Physiotherapy instrument control system and method based on artificial intelligence

Abstract

The invention discloses a physiotherapy instrument control system and method based on artificial intelligence, and relates to the technical field of intelligent medical treatment, and the method comprises the steps: collecting an electromyographic signal in real time, carrying out band-pass filtering and TK energy operator processing, and outputting a muscle activation feature vector; based on the muscle activation feature vectors, spatial-temporal features are extracted through a convolutional neural network, and a physiotherapy parameter combination is constructed; a PID controller is adopted to adjust the physical therapy parameter combination in real time, and optimized physical therapy parameters are output; performing electromagnetic thermal coupling simulation on the optimized physiotherapy parameters through finite element analysis to obtain a physiotherapy instrument control instruction; and updating the weight of the convolutional neural network by adopting a stochastic gradient descent algorithm according to the physiotherapy efficiency evaluation report, and outputting an optimization execution instruction. PID parameters are optimized through a quantum annealing algorithm, stimulation parameters are dynamically adjusted in combination with myoelectricity beta wave band phase locking, and treatment parameters and physiological response are synchronized.

Description

Technical Field

[0001] The present invention relates to the field of intelligent medical technology, and particularly to a physiotherapy instrument control system and method based on artificial intelligence. Background Art

[0002] The control method of a physiotherapy instrument driven by electromyogram signals is an important research direction in the cross-field of rehabilitation medicine and biomedical engineering, and has high application value in the postoperative rehabilitation of anorectal diseases such as hemorrhoids, anal fissure and anal fistula surgery. Most of the current methods are to respond to the change of electromyogram signal amplitude through a preset stimulation mode. Among them, anorectal special equipment usually uses surface electrodes of the anal sphincter to collect signals, and uses low-frequency electrical stimulation to relieve pain or medium-frequency electrical stimulation to improve muscle function. The common control methods of physiotherapy instruments driven by electromyogram signals usually use linear time-invariant filters to extract signal features, combine look-up table methods and empirical formulas to generate treatment parameters, and then use an analog PID controller to achieve closed-loop regulation. At present, the existing technology has realized the basic electromyogram feedback treatment function and has been widely used in the fields of muscle rehabilitation, pain management, anorectal postoperative physiotherapy, etc.

[0003] However, the traditional method has insufficient adaptability in spatio-temporal feature extraction. The static convolution kernel is difficult to capture the non-stationary characteristics of electromyogram signals. For example, it cannot effectively identify the unique circular contraction mode of the anal sphincter, which limits the matching accuracy between treatment parameters and dynamic muscle activation states. In addition, there are also deficiencies in the coordination of electromagnetic-thermal coupling simulation and real-time control. The finite element analysis currently used is mostly used as an offline verification tool and fails to form a closed-loop optimization with online parameter regulation, affecting the dynamic guarantee ability of the treatment safety margin. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a physiotherapy instrument control method based on artificial intelligence to solve the problem of insufficient coordination between dynamic feature extraction of electromyogram signals and real-time optimization of physiotherapy parameters.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: In the first aspect, the present invention provides a physiotherapy instrument control method based on artificial intelligence, which includes: collecting electromyogram signals in real time, performing band-pass filtering and TK energy operator processing, and outputting a muscle activation feature vector; Based on the muscle activation feature vector, extracting spatio-temporal features through a convolutional neural network, and constructing a combination of physiotherapy parameters; Using a PID controller to perform real-time adjustment on the combination of physiotherapy parameters, and outputting optimized physiotherapy parameters; Performing electromagnetic-thermal coupling simulation on the optimized physiotherapy parameters through finite element analysis to obtain a physiotherapy instrument control instruction; Combined with a constant current source circuit and pulse width modulation temperature control, execute the control instructions of the physiotherapy instrument, and monitor the physiotherapy effect in real time to generate a physiotherapy efficacy evaluation report; According to the physiotherapy efficacy evaluation report, use the stochastic gradient descent algorithm to update the weights of the convolutional neural network and output optimized physiotherapy instrument control instructions.

[0007] As a preferred solution of the physiotherapy instrument control method based on artificial intelligence according to the present invention, wherein: the myoelectric signals are collected in real time, and band-pass filtering and TK energy operator processing are performed to output muscle activation feature vectors. The specific steps are as follows: Collect myoelectric signals in real time, combine a bioelectric amplifier and a Butterworth filter to remove noise and motion artifacts, and output pure myoelectric signals; Combine the Hilbert transform and the TK energy operator, extract the signal envelope to obtain time-varying energy features, and use the statistical features of a sliding window to extract and output muscle energy features; Adopt the dynamic window statistical method to obtain the energy integral value, use a digital filter bank to construct a spatial feature vector and combine feature splicing to output muscle activation feature vectors.

[0008] As a preferred solution of the physiotherapy instrument control method based on artificial intelligence according to the present invention, wherein: based on the muscle activation feature vectors, spatio-temporal features are extracted through a convolutional neural network to construct a physiotherapy parameter combination. The specific steps are as follows: Adopt a sliding window to divide the window, and perform a fast Fourier transform on the window signal to output time series feature data; Use dilated causal convolution to construct a spatio-temporal feature map and output a fused spatio-temporal feature vector; Use a dual-branch predictor to synchronously optimize physiotherapy parameters and control safety risks to construct a physiotherapy parameter combination.

[0009] As a preferred solution of the physiotherapy instrument control method based on artificial intelligence according to the present invention, wherein: the PID controller is used to adjust the physiotherapy parameter combination in real time to output optimized physiotherapy parameters. The specific steps are as follows: Use the quantum annealing algorithm to optimize the initial PID parameters under the Lyapunov stability constraint and output preliminary optimized parameters; Combine the preliminary optimized parameters with the real-time pure myoelectric signals, dynamically adjust the stimulation parameters according to the phase of the myoelectric β band, and output the dynamically adjusted preliminary optimized parameters; Use incremental trial-and-error optimization and parameter fallback for feature comparison to output optimized physiotherapy parameters.

[0010] As a preferred solution of the physiotherapy device control method based on artificial intelligence according to the present invention, wherein: electromagnetic-thermal coupling simulation is performed on the optimized physiotherapy parameters through finite element analysis to obtain physiotherapy device control instructions. The specific steps are as follows: Convert the optimized physiotherapy parameters into standard units, and use middleware to convert the input format to output a standardized physiotherapy parameter set; Combine the standardized physiotherapy parameter set with the real-time pure myoelectric signal, and use Maxwell's equations to obtain the spatial electromagnetic field distribution, and output the instantaneous electromagnetic field distribution; Based on the instantaneous electromagnetic field distribution, discretize the tissue region using the finite difference method, and correct the real-time temperature value through weighted fusion to output a corrected temperature field; Based on the corrected temperature field, use threshold control to monitor and dynamically adjust the physiotherapy parameters in real time, and output a safety control instruction; Execute safety constraints on the safety control instruction and generate an executable instruction, and output the physiotherapy device control instruction As a preferred solution of the physiotherapy device control method based on artificial intelligence according to the present invention, wherein: combining a constant current source circuit and pulse width modulation temperature control, execute the physiotherapy device control instruction, and monitor the physiotherapy effect in real time to generate a physiotherapy efficacy evaluation report. The specific steps are as follows: Combine the constant current source circuit with the microcontroller to receive and issue the physiotherapy device control instruction, and output an electrical stimulation signal; Monitor the temperature of the physiotherapy area in real time, and use a fuzzy PID controller to dynamically adjust the physiotherapy power to output a safe physiotherapy temperature; Collect the patient's subjective feedback and physiological index data, and integrate the physiotherapy parameters and the safe physiotherapy temperature to output a therapeutic effect monitoring data set; Use a weighted statistical algorithm to analyze the therapeutic effect monitoring data set to generate a physiotherapy efficacy evaluation report.

