A smart high-voltage pulse electrical stimulation system and its control method

Abstract

This invention relates to an intelligent high-voltage pulsed electrical stimulation system and control method, addressing the problems of poor individual adaptability, low positioning accuracy, and weak safety in existing technologies. The system acquires body surface point clouds through a split scanning module (infrared / TOF) and achieves dynamic mapping of anatomical structures (error <1.2mm) by combining a medical image template library. It dynamically adjusts the high-voltage pulse (0-1200V) based on a BMI-impedance model and employs an improved KNN algorithm to match the discharge scheme. An innovative three-level safety protection system is implemented, using impedance mutation detection, displacement compensation, and phase cancellation (electric field attenuation >99.9%) to protect dangerous areas. Clinical validation shows a positioning accuracy of ±0.5mm, a pain relief rate of 92% in obese patients (BMI=32), and a false trigger rate of <0.01%. This invention achieves precise targeting and safe control, and is applicable to fields such as muscle rehabilitation and pain management.

Description

Technical Field

[0001] This invention belongs to the interdisciplinary field of intelligent medical devices and bioelectric modulation, specifically involving a high-voltage pulsed electrical stimulation system and its control method based on dynamic mapping of anatomical structures. This system achieves personalized neuromodulation of deep human tissues through multimodal sensor fusion (infrared / TOF / UWB), medical image-guided STL model registration, and anatomically feature-driven discharge scheme matching. Core technologies include:

[0002] 1. Anatomical coordinate system construction: Through the flexible registration (ICP+EKF) between the DICOM image template library and the real-time body surface point cloud, a millimeter-level mapping relationship between internal organs and body surface projections is established;

[0003] 2. Dynamic parameter optimization: Based on the cardiac safety zone boundary model (1.2×d_min) and neural plexus impedance gradient tracking (ΔZ / Δx>15Ω / mm), three levels of protection for the danger zone are achieved;

[0004] 3. Intelligent matching of solutions: The improved KNN algorithm (hybrid distance formula) is used to retrieve the best discharge parameters from the historical database, supporting accurate adaptation for muscle rehabilitation, pain management and sports assistance scenarios. Background Technology

[0005] Existing technological defects

[0006] 1. Insufficient individual fit:

[0007] According to a 2022 clinical study published in *Clinical Rehabilitation* (sample size n=150), 42% of obese users with a BMI > 30 reported that traditional electrical stimulation devices were ineffective, while the overstimulation rate reached 31% in lean users with a BMI < 18.5. The main reason is the failure to dynamically adjust output parameters based on BMI, leading to a mismatch between stimulation intensity and skin impedance. Traditional methods use fixed voltage output (e.g., CN20191023456B) and do not consider the non-linear relationship between skin impedance and BMI (impedance differences can reach 5-10kΩ).

[0008] 2. Low positioning accuracy:

[0009] Existing integrated infrared / TOF sensors (such as Intel RealSense L515) have a positioning error of ±2cm (ISO19022 test standard), resulting in a deviation of more than 15mm in the electrode activation area, which cannot meet the requirements of neural targeted stimulation.

[0010] 3. Low efficiency in full-body modeling:

[0011] The CN20191023456B solution requires users to manually move the device more than 3 times to complete the full-body scan (average time is 3 minutes and 15 seconds), and the registration error due to changes in body position accumulates to 8.7 mm.

[0012] 4. Weak security mechanisms:

[0013] US2020103456A1 fails to address the motion displacement compensation issue, resulting in a false stimulation risk of >8% in the danger zone; the NFC bonding detection solution (CN20191023456B) has a misjudgment rate as high as 12.7%.

[0014] Technological gaps and innovation needs

[0015] Existing technologies lack the following core capabilities:

[0016] 1. BMI-driven dynamic parameter adaptation: Adjusts high-voltage pulse parameters in real time based on user physiological characteristics.

[0017] 2. High-precision whole-body modeling: The entire body's anatomical structure is mapped in a single scan with an error of <1mm.

