distribution relies primarily on the Laplace potential distribution model. The governing equation for the positive conduction of the electric field within non-uniform and anisotropic biological tissue is expressed as: ; in This represents the gradient operator in three-dimensional space. This represents the divergence operator. It represents the time-varying dielectric tensor matrix that characterizes the anisotropic conduction properties of an organism, and its internal elements are dynamically updated over time. It represents the distributed electric potential at any location in three-dimensional space.

[0027] The conductivity data is reconstructed inversely based on the measured boundary potential response matrix, which is essentially solving a severely ill-conditioned nonlinear inverse problem. The iterative solution process is based on minimizing the objective functional. ; in This represents the error functional used to reconstruct conductivity. This represents the boundary potential response matrix obtained by actual acquisition via multiplexed electrodes. This represents the predicted potential operator obtained by forward solving based on the dielectric tensor of the current iteration step. This represents the regularization penalty coefficient, used to control the stability of the solution. The operator representing the prior regularization term in the constraint solution space.

[0028] The control logic of conventional electrotherapy devices is based on the physical assumption that human tissue is a uniform, static, and purely resistive substance. However, different layers of real biological tissue exhibit extremely complex cell membrane capacitance effects and fascial variations, leading to a strong directional dependence of current conduction. Furthermore, changes in local microcirculation over time can cause dynamic shifts in impedance distribution.

[0029] By outputting a sweeping weak excitation current containing different frequency components, the system can utilize the physical differences in the penetration capabilities of high- and low-frequency currents into the intracellular and extracellular fluids and cell membranes of biological cells to obtain a boundary potential response matrix with depth detection dimensions. Mesh discretization divides the continuous anatomical physical space into discrete finite element nodes, providing a spatial basis for subsequent numerical calculations.

[0030] Spatial gradient and temporal evolution parameters are extracted and input into a Laplace potential distribution model for iterative calculation, directly mapping the externally measured two-dimensional electrical signal into a true three-dimensional anisotropic conductivity distribution within the lesion. Based on the time-varying dielectric tensor matrix generated from the anisotropic conductivity data, the error interference of the static uniform model is eliminated, providing a precise underlying environmental benchmark for the subsequent dynamic phase compensation algorithm for the interference electric field. Relying on this tensor matrix, the system can accurately predict the refraction and distortion paths of the electric field at complex tissue interfaces, fundamentally eliminating the defects of disordered wandering and energy scattering of interference antinodes deep within the lesion.

[0031] S2. Define the actual drift coordinates of the interference antinodes as the state vector, define the compensation phase array of the three-dimensional dynamic interference electrical signal and the intermediate frequency pulse therapy signal as the control vector, and construct a nonlinear state evolution equation using the time-varying dielectric tensor matrix. Furthermore, in S2, the actual drift coordinates of the interference antinodes are defined as the state vector, and the compensated phase array of the three-dimensional dynamic interference electrical signal and the intermediate frequency pulse therapy signal is defined as the control vector, including: Extract the three-dimensional carrier frequency corresponding to the three-dimensional dynamic interference electrical signal and the reference frequency corresponding to the intermediate frequency pulse therapy signal; The spatial interference energy distribution matrix of the three-dimensional carrier frequency and the reference frequency in the preset three-dimensional coordinate system is calculated by combining the time-varying dielectric tensor matrix. Extract the three-dimensional coordinate parameters of the highest interference energy point in the spatial interference energy distribution matrix, and designate the three-dimensional coordinate parameters as the state vector; Extract the initial phase offset set of each output channel within the reference clock cycle of the digital signal processor, and designate the initial phase offset set as the control vector.

[0032] In S2, the nonlinear state evolution equations constructed using the time-varying dielectric tensor matrix include: Extract the positional drift of the state vector in a continuous evolution time series, and calculate the first derivative matrix of the state vector with respect to evolution time; Extract the phase modulation weights of the control vector on the position drift; By combining the physical space attenuation condition defined by the time-varying dielectric tensor matrix, the first derivative matrix is ​​expressed as a mapping function containing the state vector, control vector and phase modulation weight, thus generating a nonlinear state evolution equation.