[0011] Updating the weights of the convolutional neural network according to the physiotherapy efficacy evaluation report using the stochastic gradient descent algorithm to output an optimized physiotherapy device control instruction. The specific steps are as follows: Use sliding window statistics and standardization processing to extract physiotherapy parameters and physiological index data from the physiotherapy efficacy evaluation report, and output a standardized feature vector and an optimized target parameter; Apply the gradient descent algorithm to obtain the update direction, and perform parameter adjustment within the safety range to output an optimized combination of physiotherapy parameters; Use structured data encapsulation and convert the optimized combination of physiotherapy parameters into a device execution instruction to output an optimized physiotherapy device control instruction.

[0012] Second aspect, the present invention provides a physiotherapy instrument control system based on artificial intelligence, including a signal processing module, a feature extraction module, a parameter regulation module, a simulation analysis module, an execution feedback module, and an optimization iteration module; The signal processing module is used to collect electromyogram signals in real time, perform band-pass filtering and TK energy operator processing, and output muscle activation feature vectors; The feature extraction module is used to extract spatio-temporal features based on the muscle activation feature vectors through a convolutional neural network and construct a combination of physiotherapy parameters; The parameter regulation module is used to use a PID controller to adjust the combination of physiotherapy parameters in real time and output optimized physiotherapy parameters; The simulation analysis module is used to perform electromagnetic-thermal coupling simulation on the optimized physiotherapy parameters through finite element analysis to obtain physiotherapy instrument control instructions; The execution feedback module is used to combine a constant current source circuit and pulse width modulation temperature control, execute the physiotherapy instrument control instructions, and monitor the physiotherapy effect in real time to generate a physiotherapy efficacy evaluation report; The optimization iteration module is used to update the weights of the convolutional neural network by using the stochastic gradient descent algorithm according to the physiotherapy efficacy evaluation report and output optimized physiotherapy instrument control instructions.

[0013] Third aspect, the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and: when the computer program is executed by the processor, any step of the physiotherapy instrument control method based on artificial intelligence as described in the first aspect of the present invention is implemented.

[0014] Fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and: when the computer program is executed by the processor, any step of the physiotherapy instrument control method based on artificial intelligence as described in the first aspect of the present invention is implemented.

[0015] The beneficial effects of the present invention are as follows: The PID parameters are optimized through the quantum annealing algorithm, and the stimulation parameters are dynamically adjusted in combination with the phase of the electromyogram β band, so that the treatment parameters are synchronized with the physiological response; The time series features of the electromyogram signals are extracted by using sliding window segmentation and fast Fourier transform, and the spatio-temporal correlation is enhanced in combination with dilated causal convolution, so that the feature extraction is more suitable for the dynamic muscle activation state. Description of the Drawings

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0017] Figure 1 It is a flowchart of a control method for a physiotherapy instrument based on artificial intelligence.

[0018] Figure 2 It is a schematic diagram of a control system for a physiotherapy instrument based on artificial intelligence.

[0019] Figure 3 It is a flowchart for electromyogram signal processing.

[0020] Figure 4 It is a flowchart for optimizing physiotherapy parameters. Specific implementation manners

[0021] To make the above objects, features, and advantages of the present invention more obvious and understandable, the specific implementation manners of the present invention will be described in detail below with reference to the accompanying drawings of the specification.

[0022] In the following description, many specific details are set forth in order to fully understand the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art can make similar generalizations without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0023] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation manner of the present invention. The appearances of "in one embodiment" in different places in this specification do not all refer to the same embodiment, nor are they separate or alternative embodiments that exclude each other with other embodiments.

[0024] Referring to Figures 1 to 4 , it is an embodiment of the present invention. This embodiment provides a control method for a physiotherapy instrument based on artificial intelligence, including the following steps: S1. Real-time collect electromyogram signals, perform band-pass filtering and TK energy operator processing, and output muscle activation feature vectors.

[0025] Furthermore, real-time collect electromyogram signals, combine a bioelectric amplifier and a Butterworth filter to remove noise and motion artifacts, and output pure electromyogram signals.

[0026] Specifically, a high-precision bioelectric amplifier and a surface electrode array are used for real-time EMG signal acquisition. The original EMG signal is amplified 1000 times by the high-precision bioelectric amplifier, and common-mode interference is suppressed. The amplified differential signal is obtained and a 16-bit analog-to-digital converter is used to digitize the EMG signal at a sampling rate of 2kHz. The Nyquist criterion is used to meet the sampling requirements of the highest frequency of the EMG signal, and the original multi-channel digital EMG signal is output; a 4th-order Butterworth bandpass filter is used for low-frequency noise filtering and high-frequency noise suppression. The low-frequency noise filtering is eliminated by a 20Hz high-pass filter to eliminate baseline drift (such as low-frequency noise caused by breathing or changes in skin electrode contact). Frequency fluctuation), high-frequency noise suppression uses a 450Hz low-pass filter to remove electromagnetic noise outside the range of the electromyographic signal (such as radio frequency interference or equipment switching noise), uses bidirectional filtering for phase compensation to eliminate the phase delay caused by the filter and align the electromyographic signal time domain, and outputs the electromyographic signal after bandpass filtering; the limb movement signal is synchronously collected through the accelerometer as the reference input of the motion artifact, and the filter is dynamically adjusted using the least mean square algorithm to subtract the motion-related components from the electromyographic signal. The sym4 wavelet basis is used for the residual high-frequency noise of the filtered signal, and the wavelet soft threshold denoising is performed with a decomposition level of 5 to obtain a smooth electromyographic signal and output a pure electromyographic signal.

[0027] The Hilbert transform and TK energy operator are combined to extract the signal envelope and obtain the time-varying energy characteristics. Meanwhile, the statistical characteristics of the sliding window are used to extract the muscle energy characteristics.