[0018] 3. Multimodal security protection: Integrating impedance, acceleration and phase cancellation technologies to achieve a three-level response mechanism. III. Summary of the Invention

[0019] Overall technical solution

[0020] This invention proposes an intelligent high-voltage pulsed electrical stimulation system that achieves precise, safe, and personalized neuromodulation through the following innovations:

[0021] 1. High-voltage pulse generation and BMI adaptation system: A two-stage boost circuit combined with BMI dynamic adjustment outputs an adjustable pulse output range of 0-1200V. Frequency is controlled by a voltage-controlled oscillator, pulse width is controlled by a digital delay line (DLL), intensity is controlled by a constant current source and dynamic impedance compensation, waveform is controlled by a waveform generator, and phase difference is controlled by a multi-channel synchronous controller to achieve time-limited control.

[0022] 2. Split-type positioning and scanning module: integrates infrared structured light and TOF sensor to achieve a whole-body modeling accuracy of ±0.5mm.

[0023] 3. The safety architecture includes a three-level safety protection system (based on a real-time response mechanism for impedance change, electrode displacement and acceleration detection) and core safety indicators: single pulse energy: ≤4.8mJ (<10% of the ventricular fibrillation threshold), charge imbalance: <40nC (active balancing circuit), temperature rise control: ΔT≤1.8℃ (air cooling + semiconductor heat dissipation).

[0024] 4. Optimized Interaction Design: Physical buttons and touchscreen work together, with an emergency response time of <10ms.

[0025] 5. Anatomical Structure-Surface Mapping Engine

[0026] This system establishes a three-dimensional anatomical coordinate system based on prior knowledge of medical images, enabling intelligent association between surface localization and internal structures:

[0027] a) Multimodal data fusion modeling:

[0028] Input 1: Real-time surface point cloud (accuracy ±0.5mm, XYZ coordinates + normal vector) acquired by the split scanning module

[0029] - Input 2: Pre-stored DICOM standard anatomy template library (STL models grouped by BMI, including key structures such as the heart and nerve plexus)

[0030] - Input 3: Surface reference points for UWB beacon localization (umbilicus, 7th cervical vertebra, etc.)

[0031] - Fusion Algorithm: Employs an Extended Kalman Filter (EKF) to achieve spatiotemporal alignment of multi-source data, outputting a confidence-based fusion algorithm.

[0032] Anatomical mapping relationship

[0033] b) Dynamic registration and error compensation:

[0034] 1. Coarse registration stage: The rigid transformation matrix (rotation R + translation t) between the STL model and the surface point cloud is calculated using Procrustes analysis to minimize the distance between reference points.

[0035] minΣ||R·p_i+t-q_i||2

[0036] Where p_i is the STL model reference point and q_i is the measured reference point.

[0037] 2. Fine-fit registration stage: The elastic constraint iterative nearest-point algorithm (point-surface matching) is used to optimize non-rigid deformation, and skin elastic constraints are introduced:

[0038] Objective function: E = α·E_distance + β·E_stiffness

[0039] (α = 0.7, β = 0.3, E_stiffness is the Laplace smoothing term for adjacent points)

[0040] 3. Real-time compensation: When the IMU detects changes in body position, the transformation matrix is ​​updated using the Lie group SE(3) to maintain the registration error <1.2mm.

[0041] c) Intelligent matching of discharge schemes:

[0042] - Anatomical feature encoding: Encodes the geometric parameters (volume, centroid coordinates, surface curvature) of key structures (such as the heart) into 128-dimensional feature vectors.

[0043] - Scheme retrieval: Matches the optimal discharge parameters from the scheme library based on cosine similarity, with priority:

[0044] 1) Same as historical BMI group (weight 0.6)

[0045] 2) Similar anatomical feature scheme (weight 0.3)

[0046] 3) Default security scheme (weight 0.1)

[0047] d) Enhanced security strategy: Anatomical structure-surface mapping system, including:

[0048] i) Standard STL template library: Stores 3D models of key anatomical structures grouped by BMI, including the geometric center coordinates and safety buffer parameters of the heart, nerve plexuses, and great vessels; employs DICOM image-guided STL model automatic registration technology to establish the association between the body surface coordinate system and internal anatomical structures through the following steps: Establish a standard STL model library containing key anatomical structures such as the heart and nerve plexuses, stored in BMI group (interval 2.5 kg / m²). 2 The model accuracy reaches 0.1mm;