[0033] Specifically, the system uses externally set three-dimensional carrier frequency, reference frequency, and time-varying dielectric tensor matrix generated by the preceding process as basic input data. The arithmetic unit uses the aforementioned frequency parameters and dielectric tensor matrix to perform spatial electric field superposition operations, outputting a spatial interference energy distribution matrix. The system performs extremum retrieval on this distribution matrix, extracts the three-dimensional coordinate parameters of the highest interference energy point, and outputs them as a state vector; it also retrieves the initial phase offset set from the underlying layer of the digital signal processor and outputs it as a control vector.

[0034] During the equation construction phase, the module receives the time-series evolution data of the state vector, calculates and outputs the time-first derivative matrix of the state vector. This derivative matrix, along with the extracted phase modulation weights and the time-varying dielectric tensor matrix, is input into the core mapping algorithm layer. After dynamic mapping calculation, the nonlinear state evolution equation is finally output.

[0035] To quantify the spatial travel path of interferometric antinodes in complex media, a nonlinear state evolution equation based on state-space form is constructed, the core mathematical logic of which is defined as follows: ; in The matrix representing the first derivative of the state vector with respect to evolution time is used to characterize the instantaneous position drift and escape direction of the interference antinode in three-dimensional space. This represents the state vector, i.e., the three-dimensional coordinate parameters of the point where the actual interference energy is at its highest. This represents the control vector, which is the set of initial phase offsets for each output channel. This represents the phase modulation weight matrix applied by the control vector to the aforementioned position drift. This represents the time-varying dielectric tensor matrix, used to define the physical spatial attenuation conditions of an electric field during conduction within real biological tissues. This represents a nonlinear dynamic mapping function that integrates multivariable coupling and boundary constraints.

[0036] Traditional multi-source electrotherapy devices often assume a homogeneous human body medium, leading to a significant spatial deviation between the theoretical focal point and the actual interference focal point. This step extracts the actual interference carrier frequency and the intermediate frequency pulse reference frequency, substitutes them into the time-varying dielectric tensor matrix that reflects the anisotropic conduction characteristics of tissue, and accurately locates the true spatial position of the interference antinode in complex media. This physical coordinate is abstracted into a state vector, and the underlying phase offset is abstracted into a control vector, essentially transforming the invisible physical electric field drift phenomenon into a computable problem of modern control theory.

[0037] By extracting the continuous positional drift of the state vector and solving for the first-order derivative matrix, the drift rate of the energy focus in the target area is intuitively quantified. A mapping function is constructed by integrating the phase modulation weight with the spatial electromagnetic attenuation condition, establishing a rigorous mathematical causal relationship between "front-end output phase change" and "deep focus spatial displacement." The generated nonlinear state evolution equation directly constitutes the underlying controlled object model of the entire closed-loop system, providing absolutely clear theoretical equation support for subsequently finding the globally optimal compensation phase. This directly eliminates the underlying technical deficiency of conventional equipment, which can only blindly adjust through trial and error due to the lack of an evolution model.

[0038] S3. Define the objective functional and introduce costate variables to construct the Hamiltonian function of the joint state vector, control vector, costate variables and nonlinear state evolution equation; Furthermore, in S3, the objective functional is defined and costate variables are introduced. The Hamiltonian functions for the joint state vector, control vector, costate variables, and nonlinear state evolution equations are constructed as follows: Obtain the coordinates of the preset ideal lesion target point, and calculate the negative square of the Euclidean distance between the actual drift coordinates of the interferometric antinode and the coordinates of the preset ideal lesion target point at the current evolution moment; The negative quadratic is set as the running benefit function. The running benefit function is integrated within the preset treatment time interval and the final state penalty function at the time of treatment termination is added to generate the target functional. Calculate the transpose tensor of the costate variables, and perform an inner product operation between the transpose tensor and the nonlinear state evolution equation to generate differential constraint terms; The operating benefit function is arithmetically added to the differential constraint term to generate the joint state vector, control vector, costate variables, and Hamiltonian function of the nonlinear state evolution equation.

[0039] Specifically, the system control unit extracts the preset ideal lesion target coordinates and the actual drift coordinates of the interference antinodes output from the previous processing stage as basic input parameters. The computation module receives the above two sets of coordinate data, calculates the negative square of the Euclidean distance of their spatial difference, and outputs it as the running benefit function. Subsequently, the module introduces the preset treatment time interval and the final state penalty function, combines them with the running benefit function to perform definite integral calculation, and finally outputs the target functional.