[0028] Specifically, a Hilbert transform is performed based on the pure EMG signal to obtain the orthogonal components of the pure EMG signal and the analytical signal. The modulus of the analytical signal is the signal envelope, which reflects the trend of the amplitude of the pure EMG signal changing over time. The phase offset of the Hilbert transform is eliminated by group delay compensation, so that the signal envelope is aligned with the original EMG signal in time domain, and a smooth pure EMG signal envelope is output. The TK energy operator is combined with envelope weighting, and the TK energy operator is used for the signal envelope and nonlinearly amplified to highlight the transient characteristics of muscle activation and output the time-varying energy characteristics of the pure EMG signal. The sliding window is divided based on the time-varying energy characteristics, and the statistical features of the sliding window are extracted to obtain the window energy mean and energy variation coefficient, and the energy rising slope and energy peak ratio are obtained by linear fitting, and the characteristic values beyond the physiological range are eliminated, the correlation between the features is checked, the redundant features are removed, and the muscle energy characteristics are output.

[0029] Based on the muscle energy characteristics, the dynamic window statistics method is used to obtain the energy integral value, and the digital filter group is used to construct the spatial feature vector and perform feature splicing to output the muscle activation feature vector.

[0030] Specifically, the dynamic window statistical method is adopted. The window is divided into analysis windows with a muscle energy feature of 200 ms and a step size of 50 ms. The energy integral value is obtained within each window to reflect the total work done by muscle activation, the energy volatility, characterize the stability and rise time of the nerve drive signal, measure the time required from the baseline to 90% of the peak and the activation duration, and verify the feature reliability through motion sensor data, and output the time-domain feature vector; use a digital filter bank to decompose the time-varying energy feature of the pure EMG signal; use a digital filter to decompose the time-varying energy feature of the pure EMG signal into the γ band (30 - 60 Hz), the β band (13 - 30 Hz) and the α band (8 - 13 Hz), and for the γ band, obtain the percentage of the energy in the total signal energy as the voluntary contraction intensity index, determine the main frequency position through the peak detection algorithm and record its offset relative to the center frequency as the main frequency offset, use bispectrum analysis to obtain the phase coupling degree between the γ-β bands, and extract the amplitude of the third harmonic component through harmonic distortion analysis. For the β band, obtain the energy proportion to reflect the intensity of the maintenance activity, detect the offset of the main frequency within the range of 13 - 30 Hz, analyze its coherence coefficient with the α band as the band coupling degree, and extract the amplitude ratio of the fundamental wave to the second harmonic; for the α band, while obtaining the basic energy parameters, monitor the distribution entropy value of the power spectral density within the 8 - 13 Hz frequency band to quantify the relaxation state feature; compose the energy proportion (3 dimensions), the main frequency offset (3 dimensions), the band coupling degree (2 dimensions for γ-β and β-α in total), and the harmonic components (take the fundamental wave and the third harmonic for each band, 6 dimensions in total), a total of 14 feature parameters of the three bands of γ, β, and α into the frequency-domain feature vector, and output the frequency-domain feature vector after range normalization; adopt the spatial pattern analysis method to analyze the activation propagation speed, the synchronization index, the spatial entropy value and the dominant conduction direction, and correct the spatial error through impedance distribution, and output the spatial feature vector; use feature splicing to merge the time-domain feature vector, the frequency-domain feature vector, and the spatial feature vector for normalization processing, perform logarithmic compression on the time-domain feature vector, perform range normalization on the frequency-domain feature vector, and use Z-value standardization on the spatial feature vector, and output the muscle activation feature vector.

[0031] It should be noted that the nerve drive signal refers to the electrical activity of motor neurons inferred from the EMG signal.

[0032] Preferably, through high-precision signal acquisition and dynamic feature extraction, the accuracy and real-time performance of EMG signal processing are improved, and the robustness of noise suppression and motion artifact elimination is enhanced; through feature fusion and adaptive optimization, the comprehensiveness of muscle activation state representation is improved, and the matching of time-frequency domain features and physiological responses is enhanced; through standardization and dynamic window analysis, the compatibility of the feature vector is improved, and the reliability of parameter optimization is enhanced.

[0033] S2. Based on the muscle activation feature vector, extract spatio-temporal features through a convolutional neural network to construct a combination of physiotherapy parameters.

[0034] Furthermore, based on the muscle activation feature vector, use Z-score normalization to eliminate the dimensional difference, adopt sliding window segmentation and perform fast Fourier transform on the window signal to output time series feature data.

[0035] Specifically, use Z-score normalization to eliminate the dimensional difference between different channels, and adopt missing value imputation and adjacent window mean filling to output the standardized feature vector; use the muscle energy feature dynamic adjustment strategy, such as 100ms for the high energy segment, 300ms for the low energy segment, the default value is 200ms, and the overlap rate is 50%; for the standardized feature vector, perform sliding interception according to the window length and step size to generate n windows, obtain the mean, variance and peak-to-peak value of the features in each window, and perform fast Fourier transform on the window signal to extract the energy proportion in the 1-500Hz frequency band. The aggregated features of K windows are arranged in chronological order to form a time series feature matrix and output time series feature data.

[0036] It should be noted that the peak-to-peak value refers to the maximum fluctuation range of the time-varying energy feature within the sliding window, which reflects the instantaneous intensity fluctuation range of muscle activation.

[0037] Reorganize the time series feature data, and use dilated causal convolution to construct a spatio-temporal feature map to output the fused spatio-temporal feature vector.

[0038] Specifically, reorganize the three-dimensional data such as continuous time steps, time domain, frequency domain, inter-channel correlation and multi-electrode channels in the time series feature data, and organize them into a spatio-temporal cube, retaining the electrode topological relationship; use a convolution kernel with interval sampling ability. The initial convolution analyzes the adjacent 3 time points, and the next layer of convolution analyzes the association of the span of 5 time points to gradually expand the analysis range. Only allow the convolution kernel to access the current and historical time step data, and do not use future data to ensure real-time performance, obtain the time step feature set, and construct an enhanced spatio-temporal feature map; for each time step feature output by the convolution kernel, convert it into three groups of vectors: query vector, key vector and value vector through linear transformation and spatial attention. According to the spatial position of the electrode, automatically obtain the importance weight between channels. For example, since the patch electrodes 1 and 3 are close in anatomical position, their signal correlation obtains a higher weight, and add a learnable time series marker for each time step to output the fused spatio-temporal feature vector.

[0039] It should be noted that the spatial position of the electrode can be directly obtained through the physical position of the electrode patch.

[0040] Preprocess the fused spatio-temporal feature vector, and use a dual-branch predictor to synchronously optimize the physiotherapy parameters and control the safety risk to construct a combination of physiotherapy parameters.