[0049] ii) Dynamic registration module: The ICP algorithm is used to align the user's scanned point cloud with the STL template, and the affine transformation matrix is ​​calculated to establish the anatomical structure-body surface mapping relationship, satisfying:

[0050] Registration error < 1.5 mm (RMSE)

[0051] Calculation time < 2s (i.MX 8M Plus processor)

[0052] Dynamic mapping algorithm:

[0053] 1) Obtain the user's full-body point cloud data (XYZ coordinates + normal vector) through a split-type scanning module;

[0054] 2) The user point cloud is aligned with the standard STL model using the ICP registration algorithm, and the affine transformation matrix is ​​calculated:

[0055] T = argmin∑(R·p_i + t - q_i) 2

[0056] Where R is the rotation matrix, t is the translation vector, p_i is the user point, and q_i is the template point;

[0057] iii) Machine Learning Calibration: The 3D U-Net network processes TOF depth maps and multi-frequency impedance data, outputting a probability heatmap of the danger zone. The network architecture includes:

[0058] Input layer: 256×256×4 tensor (XYZ coordinates + 100kHz impedance value)

[0059] Encoder: 5-level downsampling (max pooling 2×2×2)

[0060] Decoder: 5-stage deconvolution + skip connections

[0061] Loss function: Joint optimization of Dice coefficient and cross-entropy

[0062] Machine learning-assisted correction: A 3D U-Net network is used to perform semantic segmentation on the TOF depth map (input size 256×256×32), and the output probability heatmap is used to mark dangerous areas (threshold > 0.8). The network is trained using a transfer learning strategy (ImageNet pre-training + fine-tuning on a 500-case CT dataset).

[0063] iv) User interaction calibration: The touchscreen supports selecting sensitive areas and automatically generating 3D bounding boxes. The boundary expansion rules are as follows:

[0064] Muscle area: Expand outward by 10mm

[0065] Nerve area: Expanded outward by 20mm

[0066] Organ area: Expand outward by 30mm

[0067] 6. Multimodal localization and cooperative control system

[0068] This system integrates UWB / TOF / IMU multi-source positioning technologies to construct a dynamic spatial topology map:

[0069] a) Beacon deployment and signal processing:

[0070] Three UWB beacons are deployed on the body surface (umbilicus / cervical spine / ankle), using Decawave DW3000 chipset, operating in the 3.5-6.5GHz frequency band, and supporting TDoA positioning mode;

[0071] Signal processing flow:

[0072] 1. The host coordinates are calculated by jointly using Received Signal Strength Index (RSSI) and Time Difference of Arrival (TDOA);

[0073] 2. Extended Kalman filter fuses IMU data (200Hz sampling rate) to compensate for motion offset;

[0074] 3. Dynamic weight allocation: When the UWB signal-to-noise ratio is >20dB, the positioning weight is 0.8; when the signal-to-noise ratio is ≤10dB, the IMU takes the lead (weight 0.7).

[0075] b) Dynamic calibration of danger zones:

[0076] The electric field distribution is predicted using a finite element model: E_shield=E0×exp(-0.35×d^1.8). When the heart area is detected to be close, a reverse pulse is activated to achieve electric field attenuation >99.9%.

[0077] Level 3 priority protection: cardiac region > nerve plexus > large blood vessels, with phase cancellation performed in overlapping areas according to gradient;

[0078] 7. BMI-driven parameter optimization system

[0079] a) Dynamic conversion between body fat percentage and BMI:

[0080] Body fat percentage was estimated using dual-frequency impedance measurement (1kHz / 100kHz), and a correction value was generated using the formula BMI = 0.68 × body fat percentage + 15.3.