[0040] During the core control function construction phase, the system internally generates costate variables and solves for their transpose tensors. These transpose tensors, along with the nonlinear state evolution equations generated in the previous process, are input into the inner product operation node to calculate and output differential constraint terms. The adder module receives the running reward function and the differential constraint terms, performs arithmetic addition, and outputs the complete Hamiltonian function for the next extremum solution process.

[0041] The essence of this step lies in transforming the target focusing requirements in physical space into functional extremum solution conditions under optimal control theory. The operational benefit function for evaluating the degree of spatial deviation of the interferometric antinodes is established as follows: ; in This indicates that the profit function is being run. This represents the state vector that indicates the actual drift coordinates of the antinodes of the interference wave. The control vector represents the phase array for compensating the three-dimensional dynamic interference electrical signal and the intermediate frequency pulse therapy signal. This indicates the coordinates of the preset ideal lesion target point. This indicates that the Euclidean norm is calculated on the spatial error vector.

[0042] Based on the evaluation indicators at the aforementioned single point in time, the following target functional covering the entire treatment cycle is established: ; in This represents the target functional of the system's energy targeting and focusing. It represents a continuous evolutionary time variable. and These represent the start and end times of the preset treatment time interval, respectively. This represents the definite integral operator within a preset treatment time interval. The final state penalty function represents the time when treatment ends, and is used to constrain the electric field divergence boundary at the end of treatment. Indicates the time of treatment termination The final state vector.

[0043] To solve for the functional extrema constrained by the nonlinear state evolution equation, a costate variable is introduced to construct the Pontryagin Hamiltonian function as follows: ; in Represents the Pontryagin Hamiltonian function for joint multivariates; This represents the introduced continuously differentiable costate variable; This represents the transpose tensor generated by the matrix transpose operation on the costate variable; Operators representing nonlinear state evolution equations generated by preceding processes; Represents the time-varying dielectric tensor matrix; Represents a three-dimensional spatial position vector.

[0044] Conventional electrical stimulation devices typically rely on operator observation or patient subjective feedback to manually adjust output parameters. This open-loop adjustment method lacks global quantitative evaluation standards and cannot address the phenomenon of electric field focus drift caused by changes in the conductivity of deep tissues.

[0045] Calculating the negative square of the Euclidean distance between the actual drift coordinates and the ideal target coordinates directly transforms the spatial deviation in the physical dimension into a negative benefit penalty in the mathematical dimension. The larger the distance error, the closer the generated benefit function value approaches negative infinity. By integrating and superimposing the final-state penalty function to generate the target functional, the underlying defect of single-step feedback control being prone to falling into local suboptimal traps is overcome, establishing a global evaluation benchmark that considers the energy focusing effect throughout the entire treatment cycle.

[0046] A simple objective functional cannot reflect the true physical conduction resistance of the electric field within complex biological media. By introducing costate variables and performing an inner product operation between their transpose tensor and the nonlinear state evolution equation, the time-varying physical environment barrier defined by the dielectric tensor is essentially forcibly embedded into the pure mathematical optimization objective in the form of differential constraint terms. The Hamiltonian function generated by adding the payoff function and the differential constraint terms bridges the underlying barrier between the preset target and the actual physical constraints. This function provides a direct integrand basis for subsequently finding the optimal compensation phase for each channel, ensuring that the system's output control strategy can be executed accurately and without error in a real heterogeneous tissue environment.

[0047] S4. Set the allowable control set according to the hardware delay limit, and solve for the optimal control vector that makes the Hamiltonian function reach its maximum value within the allowable control set based on the maximum value principle. Furthermore, in S4, an allowable control set is set according to the hardware delay limit. Based on the maximum principle, the optimal control vector that maximizes the Hamiltonian function within the allowable control set is found, including: Extract the inherent upper limit of conduction delay time, lower limit of conduction delay time, and maximum phase resolution parameters of the multi-source phase servo hardware system; The allowable control set, including boundary constraints, is defined based on the upper limit of the conduction delay time, the lower limit of the conduction delay time, and the maximum phase resolution parameter. Perform a maximum search within the parameter boundaries defined by the allowable control set to obtain the phase parameter combination that makes the Hamiltonian function reach the upper limit of the maximum value at any time of treatment evolution, and designate the phase parameter combination as the optimal control vector; Calculate the negative value of the partial derivative matrix of the Hamiltonian function with respect to the state vector to update the co-state variables in real time during the process of solving the optimal control vector.