[0041] Specifically, based on the fused spatio-temporal feature vector, Z-score standardization is performed to eliminate the dimension difference, and it is segmented into time segments at a fixed time interval (such as 50 ms) to adapt to real-time control, and the preprocessed spatio-temporal feature vector is output; a dual-branch predictor is used, including a physical therapy parameter prediction branch and a safety constraint evaluation branch. The physical therapy parameter prediction branch inputs the preprocessed spatio-temporal feature vector into a three-layer fully connected neural network, with 256 nodes as the first layer, 128 nodes as the second layer, and 64 nodes as the third layer. Each layer uses the ReLU activation function for processing, and the output includes three indicators: stimulation frequency, current intensity, and pulse width. The current intensity is constrained within the range of 5 to 100 milliamperes through the activation function, and the pulse width is limited between 50 and 300 microseconds, outputting a preliminary combination of physical therapy parameters; the safety constraint evaluation branch uses the same feature vector as the input, but is processed using a different network structure, including two layers of fully connected networks and a long short-term memory network, which is used to analyze the temporal risk characteristics, outputting a safety probability score between 0 and 1, and also giving a specific risk type judgment, and comparing and verifying with the preset clinical rules. The risk rules include various safety limit conditions, such as automatically limiting the maximum stimulation frequency when the energy of the electromyogram signal in a specific frequency band is detected to be too high. If any parameter violates the safety rules, it is corrected and the corresponding risk type is marked; dynamic weighting is used to mix the original parameters with the default safety parameters according to the safety probability score (such as when the safety score is 0.7, 70% uses the predicted value and 30% uses the default value), forcing the physical therapy parameters not to exceed the physiological safety boundary such as the current being less than or equal to 100 mA, and fine-tuning the parameters based on real-time physiological indicators to construct a combination of physical therapy parameters.

[0042] It should be noted that safety constraint refers to a mechanism for dynamically restricting and risk assessing physical therapy parameters through physiological safety boundaries and clinical rules, making the physical therapy process conform to the human physiological tolerance range and reducing tissue damage, pain exacerbation or other adverse reactions caused by improper parameters; physiological indicators include skin impedance, local temperature, and electromyogram signal characteristics; the preset clinical rules include physiological safety boundaries: current intensity, pulse width, and temperature control; signal feature trigger rules: abnormal electromyogram signal and impedance monitoring; patient state adaptation rules: subjective feedback (pain degree score, comfort level selection, and local sensation description) and physiological indicator blood oxygen saturation value; risk rules include electrical safety rules, physiological response rules, and patient state.

[0043] Preferably, by sliding window segmentation and fast Fourier transform, the resolution of the temporal features of the EMG signal is improved, and the adaptability to non-stationary signal processing is enhanced; by dilated causal convolution and spatio-temporal feature fusion, the ability to capture the correlation of multi-electrode channels is improved, and the representation accuracy of the dynamic muscle activation state is enhanced; by a two-branch predictor and safety constraint evaluation, the safety of the physiotherapy parameter generation is improved, and the matching between the physiotherapy parameters and the physiological needs is enhanced.

[0044] S3. Use a PID controller to adjust the combination of physiotherapy parameters in real time and output the optimized physiotherapy parameters.

[0045] Furthermore, use the quantum annealing algorithm to optimize the combination of physiotherapy parameters under the Lyapunov stability constraint and output the preliminary optimized parameters.

[0046] Specifically, in combination with parameter range setting and quantum annealing hardware initialization, by defining a reasonable search space for the PID parameters and connecting the quantum processor, output the initialized parameter space and the ready quantum computing environment; use quantum bit encoding and Hamiltonian construction to convert the continuous PID parameters into discrete quantum optimization, encode each PID parameter using an 8-bit binary encoding scheme, and define a composite cost function that includes the integral of time multiplied by the absolute error criterion and the Lyapunov function, and output the set of PID parameter candidates represented in the form of quantum bits; use Lyapunov stability analysis in combination with the set of PID parameter candidates, by calculating the quadratic Lyapunov function and its derivative in real time, eliminate the parameter combinations that do not meet the stability conditions, and output the subset of PID parameters that pass the stability test; use the parallel search of the quantum annealing algorithm to synchronously evaluate 512 groups of parameter combinations within the set annealing time window of 20 ms, break through the local optimum limit through the quantum tunneling effect, select the parameter that minimizes the composite cost function, and output the preliminary optimized parameters.

[0047] It should be noted that the formula for the composite cost function is: ; where represents the composite cost function, comprehensively measuring the control error and stability; is the integral of time-weighted error, punishing the continuously existing control error (such as current, temperature deviation), ; represents the Lyapunov function, ensuring stability and preventing out-of-control caused by parameter adjustment, , is the error weight coefficient, controlling the contribution weight of the error term to adjust the response speed, adjusted according to the efficacy evaluation report, and the default value can be set to 0.7, represents the stability weight coefficient, the contribution weight of the stability term to prevent overshoot, automatically increased when an abnormal EMG signal is detected, and the default value can be set to 0.3.

[0048] Combine the preliminary optimized parameters with the real-time pure EMG signal, dynamically adjust the stimulation parameters according to the phase of the EMG beta band, and output the preliminarily optimized parameters with dynamic adjustment.

[0049] Specifically, convert the preliminary optimized parameters into physiotherapy instructions, use digital filtering technology to extract the 13 - 30 Hz characteristic frequency band in the real-time pure EMG signal, adopt signal envelope detection to identify the nerve discharge characteristics, and output the beta band characteristic signal reflecting the muscle activation state; monitor the fluctuation period of the EMG signal in real time, use time-domain feature matching, dynamically adjust the stimulation timing according to the nerve discharge characteristics of the rising or falling edge of the waveform, and maintain the synchronization of the stimulation and nerve activity through delay compensation, and output the stimulation timing scheme synchronized with the muscle contraction rhythm; set the upper and lower safety limits of the stimulation intensity, such as 10 - 50 mA, automatically adjust the output intensity according to the change of the signal amplitude and monitor the change of the skin impedance in real time. When an abnormality occurs, start protection and output the preliminarily optimized parameters with dynamic adjustment.

[0050] It should be noted that the abnormal change of the skin impedance refers to the change of the hindering effect of the skin on the current that does not conform to the normal physiological law or the expected range, including sudden change of impedance value, imbalance of polarity response, abnormal frequency response, left-right asymmetry and breakthrough of the safety threshold. When the impedance value is lower than 300 Ω, there is an electric shock risk, and when it is higher than 5000 Ω, the electrode contact failure is abnormal.

[0051] According to the preliminarily optimized parameters with dynamic adjustment, use incremental trial-and-error optimization and parameter fallback for feature comparison, and combine skin impedance and muscle tremor analysis to correct the parameters, and output the optimized physiotherapy parameters.

[0052] Specifically, execute the biological adaptation parameters for the physiotherapy device and continuously compare the EMG signal waveform during the physiotherapy process with the effective physiotherapy waveform, obtain the characteristic differences such as signal amplitude and fluctuation period, generate parameter adjustment suggestions according to the degree of difference, output the recommended adjustment direction, use incremental trial-and-error optimization, take the current parameters as the benchmark, conduct a small-range parameter exploration of ±5%, and monitor the response change trends such as muscle contraction force and fatigue degree in real time, retain the parameter changes with improved response, revoke the invalid adjustments, and output the effective parameter fine-tuning scheme; synchronously track indicators such as skin impedance and muscle tremor. When any indicator exceeds the normal fluctuation range, automatically pause the physiotherapy, restore the safe state through the parameter fallback mechanism, compare the current physiotherapy data with the historical successful cases in terms of characteristics, and adopt the effective parameter combination in the similar cases, and output the optimized physiotherapy parameters.