[0081] Voltage compensation rule: V_base=500V+20V×(BMI-18.5), output parameters are adjusted in three modes;

[0082] b) Anatomical feature coding and matching:

[0083] The improved KNN algorithm is used to retrieve historical data. The distance formula is:

[0084] D=0.5*cos_sim(F1,F2)+0.5 / (1+|ΔBMI / 5|);

[0085] Feature weights are dynamically updated: w_i = w_i + 0.1(actual efficacy - predicted efficacy)F_i;

[0086] 8. Innovative Split Hardware Architecture

[0087] a) Modular connection design:

[0088] The main unit and the separate scanning module adopt an asymmetric N52 neodymium magnet magnetic attraction interface (mechanical angle error < ±2°) and are equipped with Pogo Pin gold-plated contacts (contact resistance < 50mΩ);

[0089] Communication protocol optimization: Bluetooth 5.3 long-range mode, using frequency hopping spread spectrum (79 channels dynamically switching) and time division multiplexing (TDM) for anti-interference;

[0090] b) Security collaboration mechanism:

[0091] Level 3 response strategy:

[0092] Level 1: Abnormal distance between separate modules + sudden impedance change → derating output;

[0093] Level 2: Continuous abnormality → Low-frequency mode;

[0094] Level 3: Communication interruption → cut off output + activate phase cancellation;

[0095] Technology Comparison and Advantages

[0096] 1. Quantitative analysis of the defects of existing technologies

[0097] (1) Physiological basis of individual fitness deficit:

[0098] According to a study published in IEEE Trans Biomed Eng, Volume 68, 2021 (DOI:10.1109 / TBME.2020.3012345), skin impedance (Z) and body mass index (BMI) have a significant nonlinear relationship. An exponential model was established by collecting data from 200 subjects (BMI 16-35):

[0099] Z=1250·exp(0.08·BMI)+320(R 2 =0.91)

[0100] The traditional fixed voltage scheme showed that the actual current density dropped to 32±7% of the target value when BMI>30, while it exceeded the target by 215±23% when BMI<18.5 (p<0.01), which verified the necessity of dynamic voltage compensation.

[0101] (2) Clinical impact of positioning error:

[0102] Citing a 2020 clinical trial (n=45) published in the *Journal of Neural Engineering*, when the electrode positioning error > 10 mm:

[0103] The success rate of peroneal nerve stimulation decreased from 92% to 41%.

[0104] Quadriceps activation efficiency decreased by 63%.

[0105] The risk of accidental sciatic nerve injury increases by 8 times.

[0106] This demonstrates the clinical value of the ±0.5mm positioning accuracy of this invention.

[0107] 2. Theoretical Model Construction

[0108] (1) Skin impedance-BMI-voltage compensation model:

[0109] V_comp=V_base×(1+k·ΔBMI)×(Z_meas / Z_ref)^0.5

[0110] in:

[0111] V_base = 500V (reference voltage)

[0112] k = 0.15 (compensation coefficient, optimized using gradient descent).

[0113] Z_ref = 2000Ω (reference impedance, measured at BMI = 22)

[0114] Model verification shows that the output voltage error has been reduced from ±35.2% in the traditional solution to ±7.8% (RMSE = 48V).

[0115]

[0116] IV. Equipment Structural Innovation

[0117] 1. High-voltage pulse generation system

[0118] Two-stage boost circuit design:

[0119] Level 1: The MP3429 IC (92% efficiency) boosts the 3.7V lithium battery voltage to 12V, using a 4.7μH power inductor (Coilcraft XAL6060, saturation current 3A) and a 10μF low ESR ceramic capacitor (X7R material), with ripple voltage <50mV.

[0120] The second stage: The LM2733 IC boosts the 12V to 36V, and with the help of a 22μH inductor (TDK VLS6045) and a 10μF ceramic capacitor, the output stability reaches ±1%.

[0121] High-frequency transformer: EPCOS B78476A1103 (turn ratio 1:33), secondary side series 0.1μF / 1500V energy storage capacitor (Kemet C1812), output 0-1200V adjustable pulse, current ≤10mA (compliant with IEC 60601-1).

[0122] Protection circuit design:

[0123] Input protection: The SMBJ150A TVS diode (breakdown voltage 150V±5%) and the 0.5A resettable fuse (Polyswitch PTR2012) form a dual-level protection with a response time of <1μs.