[0048] Specifically, the system microprocessor extracts pre-set multi-source phase servo hardware system parameters from the underlying memory as initial input data for physical constraints. The extracted data includes the upper and lower limits of the conduction delay time, as well as the maximum phase resolution parameter. Based on these hardware characteristic parameters, the microprocessor generates an admissible control set containing boundary constraints and loads it into the memory register. The system retrieves the Hamiltonian function equation, along with the associated state vector and costate variables, output from the previous processing stage. Within the parameter boundaries defined by the admissible control set, it initiates a numerical optimization module to perform extreme value retrieval. After computation, the numerical optimization module outputs the phase parameter combination that maximizes the Hamiltonian function, defining it as the optimal control vector for subsequent discharge modulation. The computation module calculates and outputs the negative value of the partial derivative matrix of the Hamiltonian function with respect to the state vector, and sends this negative value to the costate variable update register to complete the numerical overwrite.

[0049] To ensure that pure mathematical solutions can be adapted to the underlying hardware driver capabilities, the following allowable control set is established: ; in This represents the permissible control set that includes entity boundary constraints. Indicates evolution time Any control vector under [the specified conditions]. This represents the lower limit of the conduction delay time determined by the underlying multi-source phase servo hardware system. This represents the upper limit of the conduction delay time determined by the underlying multi-source phase servo hardware system. An arbitrary control vector is given here by default. All have satisfied the discretization step constraint of the maximum phase resolution parameter.

[0050] Within the set permissible control set, the core solution conditions for solving the objective parameters based on the maximum principle are constructed as follows: ; in This represents the Pontryagin Hamiltonian function generated by the preceding processing steps. Indicates the current evolutionary time The state vector that maintains the optimal evolutionary trajectory. This represents the optimal control vector that maximizes the Hamiltonian function, and is the core objective of this process. This represents a costate variable that strictly matches the optimal state trajectory. Indicates that in the allowable control set A mathematical operator that performs maximum search within the bounded range of a given condition. Indicates that in the allowable control set Combination of test control parameters for inner traversal testing.

[0051] The continuous evolution of the system requires the synchronous updating of costate variables, and its conjugate evolution partial differential equation is: ; in Representing costate variables The first derivative matrix with respect to continuous evolution time. Representing the Hamiltonian function For the state vector Solve for the partial derivative matrix. The negative sign indicates the inverse conjugate characteristic of the gradient update of the costate variable.

[0052] Solving optimal control algorithms in the theoretical dimension is prone to generating extreme parameters that physical devices cannot execute. For example, calculating tiny phase deflections below the hardware's nanosecond response limit, or extremely large phase shift commands exceeding the transformer core saturation boundary. By extracting upper and lower limits for conduction delay time and maximum phase resolution parameters to define the allowable control set, a non-overreach barrier is directly established between the algorithm solution layer and the underlying transmission hardware layer. This constraint eliminates invalid theoretical solutions that could lead to hardware shutdown or output waveform distortion, ensuring that the calculated control commands fall entirely within the safe transmission envelope of the signal generator.

[0053] Performing a maximum optimization search within strictly defined hardware parameter boundaries essentially involves squeezing out the actual physical control sequence that maximizes the alignment of the interference energy focus with the target area coordinates within the limits of available hardware tuning capabilities. The optimized combination of phase parameters is designated as the optimal control vector, bridging the fundamental gap between the macroscopic therapeutic target and the microscopic electrical commands.

[0054] The negative values ​​of the partial derivative matrix of the Hamiltonian function with respect to the state vector are calculated to update the costate variables, establishing a closed-loop self-verification mechanism within the control system. In optimal control theory, costate variables characterize the marginal sensitivity of the objective functional to state vector drift. With the passage of treatment time and the actual movement of the interference antinode position, fixed costate variables can cause the Hamiltonian function to detach from the actual physical environment. By using the negative values ​​of the partial derivatives to update the costate variables in real time, the mathematical form of the Hamiltonian function dynamically evolves along with the impedance drift of the physical lesion, preventing the optimization algorithm from diverging under long-term continuous treatment conditions.