[0053] It should be noted that for the definition criteria of the normal fluctuation range, skin impedance: the baseline range is 800 - 3000 Ω, the change amplitude during physical therapy is less than or equal to 25%, a sudden drop greater than 30% may be due to electrode detachment or skin damage, and a continuous rise greater than 40% may indicate poor contact or too thick cutin layer; local temperature: the skin surface temperature in the physical therapy area should be 32 - 37 °C, the temperature rise per single physical therapy session is less than or equal to 3 °C, the temperature rise per minute is less than or equal to 0.5 °C, stop the machine immediately when the absolute temperature is greater than 40 °C, and trigger an alarm when the local temperature difference is greater than 2 °C / cm²; muscle tremor: 50 - 500 μV in the resting state, the normal physical therapy response amplitude increase is less than or equal to 300%, and the abnormal high-frequency tremor with a component greater than 50 Hz accounting for more than 15%.

[0054] Preferably, through the quantum annealing algorithm and Lyapunov stability constraint, the global convergence of PID parameter optimization is improved to enhance stability; through β-band synchronization adjustment, the coordination between stimulation parameters and physiological rhythms is improved to enhance the biological adaptability of the physical therapy plan; through incremental trial-and-error optimization and parameter fallback mechanism, the accuracy of dynamic adjustment of physical therapy parameters is improved to enhance the safety and fault tolerance in abnormal states.

[0055] S4. Conduct electromagnetic-thermal coupling simulation on the optimized physical therapy parameters through finite element analysis to obtain the control instructions for the physical therapy device.

[0056] Furthermore, convert the optimized physical therapy parameters into standard units, and use middleware to convert the input format to output a standardized set of physical therapy parameters.

[0057] Specifically, use regular expression matching to identify variables from the optimized physical therapy parameters and convert them into standard units, such as converting current to amperes, frequency to hertz, and duty cycle to a decimal, and output a parameter set with a unified dimension; Z-score standardization is suitable to eliminate the dimension differences between parameters, perform linear transformation according to the normal fluctuation range, and standardize and normalize the frequency and duty cycle to output a dimensionless parameter vector; use tensor filling to construct spatio-temporal correlation data, arrange the dimensionless parameter vector into a time series matrix according to the time window, use interpolation method to supplement missing values, and construct a two-dimensional parameter matrix in a standard format; use middleware conversion to convert the two-dimensional parameter matrix in the standard format into an input format supported by finite element analysis, and embed the boundary conditions of electrode contact impedance to output a standardized set of physical therapy parameters.

[0058] Combine the standardized set of physical therapy parameters with real-time pure electromyogram signals, and use Maxwell's equations to obtain the spatial electromagnetic field distribution and output the instantaneous electromagnetic field distribution.

[0059] Specifically, by using structured data extraction and physical quantity conversion, the standardized physiotherapy parameter group and the pure EMG signal amplitude are directly parsed, the current intensity is converted into the electrode surface current density, the tissue conductivity is adjusted according to the EMG signal amplitude, the boundary conditions for electromagnetic field calculation are obtained, and the Maxwell's equations are used in combination with the finite difference method to directly solve, outputting the three-dimensional electric field distribution and magnetic field distribution. The electrode surface current density is set, and the remaining boundaries are insulation or impedance matching conditions to obtain the maximum electric field strength and maximum magnetic field strength. The local gradient method is used to identify the field strength spatial region and detect the field strength concentration region. The region with a sudden change in the rate of change is used to obtain the hotspot map of the field strength distribution, and the instantaneous electromagnetic field distribution is output.

[0060] Based on the instantaneous electromagnetic field distribution, the finite difference method is used to discretize the tissue region, and weighted fusion is combined to correct the real-time temperature value, outputting the corrected temperature field.

[0061] Specifically, based on the instantaneous electromagnetic field distribution, according to the electric field strength and tissue conductivity (dynamically adjusted by the EMG signal amplitude), the heat generation per unit volume is obtained, and the three-dimensional heat source distribution is output; combined with the three-dimensional heat source distribution and real-time temperature monitoring data (such as body surface temperature), the finite difference method is used to discretize the tissue region, the temperature change of each grid point is obtained, and the blood flow heat dissipation (the blood flow takes away 0.5 W / m³ of heat) is considered to obtain the temperature distribution, and the predicted temperature field is output; based on the predicted temperature field and real-time temperature monitoring data, direct weighted fusion is used to correct the temperature value near the monitoring point, and the corrected temperature field is output.

[0062] It should be noted that the weight coefficient can be adjusted according to the sensor accuracy and can be constructed as 30% of the predicted temperature field and 70% of the real-time temperature monitoring data; the tissue region refers to the three-dimensional biological tissue space where the physiotherapy instrument acts, including layered structures such as skin, fat, and muscle, which has conductivity and needs to be discretized into grids by the finite difference method.

[0063] Based on the corrected temperature field, threshold control is used to monitor in real time and dynamically adjust the standardized physiotherapy parameters, outputting the safety control instruction.

[0064] Specifically, by combining the corrected temperature field data and the instantaneous electromagnetic field distribution data, directly check the extreme values. Check the temperature values at all positions in the temperature field data and record the maximum value. If the highest temperature exceeds 41 degrees Celsius, mark it as needing to be processed. Check the electric field intensity values at all positions in the electromagnetic field distribution data and record the maximum value. If the maximum electric field intensity exceeds 100 volts per meter, mark it as needing to be processed. Obtain the temperature difference between adjacent positions. If the temperature change rate exceeds 3 degrees Celsius per centimeter, mark it as needing to be processed. Output a safety status report. Combine the safety status report with the current standardized physiotherapy parameters (current value, frequency value, duty cycle). According to the abnormal conditions in the safety status report, directly modify the standardized physiotherapy parameters. If the temperature is too high and needs to be processed, the new current value is 90% of the current current value, and the new frequency value is 105% of the current frequency value. If the electric field intensity is too large and needs to be processed, the new duty cycle is 85% of the current duty cycle, and forcefully limit each parameter within the safe range (such as keeping the current between 5 - 100 milliamperes). Output a safety control instruction.

[0065] It should be noted that the safe range refers to the range of parameter values that have been clinically verified and can ensure the balance between therapeutic efficacy and safety during the operation of the physiotherapy device.

[0066] Execute safety constraints on the safety control instruction and generate an executable instruction, and output a physiotherapy device control instruction.