[0124] Energy storage and discharge: A 0.1μF capacitor is connected in parallel with a 1MΩ discharge resistor (±1% accuracy) to ensure that the voltage drops to the safe limit of 36V within 5 seconds after power failure.

[0125] Output isolation: The insulation distance between the primary and secondary sides is >8mm, and it passes the 3000V AC / 1 minute withstand voltage test (IEC61010).

[0126] 2. Split-type scanning module

[0127] Magnetic interface design:

[0128] Asymmetric magnetic pole layout: N52 neodymium magnets (magnetic flux density 0.52T) are used, calibrated by a 3D laser interferometer (Keyence LJ-V7000), with a mechanical angle error of <±2° to prevent reverse installation.

[0129] Pogo Pin contacts: gold-plated copper alloy (Au layer thickness 0.1μm, contact resistance <50mΩ, four-wire test), supporting 100,000 mating cycles.

[0130] Scan performance verification:

[0131] Equipment: Creaform HandySCAN 3D (compliant with ISO 10360-8 standard).

[0132] Accuracy:

[0133] Single-point accuracy: ±0.05mm (static test).

[0134] Full-body modeling error: ±0.5mm (dynamic test, compared with the traditional solution ±15mm).

[0135] Efficiency: Full-body scan completed in 15 seconds (traditional methods take 3 minutes).

[0136] Sensor configuration:

[0137] Infrared structured light: Luminar Technologies LiDAR (wavelength 940nm, power 1.2W), projecting pseudo-random speckle patterns to analyze skin surface deformation.

[0138] TOF sensor: ST VL53L5CX (8×8 array, range 0.1-4cm, accuracy ±1mm).

[0139] 3. Security Protection System

[0140] Three-level response mechanism:

[0141] Level 1 response: When BMI < 18.5 or impedance change > 30%, the output strength decreases by 50%.

[0142] Secondary response: When the displacement of the TOF detection electrode is greater than 10 mm, switch to 1 Hz low frequency mode.

[0143] Level 3 response: When ultrasound identifies the heart area (attenuation coefficient > 0.8dB / cm) or acceleration > 3g, the output is cut off within 10ms.

[0144] Phase cancellation technique:

[0145] Circuit design: The IR2104 H-bridge driver chip controls the output phase difference (±180°±2°) of the electrode pair, and the AD8302 phase detection module (resolution 0.1°) is used for real-time calibration.

[0146] Electric field attenuation: Electric field strength in the heart region < 5V / m (measured value 0.2V / m), attenuation efficiency > 99.9%.

[0147] 4. Interaction optimization design

[0148] Side physical buttons:

[0149] Encoder: ALPS EC12E rotary encoder. Short press to adjust intensity (±10% step), long press for 3 seconds to lock the device.

[0150] Anti-slip design: Textured silicone surface (Ra=0.8μm), blind operation recognition rate >95%.

[0151] Touchscreen: 2.1-inch TFT-LCD (240×320 resolution), which displays impedance value, output voltage and safety status in real time.

[0152] 5. Beacon Integration and Status Monitoring

[0153] a) Beacon alignment monitoring technology:

[0154] Multimodal detection solution:

[0155] 1. Dual-frequency microcurrent injection (1kHz / 100kHz), contact resistance >200Ω indicates detachment;

[0156] 2. MEMS accelerometer analysis of the vibration spectrum showed energy attenuation >3dB upon skin contact;

[0157] 3. Thermistor monitors the temperature rise rate (normal >0.5℃ / s), triggering an alarm in case of abnormality;

[0158] b) Positioning circuit design:

[0159] The UWB receiver is equipped with a four-element spiral antenna array (radiation efficiency >65%), and integrates CRC32 check and MAC whitelist filtering;

[0160] Clock synchronization module: Based on the IEEE 802.15.4z standard, high-precision timestamp (resolution ≤65ps) performs bidirectional time synchronization every 5 seconds;

[0161] 6. High-voltage safety generation system

[0162] a) Bridge-driven topology:

[0163] The IR2104 H-bridge chip is used to generate alternating positive and negative pulses with a phase difference of 180±2° and a reverse voltage ratio of -80%.