[0055] S5. The optimal control vector is analyzed into a phase delay parameter, and the modulated stereo dynamic interference electrical signal and the intermediate frequency pulse therapy signal are output synchronously. The electrophysiological fluctuation residual vector is extracted and transmitted back to drive the iterative update of the time-varying dielectric tensor matrix. Furthermore, in S5, the optimal control vector is analyzed into a phase delay parameter, and the modulated stereo dynamic interference electrical signal and the intermediate frequency pulse therapy signal are output synchronously, including: Extract the multi-channel phase control components from the optimal control vector and map the multi-channel phase control components to the hardware clock register to generate microsecond-level delay trigger instructions; Adjust the pulse emission start time of each output channel signal generator according to the microsecond-level delay trigger command; The multi-channel transformer-coupled output circuit drives the electrode array attached to the patient's body surface to synchronously output a three-dimensional dynamic interference electrical signal and a mid-frequency pulse therapy signal modulated by a microsecond-level delay trigger command.

[0056] In S5, the extraction and backpropagation of the electrophysiological fluctuation residual vector to drive the iterative update of the time-varying dielectric tensor matrix includes: Extract the wave node moments at which the three-dimensional dynamic interference electrical signal and the intermediate frequency pulse therapy signal interfere and cancel each other out; Within the same frequency microsecond-level time window corresponding to the wave node time, the electrophysiological acquisition circuit is activated to acquire the composite potential fluctuation signal on the surface of the patient's target area; The high-voltage stimulation artifact component in the composite potential fluctuation signal is removed by blind source separation algorithm, and the pure induced surface electromyographic response signal is extracted. An electrophysiological fluctuation residual vector is constructed based on the induced surface electromyography response signal, and then transmitted to the input of the dielectric tensor inversion module to drive the iterative update of the time-varying dielectric tensor matrix in the next control cycle.

[0057] Specifically, the control module extracts the optimal control vector from the previous solution process as the initial input parameter. The analysis unit extracts the multi-channel phase control components, loads them into the underlying hardware clock register, and outputs a microsecond-level delay trigger command. The signal generators of each output channel reset the pulse transmission start time according to the received delay command, and output physically modulated three-dimensional dynamic interference electrical signals and intermediate frequency pulse therapy signals to the body surface electrode array via a multi-channel transformer coupling circuit.

[0058] At the feedback loop, the system clock extracts the wave node time when the two types of treatment signals interfere and cancel each other out, and sends a trigger command to the electrophysiological acquisition loop. The acquisition loop reads the surface composite potential fluctuation signal within the corresponding microsecond-level time window and sends it to the blind source separation algorithm node. After the algorithm node processes and removes the high-voltage stimulation artifact component, it outputs the induced surface electromyographic response signal. The system encapsulates this response signal into an electrophysiological fluctuation residual vector and transmits it directionally back to the data receiving end of the dielectric tensor inversion module as the environmental input parameter driving the next round of loop calculations.

[0059] To accurately translate the optimal control vector at the algorithm level into a discharge action at the hardware execution level, a time-domain delay conversion model is established as follows: ; in Indicates the first The hardware clock register of each output channel receives the specific time value of the microsecond-level delay trigger instruction. Represents the corresponding first element extracted from the optimal control vector. Multi-channel phase control components for each output channel. Indicates the first The reference carrier frequency transmitted by each output channel. This represents the constant value of pi.

[0060] Processing mixed electrophysiological signals acquired from the body surface relies on blind source separation and demixing operations, the core mathematical expression of which is established as follows: ; in This represents the pure induced surface electromyographic response signal matrix obtained after separation and extraction by the algorithm. This represents the unmixed weight matrix generated by the blind source separation algorithm through iterative solution of independent components maximization. This represents the matrix multiplication operator. This represents the original composite potential fluctuation signal matrix acquired through the acquisition loop at the moment of interference phase elimination.

[0061] The mathematical relationship for constructing the feedback driving vector based on pure electromyographic signals is as follows: ; in This represents the electrophysiological fluctuation residual vector that has been constructed and transmitted to the input of the inversion module. This represents the baseline electromyographic response signal matrix of the ideal tissue in a non-desensitized state, as preset by the system.