[0067] Specifically, based on the safety control instruction, use parameter standardization to convert all safety control instruction parameters into relative values within the range of 0 - 1. Set weights according to the physiotherapy stage. For example, the weight for the acute stage is 50% for current, 30% for frequency, and 20% for duty cycle. The weight for the recovery stage is 30% for current, 40% for frequency, and 50% for duty cycle. Output the integrated control parameters after fusion. Combine the integrated control parameters after fusion with the device safety specifications to conduct electrical safety verification. Check whether the current is within the range of 5 - 100 mA, verify whether the frequency is within the safe range of 10 - 150 Hz, confirm that the power does not exceed the rated maximum value, the single - time continuous duration is less than 30 minutes, and the intermittent time is greater than 5 minutes. Output the parameters that pass the safety verification. Use medical device communication protocol conversion for standard format encapsulation and output a safety control instruction.

[0068] It should be noted that setting weights according to the physiotherapy stage is based on medical rehabilitation principles, clinical empirical data, and device safety requirements; the electrical safety verification comes from medical device safety specifications, research results in bioelectromagnetics, and clinical practice evidence.

[0069] Preferably, through regular expression matching and standardization conversion, the uniformity of the physical therapy parameter format is improved, and the compatibility of data interaction is enhanced; through Maxwell's equations and finite element analysis, the calculation accuracy of the electromagnetic field distribution is improved, and the prediction reliability of tissue thermal effects is enhanced; through threshold control and dynamic parameter adjustment, the safety of the physical therapy process is improved, and the real-time response ability to abnormal situations is enhanced.

[0070] S5. Combine a constant current source circuit and pulse width modulation temperature control to execute the physical therapy device control instruction, and monitor the physical therapy effect in real time to generate a physical therapy efficacy evaluation report.

[0071] Furthermore, combine a constant current source circuit with a microcontroller to receive and execute the physical therapy device control instruction and output an electrical stimulation signal.

[0072] Specifically, an operational amplifier and a power field effect transistor are used to form a feedback control loop to monitor the output current in real time and adjust the driving voltage. A precision current sampling resistor is used to detect the actual current, and it is fed back to the control end through a differential amplifier to output a constant current source circuit; the microcontroller receives the physical therapy device control instruction to generate a corresponding PWM signal, and an isolation drive circuit is used to enhance safety and prevent high-voltage backflow from damaging the control circuit. A low-pass filter is used to smooth the PWM signal to reduce the interference of high-frequency noise on the nerves; by executing the physical therapy device control instruction, the microcontroller analyzes the instruction, sets the PWM duty cycle and frequency parameters, and the microcontroller outputs the PWM signal to the field effect transistor drive circuit to control the on and off of the current. The current sampling resistor monitors the output in real time and feedback-adjusts the gate voltage to maintain the target current. If the detected current exceeds the limit, the output is immediately turned off and an alarm is given to output an electrical stimulation signal.

[0073] It should be noted that the electrical stimulation signal is an adjustable pulse waveform that meets medical standards. For example, in the low-frequency analgesia mode: 1 - 10 Hz is used to relieve chronic pain, in the medium-frequency muscle strength training mode: 20 - 50 Hz is used for muscle activation, and in the high-frequency nerve inhibition mode: greater than 100 Hz is used for spasm control.

[0074] Monitor the temperature of the physical therapy area in real time, and based on the electrical stimulation signal, use a fuzzy PID controller to dynamically adjust the power of the physical therapy device to output a safe physical therapy temperature.

[0075] Specifically, an infrared temperature sensor is used to non - contact measure the skin surface temperature, which is collected by an analog - to - digital converter and transmitted to the main control unit of the physiotherapy device. A moving average filter is used to eliminate instantaneous noise, making the skin surface temperature data stable. The main control unit compares the real - time temperature with the target temperature to obtain the error, and dynamically adjusts the PWM duty cycle according to the error. If the temperature is too high and the error is less than - 1°C, the duty cycle is reduced, such as by 5%. If the temperature is insufficient and the error is greater than 1°C, the duty cycle is increased, such as by 5%. And the on - off of the heating element of the physiotherapy device is controlled through a field - effect transistor drive circuit; if the temperature exceeds the safety threshold (such as 42°C), the power output is immediately cut off and an audible and visual alarm is triggered, and the temperature rise rate is controlled, such as less than or equal to 1°C / min, to reduce discomfort caused by thermal shock; the infrared sensor and the thermocouple synchronously collect the temperature of the physiotherapy area, which is converted by the analog - to - digital converter and then input into the main control unit. Based on the temperature error and its change trend, the fuzzy PID controller of the electrical stimulation signal outputs a PWM duty - cycle adjustment instruction. In the low - temperature state: rapidly increase the temperature and increase the duty cycle (such as 80%); approaching the target: finely adjust the duty cycle (such as 45% - 55%); risk of over - temperature: forcibly cool down, reduce the duty cycle to 0% and start the cooling fan. The PWM signal drives the heating element, and at the same time, the temperature is continuously monitored to form a closed - loop control and output a safe physiotherapy temperature.

[0076] It should be noted that the setting of the safety threshold (such as 42°C) is based on a comprehensive consideration of the critical value of biological tissue thermal damage, clinical safety standards, and equipment reliability requirements.

[0077] Collect the subjective feedback and physiological index data of the patient, integrate the physiotherapy parameters and the safe physiotherapy temperature, and output a dataset for efficacy monitoring.

[0078] Specifically, during the physiotherapy process, the pain degree score (0 - 10 points), comfort level selection (1 - 5 levels), and local sensation description are obtained through patient prompts, and formatted subjective evaluation data is output; 8 - channel differential input is performed through surface electromyogram signal acquisition, blood oxygen saturation is measured through infrared light detection, and skin impedance measurement is obtained, and physiological parameter time - series data and characteristic parameters (such as electromyogram intensity, blood oxygen value, etc.) are output; the physiotherapy parameters are obtained and recorded in real - time, including electrical stimulation intensity, temperature control parameters, and the spatial position of the treatment head, and time - synchronized recording is performed, and physiotherapy parameter data and device operation logs are output; the environmental sensor group is used for periodic environmental monitoring, including temperature and humidity measurement, atmospheric pressure detection, and electromagnetic field intensity scanning, and noise filtering and abnormal data elimination are performed, and environmental parameter records and electromagnetic environment reports are output; combining the formatted subjective evaluation data, physiological parameter time - series data, characteristic parameters, physiotherapy parameter database, and device operation logs, data alignment and standardized format conversion are adopted, a unified time reference is established, and numerical normalization processing and format - unified conversion are performed to output a dataset for efficacy monitoring.

[0079] It should be noted that the subjective feedback of the patient includes pain degree scoring, comfort level selection, and local sensation description; the physiological index data includes surface electromyogram signal, blood oxygen saturation measurement, and skin impedance measurement.

[0080] Use a weighted statistical algorithm to analyze the efficacy monitoring data set and generate a physical therapy efficacy evaluation report.

[0081] Specifically, based on the efficacy monitoring data set, data cleaning and standardization calculations are performed. Data validity checks are carried out on the efficacy monitoring data set, blank records are deleted, and values beyond the reasonable range are corrected. For example, values of blood oxygen saturation greater than 100% are set to 100. The standard deviation is calculated for continuous variables, and numerical coding is performed for categorical variables. A standardized data set and a data quality description are output; the coefficient of variation method is used to obtain the coefficient of variation, combined with the preset basic weights, and an index weight distribution table is output; linear weighted summation is used to obtain the original score, which is then converted to a percentage system, and an individual efficacy score is output; using basic statistical functions, select indicators with a change amplitude greater than 15%, and obtain the mean difference before and after physical therapy, and use a report template to generate a physical therapy efficacy evaluation report.