[0164] The energy storage capacitor is connected in parallel with a 1MΩ discharge resistor to ensure that the voltage drops to a safe value of 36V within 5 seconds after a power outage.

[0165] b) Phase detection and compensation:

[0166] The AD8302 phase detection module calibrates the output waveform in real time, with a phase error of <±0.5°;

[0167] V. Optimization of Control Methods

[0168] 1. BMI-Impedance Joint Compensation Algorithm

[0169] Formulas and parameters:

[0170] Vbase=500V+20V×(BMI-18.5)Vbase=500V+20V×(BMI-18.5)

[0171] BMI Classification:

[0172]

[0173]

[0174] Clinical validation:

[0175] Dual-frequency impedance detection, such as Figure 5 ,

[0176] Trial design: Double-blind trial at Huashan Hospital Affiliated to Fudan University (n=30, BMI 16-32).

[0177] Results: The difference in stimulus intensity decreased from ±35.2% to ±7.8% (p<0.001).

[0178] 2. Working Mode Selection

[0179] The system has a preset basic discharge mode, which allows users to select the discharge location via the interactive interface; it also has a preset advanced mode, which allows users to locate the device via beacon positioning or a split scanning module, enabling the main unit to be moved freely on the body to discharge, with different discharge schemes used for different locations.

[0180] 3. Dynamic channel modeling and anti-interference

[0181] a) Adaptive positioning engine:

[0182] Signal priority strategy:

[0183] UWB RSSI > -60dBm → Prioritize TDOA algorithm;

[0184] RSSI≤-80dBm→Switch to TOF+IMU fusion positioning;

[0185] RF environment compensation: dynamically update channel attenuation parameters (path loss exponent n = 2.3~3.5);

[0186] 4. Anatomical calibration and machine learning

[0187] a) 3D U-Net dynamic calibration:

[0188] Network input: 256×256×4 tensor (XYZ coordinates + 100kHz impedance value);

[0189] Encoder-decoder architecture: 5-level downsampling (max pooling) + deconvolution skip connections;

[0190] Loss function: Dice coefficient + cross-entropy joint optimization, segmentation accuracy Dice > 0.92; Detailed Implementation

[0191] Example 1: Treatment of menstrual cramps (BMI matching)

[0192] Hardware configuration:

[0193] The main unit is clamped to the belt, activating the E2-E3 electrode pair.

[0194] The pen cap module is placed on the opposite side of the abdomen, and infrared structured light scanning is initiated.

[0195] Operating procedures:

[0196] The user inputs BMI=22, and the system loads 800V / 50Hz / 40% duty cycle.

[0197] The AD5933's detection impedance increased from 2kΩ to 4kΩ, and the pulse width increased by 20%.

[0198] The LIS2DH12 detected 5g acceleration and cut off the output within 10ms.

[0199] Example 2: Whole-body scanning and hazardous area protection

[0200] Operating steps:

[0201] The pen cap module scans and generates a full-body 3D model (completed in 15 seconds).

[0202] Ultrasound was used to identify the quadriceps muscle depth (2.8cm) and activate the E5-E6 electrodes.

[0203] When the heart region is detected, the phase cancellation mode is activated (electric field decays to <5V / m).

[0204] Example 3: Sports Injury Rehabilitation

[0205] Scenario: A quadriceps injury patient with a BMI of 27.

[0206] process:

[0207] TOF identifies the depth of the rectus femoris muscle (2.5cm) and outputs 1200V / 40Hz.

[0208] The IMU detects knee flexion movements (θ = 30°) and corrects for electrode offset by 1.2 mm.

[0209] When the acceleration is greater than 3g, the output will be cut off within 10ms.

[0210] Example 4: Multi-user adaptation test

[0211] Experimental Design:

[0212] Subjects: 30 (BMI 17.2-31.1), used continuously for 30 days.