[0062] The optimal control vector is parsed into multi-channel phase control components and mapped to the hardware clock register, bridging the execution gap between the control algorithm and the underlying hardware driver layer. A microsecond-level delay trigger instruction resets the pulse emission start time of the signal generator, precisely translating the abstract theoretical compensation phase into the physical discharge timing of the physical circuit, ensuring that the various therapeutic electric fields strictly interfere and superimpose according to the target coordinates at deep lesions in the human body.

[0063] Acquiring weak bioelectrodes under strong electric field therapy conditions can easily cause saturation and blockage of the acquisition amplifier. The system specifically extracts the wave node moment of the destructive interference of electric field energy, using this extremely short time window as the hardware trigger condition to activate the electrophysiological acquisition circuit. Utilizing the instantaneous zero-potential characteristic at the wave node, a natural electromagnetic shielding gap is constructed in situ. Combined with a blind source separation algorithm to remove residual artifacts, this eliminates the measurement obstacle that makes it difficult to extract weak physiological responses under strong stimulation interference.

[0064] Example 2: In long-duration, multi-source composite electrical stimulation therapy, the time-varying and anisotropic dielectric properties of deep human biological tissues cause the antinodes of the interference waves formed by the superposition of the three-dimensional dynamic interference electrical signal and the mid-frequency pulse therapy signal in the body to detach from the preset lesion target point, resulting in uncontrollable spatial drift and energy scattering. Simultaneously, the physiological desensitization tolerance of the local motor endplates of the lesion leads to a precipitous decrease in the effective therapeutic efficacy of the deep lesion target area. Furthermore, the free-floating high-voltage interference electric field easily forms abnormal energy accumulation in superficial healthy tissue or non-target nerve trunks, triggering unexpected severe stinging pain and local muscle rigidity spasms in patients. To solve these problems, the control system for the three-dimensional dynamic interference electrical therapy device provided by this invention is adopted, the structure of which is as follows: Figure 3 As shown. The specific implementation process of this system is as follows: The tissue dielectric tensor inversion module reconstructs the time-varying dielectric tensor matrix using the measured boundary potential matrix, abandoning the erroneous assumption of homogenization of human tissue in traditional devices and providing accurate time-varying physical and electrical boundaries. The spatial interferometric state observation module uses this tensor matrix to construct a nonlinear state evolution equation, establishing the intrinsic dynamic causal relationship between the front-end output phase change and the spatial walk of the deep interference antinodes. The Hamiltonian function construction module introduces costate variables and the target functional to generate a Hamiltonian function, establishing a global extremum solution index considering the entire treatment cycle. The maximum optimization module defines a safe transmission envelope based on the hardware delay limits of the underlying devices, solving for the optimal output control vector within the allowable control set to prevent output distortion caused by invalid theoretical solutions. The multi-source phase servo compensation module is responsible for parsing instructions and driving the signal generator to execute precise synchronous transmission of multiple treatment signals, extracting the surface electromyographic response induced in the target area to construct the electrophysiological fluctuation residual vector and transmitting it back to the headend. The underlying dielectric tensor matrix is ​​iteratively updated based on the residual. The system relies entirely on the real neurological tolerance physiological feedback of the lesion muscle layer to continuously correct the electric field compensation phase, eliminating the underlying defects of interference wave antinode wandering scattering and deep energy cliff drop under long-term treatment conditions.

[0065] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A control method for a three-dimensional dynamic interference electrotherapy device, characterized in that, Includes the following steps: S1. Inject a current sequence into the target lesion area and measure the boundary potential response matrix to reconstruct the time-varying dielectric tensor matrix characterizing the anisotropic conduction characteristics of the tissue. S2. Define the actual drift coordinates of the interference antinodes as the state vector, define the compensation phase array of the three-dimensional dynamic interference electrical signal and the intermediate frequency pulse therapy signal as the control vector, and construct a nonlinear state evolution equation using the time-varying dielectric tensor matrix. S3. Set the objective functional and introduce costate variables to construct a Hamiltonian function that combines the state vector, the control vector, the costate variables, and the nonlinear state evolution equation; S4. Set the allowable control set according to the hardware delay limit, and solve for the optimal control vector that makes the Hamiltonian function reach its maximum value within the allowable control set based on the maximum value principle. S5. The optimal control vector is parsed into a phase delay parameter, and the modulated stereo dynamic interference electrical signal and the intermediate frequency pulse therapy signal are output synchronously. The electrophysiological fluctuation residual vector is extracted and transmitted back to drive the iterative update of the time-varying dielectric tensor matrix.