[0082] It should be noted that the coefficient of variation is the ratio of the standard deviation to the mean, and a larger coefficient of variation is given a higher weight; the formula for the original score is the sum of the standardized values of each indicator multiplied by the corresponding weights.

[0083] Preferably, through a constant current source circuit and PWM temperature control technology, the stability of the electrical stimulation output is improved, and the accuracy and controllability of the physical therapy dose are enhanced; through fuzzy PID control and multi-sensor monitoring, the dynamic response speed of temperature regulation is increased, and the safety of thermal management in the physical therapy area is enhanced; through multi-modal data acquisition and weighted statistical algorithms, the objectivity of efficacy evaluation is improved, and the comprehensiveness and scientificity of physical therapy feedback are enhanced.

[0084] S6. According to the physical therapy efficacy evaluation report, use the stochastic gradient descent algorithm to update the weights of the convolutional neural network and output optimized physical therapy device control instructions.

[0085] Furthermore, sliding window statistics and standardization processing are adopted to extract physical therapy parameters and physiological index data from the physical therapy efficacy evaluation report, and a standardized feature vector and optimized target parameters are output.

[0086] Specifically, based on the physical therapy efficacy evaluation report, overlapping time windows are used to apply sliding windows to each data channel (such as current, temperature) respectively, extract statistical features such as the mean, variance, and slope within the window, use clustering algorithms to detect and remove outlier windows, and output a segmented time series dataset; extract time-domain features and frequency-domain features from the electromyogram signals, align the temperature data with the physical therapy parameters according to the timestamps, calculate the cross-correlation (such as the phase difference between temperature lagging behind current), use principal component analysis to compress the high-dimensional features into a low-dimensional space, output a fused feature matrix, perform Z-value standardization on the fused feature matrix, use SHAP value analysis to obtain the contribution degree of the features to the efficacy (such as the pain relief rate), generate a parameter adjustment coefficient (such as the current increase), and output a standardized feature vector and optimized target parameters.

[0087] It should be noted that each window of the segmented time series dataset corresponds to a set of statistical features; each row of the fused feature matrix represents the feature vector of a time window; the standardized feature vector is a numerical matrix with a mean of 0 and a variance of 1; the optimized target parameters are such as current +5mA, frequency -2Hz.

[0088] Combining the standardized feature vector and the optimized target parameters, apply the gradient descent algorithm to obtain the update direction and perform parameter adjustment within the safe range, and output the optimized combination of physical therapy parameters.

[0089] Specifically, combining the standardized feature vector and the optimized target parameters, make a small adjustment to each physical therapy parameter such as current ±1mA, compare the efficacy differences before and after the adjustment, determine the best adjustment direction for each parameter, and output the adjustment direction and amplitude of each parameter; make adjustments according to the adjustment direction and amplitude of each parameter, check whether the adjusted parameters are within the safe range, and if they exceed the safe range, set them to the boundary values, and output the preliminarily optimized parameter combination; combine the preliminarily optimized parameter combination with the real-time monitoring data, check whether the temperature is close to the safe range, if the temperature is higher than the safe range, reduce the stimulation intensity proportionally, and check whether the electromyogram signal is abnormal, and fine-tune the frequency according to the electromyogram signal, and output the optimized combination of physical therapy parameters.

[0090] Use structured data to encapsulate and convert the optimized combination of physical therapy parameters into device execution instructions, and output optimized physical therapy device control instructions.

[0091] Specifically, standardize the optimized physical therapy parameter combination. Round the parameter values according to the minimum resolution of the device, uniformly convert all time parameters to the second-level unit so that the physical therapy device can directly recognize them, and check whether each parameter value is within the range allowed by the physical therapy device. If it exceeds the range, automatically correct it to the closest allowed value and output a set of standardized parameter values. Select the corresponding communication protocol according to the connected device type, fill each parameter value into the corresponding instruction field according to the format requirements of the selected protocol, and add a check code at the end of the instruction to output the optimized physical therapy device control instruction.

[0092] Preferably, through sliding window statistics and feature standardization processing, improve the effectiveness of efficacy data extraction and enhance the pertinence of parameter adjustment; through the gradient descent algorithm and safety range constraint, improve the adaptability of neural network weight update and enhance the stability of physical therapy parameter optimization; through structured data encapsulation and protocol conversion, improve the standardization degree of instruction generation and enhance the reliability and compatibility of device execution.

[0093] This embodiment also provides an artificial intelligence-based physical therapy device control system, including: a signal processing module for real-time collecting electromyogram signals, performing band-pass filtering and TK energy operator processing, and outputting a muscle activation feature vector; a feature extraction module for extracting spatio-temporal features based on the muscle activation feature vector through a convolutional neural network and constructing a physical therapy parameter combination; a parameter regulation module for using a PID controller to adjust the physical therapy parameter combination in real time and outputting optimized physical therapy parameters; a simulation analysis module for performing electromagnetic-thermal coupling simulation on the optimized physical therapy parameters through finite element analysis to obtain a physical therapy device control instruction; An execution feedback module for combining a constant current source circuit and pulse width modulation temperature control to execute the physical therapy device control instruction, monitoring the physical therapy effect in real time, and generating a physical therapy efficacy evaluation report; an optimization iteration module for updating the weights of the convolutional neural network using the stochastic gradient descent algorithm according to the physical therapy efficacy evaluation report and outputting an optimized physical therapy device control instruction.

[0094] This embodiment also provides a computer device applicable to the case of the artificial intelligence-based physical therapy device control method, including: a memory and a processor; the memory is used for storing computer-executable instructions, and the processor is used for executing the computer-executable instructions to implement the artificial intelligence-based physical therapy device control method proposed in the above embodiment.

[0095] The computer device may be a terminal, which includes a processor, a memory, a communication interface, a display screen, and an input device connected via a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner. The wireless manner can be achieved through WIFI, carrier networks, NFC (Near Field Communication), or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads provided on the housing of the computer device. It can also be an external keyboard, touchpad, or mouse, etc.

[0096] This embodiment also provides a storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the method for controlling a physiotherapy device based on artificial intelligence as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device, or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic memory, flash memory, magnetic disks, or optical discs.

[0097] In summary, the present invention: optimizes PID parameters through the quantum annealing algorithm, combines the dynamic adjustment of stimulation parameters with the phase of the myoelectric β band to synchronize the treatment parameters with the physiological response; uses sliding window segmentation and fast Fourier transform to extract the temporal characteristics of myoelectric signals, and combines dilated causal convolution to enhance spatio-temporal correlation, making the feature extraction more suitable for the dynamic muscle activation state.