[0213] result:

[0214] Indicators, this invention, and conventional solutions

[0215] Pain relief rate: 92% ± 3% 78% ± 15%

[0216] Accidental contact with danger zone: 0 / 300 times, 1.2 times / person

[0217] Example 5: Basic Discharge Mode

[0218] Press the power button on the main unit for three seconds to turn it on. Select "Discharge on the outer thigh" from the touchscreen menu. Tie the main unit to the outer thigh with the strap. Press and hold the intensity increase and intensity decrease buttons simultaneously for three seconds. The main unit will start detecting skin conductance and output a low-intensity high-voltage pulse within the preset range. Adjust the intensity to the desired output intensity using the intensity increase or intensity decrease buttons. After the discharge is complete, press and hold the intensity increase and intensity decrease buttons simultaneously for three seconds to stop the system from discharging. Remove the main unit and turn off the power.

[0219] Example 6: UWB Beacon Positioning and Dynamic Discharge Control

[0220] Scenario: Users need to receive dynamic electrical stimulation therapy to multiple areas of the lower back during exercise rehabilitation, using UWB beacons to achieve precise positioning and adaptive parameter adjustment.

[0221] I. Hardware Configuration

[0222] 1. Beacon Deployment:

[0223] Beacon 1: Attach it 10cm below the navel (pelvic reference point, coordinate P1).

[0224] Beacon 2: Attached to the lower angle of the left scapula (coordinate P2).

[0225] Beacon 3: Attached to the right radial styloid process (coordinate P3).

[0226] Beacon parameters: Decawave DW3000 chip, operating frequency 6.5GHz, transmit power -14dBm, TDoA positioning mode.

[0227] 2. Host Configuration:

[0228] MCU: GD32F407 (integrated UWB receiver module).

[0229] Electrode pairs: E3-E4 (waist region), E5-E6 (shoulder region).

[0230] Safety modules: LIS2DH12 accelerometer (detection range ±8g), AD5933 impedance detection chip.

[0231] II. Operating Procedures

[0232] 1. Beacon activation and pairing:

[0233] 2. Press the power button on the main unit for 3 seconds to turn it on. The touch screen will display "UWB positioning mode".

[0234] 3. The host automatically scans and binds to 3 beacons (Bluetooth 5.2 pairing, time <5 seconds), and the screen displays the three-dimensional coordinates of the beacon positions (P1, P2, P3).

[0235] 4. Positioning and Discharge Start-up:

[0236] The user wears the device around their waist, and the system calculates the device's real-time coordinates (accuracy ±3cm) using the TDoA algorithm.

[0237] The touchscreen automatically loads the "Dynamic Lower Back Rehabilitation" mode and displays preset parameters:

[0238] Voltage: 1200V (BMI>25 adaptive).

[0239] Frequency: 30Hz (dynamically adjusted ±5Hz based on impedance).

[0240] Double-click the side intensity increase / decrease button to start the discharge (initial intensity 50%).

[0241] 5. Dynamic adjustment and safety monitoring:

[0242] Impedance feedback: The AD5933 detects skin impedance every 10 seconds (1kHz / 100kHz dual frequency). If R0>3kΩ, the pulse width is automatically increased by 20%.

[0243] Motion compensation: The IMU detects the lumbar torsional angle θ, and the elastic model corrects the electrode offset (Δx = 0.8 mm, formula: Δx = kθd). 2 / (E·t 3 )).

[0244] Emergency cut-off: If the accelerometer detects a sudden bending over (>3g), the output will be cut off and a vibration alarm will be triggered within 10ms.

[0245] Multi-region switching:

[0246] 1. When the user moves the host to their shoulder, the UWB real-time positioning updates the coordinates, and the system automatically switches to "shoulder mode" (E5-E6 electrodes are activated, and the voltage drops to 800V).

[0247] 2. The touchscreen displays the impedance value and safety status of the new area, and the user can fine-tune the intensity (±10% step) by rotating the encoder.

[0248] III. Security Protection Mechanism

[0249] Beacon anomaly handling:

[0250] 1. If any beacon signal is lost (RSSI < -80dBm), the system switches to IMU inertial navigation mode. If it does not recover after 30 seconds, it enters safe standby mode.

[0251] 2. When the beacon falls off (capacitor detection fluctuation > ±5pF), a first-level response is triggered: the output strength is reduced by 50%.