2. The control method for a three-dimensional dynamic interference electrotherapy device according to claim 1, characterized in that, In S1, injecting a current sequence into the target lesion region and measuring the boundary potential response matrix includes: Acquire preset scanning frequency band parameters and activate the multi-channel excitation probe array; The multi-channel excitation probe array is controlled to alternately output weak excitation currents with different frequency components to the target lesion area according to the preset scanning frequency band parameters; The voltage distribution characteristics of the surface of the target lesion area are simultaneously acquired using multiplexed electrodes in a multi-dimensional space. The voltage distribution characteristics are subjected to analog-to-digital conversion and filtering noise reduction to generate the corresponding boundary potential response matrix.

3. The control method for a three-dimensional dynamic interference electrotherapy device according to claim 1, characterized in that, In S1, the reconstruction of the time-varying dielectric tensor matrix characterizing the anisotropic conduction properties of the tissue includes: Obtain the anatomical geometric boundary model of the target lesion region, and perform mesh discretization on the anatomical geometric boundary model to generate a three-dimensional spatial mesh model; Extract the spatial gradient parameters and temporal evolution parameters of the boundary potential response matrix on the nodes of the three-dimensional spatial grid model; The spatial gradient parameter and the time evolution parameter are input into the Laplace potential distribution model to perform iterative conductivity calculation, thereby obtaining the anisotropic conductivity data at each grid node location. The time-varying dielectric tensor matrix is ​​generated based on the combination of the anisotropic conductivity data.

4. The control method for a stereoscopic dynamic interference electrotherapy device according to claim 1, characterized in that, In S2, defining the actual drift coordinates of the interference antinodes as a state vector and defining the compensated phase array of the three-dimensional dynamic interference electrical signal and the intermediate frequency pulse therapy signal as a control vector includes: Extract the three-dimensional carrier frequency corresponding to the three-dimensional dynamic interference electrical signal and the reference frequency corresponding to the intermediate frequency pulse therapy signal; The spatial interference energy distribution matrix of the three-dimensional carrier frequency and the reference frequency in the preset three-dimensional coordinate system is calculated by combining the time-varying dielectric tensor matrix. Extract the three-dimensional coordinate parameters of the highest interference energy point in the spatial interference energy distribution matrix, and designate the three-dimensional coordinate parameters as the state vector; Extract the initial phase offset set of each output channel within the reference clock cycle of the digital signal processor, and designate the initial phase offset set as the control vector.

5. The control method for a stereoscopic dynamic interference electrotherapy device according to claim 1, characterized in that, In S2, constructing the nonlinear state evolution equation using the time-varying dielectric tensor matrix includes: Extract the positional drift of the state vector in the continuous evolution time series, and calculate the first derivative matrix of the state vector with respect to evolution time; Extract the phase modulation weights of the control vector on the position drift amount; Combining the physical space attenuation condition defined by the time-varying dielectric tensor matrix, the first derivative matrix is ​​expressed as a mapping function containing the state vector, the control vector, and the phase modulation weight, thereby generating the nonlinear state evolution equation.

6. The control method for a three-dimensional dynamic interference electrotherapy device according to claim 1, characterized in that, In S3, the step of setting the objective functional and introducing costate variables to construct a Hamiltonian function that combines the state vector, the control vector, the costate variables, and the nonlinear state evolution equation includes: Obtain the coordinates of the preset ideal lesion target point, and calculate the negative square of the Euclidean distance between the actual drift coordinates of the interference antinode and the coordinates of the preset ideal lesion target point at the current evolution moment; The negative square is set as the operating benefit function, and the operating benefit function is integrated within a preset treatment time interval and the final state penalty function at the time of treatment termination is added to generate the target functional. Calculate the transpose tensor of the costate variable, and perform an inner product operation between the transpose tensor and the nonlinear state evolution equation to generate differential constraint terms; The operating benefit function is arithmetically added to the differential constraint term to generate the Hamiltonian function that combines the state vector, the control vector, the costate variables, and the nonlinear state evolution equation.