[0098] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A control method for a physiotherapy instrument based on artificial intelligence, characterized in that: including, real-time collecting electromyography (EMG) signals, performing band-pass filtering and TK energy operator processing, and outputting muscle activation feature vectors; extracting spatio-temporal features through a convolutional neural network based on the muscle activation feature vectors to construct a combination of physiotherapy parameters; using a PID controller to adjust the combination of physiotherapy parameters in real time and outputting optimized physiotherapy parameters; performing electromagnetic-thermal coupling simulation on the optimized physiotherapy parameters through finite element analysis to obtain physiotherapy device control instructions; combining a constant current source circuit and pulse width modulation temperature control, executing the physiotherapy device control instructions, and monitoring the physiotherapy effect in real time to generate a physiotherapy efficacy evaluation report; updating the weights of the convolutional neural network using the stochastic gradient descent algorithm according to the physiotherapy efficacy evaluation report and outputting optimized physiotherapy device control instructions.

2. The control method of the physiotherapy apparatus based on artificial intelligence according to claim 1, characterized in that: The steps for real-time collecting electromyography (EMG) signals, performing band-pass filtering and TK energy operator processing, and outputting muscle activation feature vectors are as follows. Real-time collecting electromyography (EMG) signals, combining a bioelectric amplifier and a Butterworth filter to remove noise and motion artifacts, and outputting pure EMG signals; Combining the Hilbert transform and the TK energy operator, extracting the signal envelope to obtain time-varying energy features, and using the statistical features of a sliding window to extract and output muscle energy features; Using the dynamic window statistics method to obtain energy integral values, constructing a spatial feature vector using a digital filter bank, and combining feature splicing to output muscle activation feature vectors.

3. The method for controlling a physiotherapy apparatus based on artificial intelligence according to claim 2, wherein: The steps for extracting spatio-temporal features through a convolutional neural network based on the muscle activation feature vectors to construct a combination of physiotherapy parameters are as follows. Using a sliding window to segment the window and performing a fast Fourier transform on the window signal to output time series feature data; Using dilated causal convolution to construct a spatio-temporal feature map and outputting a fused spatio-temporal feature vector; Using a two-branch predictor to synchronously optimize physiotherapy parameters and control safety risks to construct a combination of physiotherapy parameters.

4. The method for controlling a physiotherapy apparatus based on artificial intelligence according to claim 3, wherein: The steps for using a PID controller to adjust the combination of physiotherapy parameters in real time and outputting optimized physiotherapy parameters are as follows. Using a quantum annealing algorithm to optimize the initial PID parameters under the Lyapunov stability constraint and outputting preliminary optimized parameters; Combining the preliminary optimized parameters with the real-time pure EMG signals, dynamically adjusting the stimulation parameters according to the phase of the EMG beta band, and outputting the preliminarily optimized parameters with dynamic adjustment; Using incremental trial-and-error optimization and parameter backtracking for feature comparison and outputting optimized physiotherapy parameters.

5. The method for controlling a physiotherapy device based on artificial intelligence according to claim 4, wherein: The steps for performing electromagnetic-thermal coupling simulation on the optimized physiotherapy parameters through finite element analysis to obtain physiotherapy device control instructions are as follows. Converting the optimized physiotherapy parameters to standard units and using middleware to convert the input format to output a standardized set of physiotherapy parameters; Combining the standardized set of physiotherapy parameters with the real-time pure EMG signals, using Maxwell's equations to obtain the spatial electromagnetic field distribution, and outputting the instantaneous electromagnetic field distribution; Based on the instantaneous electromagnetic field distribution, discretizing the tissue region using the finite difference method, and combining weighted fusion to correct the real-time temperature value to output a corrected temperature field; Based on the corrected temperature field, using threshold control to monitor and dynamically adjust the physiotherapy parameters in real time and outputting safety control instructions; Performing safety constraints on the safety control instructions and generating executable instructions to output physiotherapy device control instructions.

6. The method for controlling a physiotherapy apparatus based on artificial intelligence according to claim 5, wherein: Combining the constant current source circuit and pulse width modulation temperature control, executing the control instructions of the physiotherapy device, and real-time monitoring the physiotherapy effect to generate a physiotherapy efficacy evaluation report. The specific steps are as follows: The constant current source circuit is combined with the microcontroller to receive and issue the control instructions of the physiotherapy device, and output an electrical stimulation signal; Real-time monitor the temperature of the physiotherapy area, and use a fuzzy PID controller to dynamically adjust the physiotherapy power to output a safe physiotherapy temperature; Collect the subjective feedback and physiological index data of the patient, and integrate the physiotherapy parameters and the safe physiotherapy temperature to output a therapeutic effect monitoring data set; Use the weighted statistical algorithm to analyze the therapeutic effect monitoring data set and generate a physiotherapy efficacy evaluation report.

7. The method for controlling a physiotherapy device based on artificial intelligence according to claim 6, characterized in that: According to the physiotherapy efficacy evaluation report, use the stochastic gradient descent algorithm to update the weights of the convolutional neural network and output the optimized control instructions of the physiotherapy device. The specific steps are as follows: Adopt sliding window statistics and normalization processing to extract the physiotherapy parameters and physiological index data from the physiotherapy efficacy evaluation report, and output the normalized feature vector and the optimized target parameters; Apply the gradient descent algorithm to obtain the update direction and perform parameter adjustment within the safe range, and output the optimized combination of physiotherapy parameters; Use structured data encapsulation and convert the optimized combination of physiotherapy parameters into device execution instructions, and output the optimized control instructions of the physiotherapy device.

8. A physical therapy instrument control system based on artificial intelligence, based on the physical therapy instrument control method based on artificial intelligence according to any one of claims 1 to 7, characterized in that: Including a signal processing module, a feature extraction module, a parameter regulation module, a simulation analysis module, an execution feedback module and an optimization iteration module; The signal processing module is used to collect the electromyogram signal in real time, and perform band-pass filtering and TK energy operator processing to output a muscle activation feature vector; The feature extraction module is used to extract spatio-temporal features based on the muscle activation feature vector through a convolutional neural network to construct a combination of physiotherapy parameters; The parameter regulation module is used to use a PID controller to adjust the combination of physiotherapy parameters in real time and output the optimized physiotherapy parameters; The simulation analysis module is used to perform electromagnetic-thermal coupling simulation on the optimized physiotherapy parameters through finite element analysis to obtain the control instructions of the physiotherapy device; The execution feedback module is used to combine the constant current source circuit and pulse width modulation temperature control, execute the control instructions of the physiotherapy device, and real-time monitor the physiotherapy effect to generate a physiotherapy efficacy evaluation report; The optimization iteration module is used to update the weights of the convolutional neural network according to the physiotherapy efficacy evaluation report by using the stochastic gradient descent algorithm and output the optimized control instructions of the physiotherapy device.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the artificial intelligence-based physiotherapy device control method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the artificial intelligence-based physiotherapy device control method according to any one of claims 1 to 7.

Images