7. The control method for a stereoscopic dynamic interference electrotherapy device according to claim 1, characterized in that, In S4, the step of setting an allowable control set based on hardware delay limits and finding the optimal control vector within the allowable control set that maximizes the Hamiltonian function according to the maximum principle includes: Extract the inherent upper limit of conduction delay time, lower limit of conduction delay time, and maximum phase resolution parameters of the multi-source phase servo hardware system; The permissible control set, which includes boundary constraints, is defined based on the upper limit of the conduction delay time, the lower limit of the conduction delay time, and the maximum phase resolution parameter. Perform a maximum search within the parameter boundaries defined by the permissible control set to obtain a combination of phase parameters that allows the Hamiltonian function to reach its maximum upper limit at any time of treatment evolution, and designate the combination of phase parameters as the optimal control vector; The negative values ​​of the partial derivative matrix of the Hamiltonian function with respect to the state vector are calculated to update the costate variables in real time during the process of solving the optimal control vector.

8. The control method for a three-dimensional dynamic interference electrotherapy device according to claim 1, characterized in that, In S5, the optimal control vector is parsed into a phase delay parameter, and the modulated stereo dynamic interference electrical signal and the intermediate frequency pulse therapy signal are synchronously output, including: Extract the multi-channel phase control components from the optimal control vector and map the multi-channel phase control components to the hardware clock register to generate microsecond-level delay trigger instructions; Adjust the pulse emission start time of each output channel signal generator according to the microsecond-level delay trigger command; The multi-channel transformer-coupled output circuit is driven to synchronously output the stereoscopic dynamic interference electrical signal and the intermediate frequency pulse therapy signal, which are modulated by the microsecond-level delay trigger command, to the electrode array attached to the patient's body surface.

9. The control method for a three-dimensional dynamic interference electrotherapy device according to claim 1, characterized in that, In S5, the step of extracting and backpropagating the electrophysiological fluctuation residual vector and driving the iterative update of the time-varying dielectric tensor matrix includes: Extract the wave node times at which the three-dimensional dynamic interference electrical signal and the intermediate frequency pulse therapy signal interfere and cancel each other out; Within the same frequency microsecond-level time window corresponding to the wave node time, the electrophysiological acquisition circuit is activated to acquire the composite potential fluctuation signal on the surface of the patient's target area; The high-voltage stimulation artifact component in the composite potential fluctuation signal is removed by blind source separation algorithm to extract the pure induced surface electromyographic response signal. The electrophysiological fluctuation residual vector is constructed based on the induced surface electromyography response signal, and the electrophysiological fluctuation residual vector is transmitted to the input of the dielectric tensor inversion module to drive the iterative update of the time-varying dielectric tensor matrix in the next control cycle.

10. A control system for a three-dimensional dynamic interference electrotherapy device, characterized in that, The control method for a stereoscopic dynamic interference electrotherapy device according to any one of claims 1-9 comprises the following modules: The tissue dielectric tensor inversion module is used to inject current sequences into the target lesion region and measure the boundary potential response matrix to reconstruct the time-varying dielectric tensor matrix characterizing the anisotropic conduction characteristics of the tissue. The spatial interferometric state observation module is used to define the actual drift coordinates of the interferometric antinodes as a state vector, define the compensation phase array of the three-dimensional dynamic interference electrical signal and the intermediate frequency pulse therapy signal as a control vector, and construct a nonlinear state evolution equation using the time-varying dielectric tensor matrix. The Hamiltonian function construction module is used to set the objective functional and introduce costate variables to construct a Hamiltonian function that combines the state vector, the control vector, the costate variables, and the nonlinear state evolution equation. The maximum value optimization module is used to set an allowable control set according to the hardware delay limit, and to find the optimal control vector within the allowable control set that makes the Hamiltonian function reach a maximum value according to the maximum value principle. The multi-source phase servo compensation module is used to resolve the optimal control vector into a phase delay parameter, synchronously output the modulated stereo dynamic interference electrical signal and the intermediate frequency pulse therapy signal, extract the electrophysiological fluctuation residual vector and transmit it back, and drive the iterative update of the time-varying dielectric tensor matrix.

Images