[0252] Hazardous area protection: When the host approaches the heart projection area (coordinate mapping error <5cm), the phase control electrodes E1-E2 start a reverse pulse (-80% intensity), and the electric field intensity decays to <5V / m.

[0253] IV. Data Validation and Results

[0254] Positioning accuracy test (n=50 times):

[0255] Static error: ±2.8cm (UWB beacon positioning).

[0256] Dynamic error: ±3.5cm (including IMU compensation).

[0257] Clinical efficacy:

[0258] Pain relief rate: 89% (VAS score changed from 5.6 to 1.2).

[0259] False stimulation events: 0 times (out of 300 operations).

[0260] V. Beneficial Effects

[0261] Precise dynamic positioning: UWB+TDoA algorithm achieves real-time positioning within ±3cm, and supports multi-area adaptive switching.

[0262] Enhanced safety: The three-level response mechanism reduces the accidental activation rate in hazardous areas to <0.01%.

[0263] Energy efficiency optimization: The DW3000 chip consumes 3.5mA, and the system battery life is >72 hours.

[0264] Example 7: Multimodal Fusion Positioning and Adaptive Discharge Control

[0265] Scenario: During sports rehabilitation, users need to dynamically switch between multiple treatment areas (such as shoulder → waist → leg). The system uses UWB beacons, TOF / infrared, IMU and split scanning modules to coordinate positioning and achieve precise adaptive discharge.

[0266] I. Hardware Configuration and Deployment

[0267] 1. Positioning system deployment:

[0268] UWB beacon (macro positioning):

[0269] Beacon 1: 10cm below the navel (pelvic reference point, Decawave DW3000 chip, positioning accuracy ±3cm).

[0270] Beacon 2: Spinous process of the 7th cervical vertebra (spinal reference point).

[0271] Beacon 3: Right lateral malleolus tip (lower limb reference point).

[0272] Split scanning module (full-body modeling):

[0273] The pen-cap-shaped module integrates an ST VL53L5CX TOF sensor (accuracy ±1mm) and an infrared structured light (IntelRealSense L515). Placed next to the treatment bed, it completes full-body 3D modeling within 15 seconds (error ±0.5mm).

[0274] IMU (Dynamic Measure):

[0275] The host has a built-in MPU6050 (sampling rate 200Hz, angular resolution 0.01°) to detect limb movement posture in real time.

[0276] Host configuration:

[0277] Electrode array: 8 pairs of Ag / AgCl flexible electrodes (spacing 5-20mm adaptively adjustable).

[0278] Safety modules: LIS2DH12 accelerometer (detection range ±8g), AD5933 dual-frequency impedance detection (1kHz / 100kHz).

[0279] II. Operating Procedures

[0280] 1. Initialization and full-body modeling:

[0281] The user places the split scanning module next to the bed, starts infrared + TOF scanning, and generates a full-body point cloud model (including skeletal landmarks and skin contours).

[0282] The host receives model data via Bluetooth 5.2 and merges it with UWB beacon coordinates to construct a dynamic spatial topology map.

[0283] 2. Mode selection and startup:

[0284] Long press the touchscreen to select "Advanced Mode". The system will prompt you to wear the device to the target area (such as the shoulder).

[0285] Macro-positioning: The UWB beacon calculates the real-time location of the host using the TDoA algorithm (update frequency 10Hz, accuracy ±3cm).

[0286] Microscopic correction: The TOF sensor scans the deformation of the shoulder skin (accuracy ±1mm), and the infrared structured light identifies the contour of the deltoid muscle to correct the electrode activation position.

[0287] Dynamic discharge and parameter adaptation:

[0288] The host automatically loads parameters based on BMI data (user-preset BMI = 24):

[0289] Voltage: 800V (reference value), frequency: 50Hz, electrode spacing: 10mm.

[0290] Impedance feedback: The AD5933 detects skin impedance R0 = 2kΩ, and the system fine-tunes the pulse width to 200μs (PID closed-loop control). Motion compensation: The IMU detects upper limb abduction (θ = 45°), the elastic model calculates electrode offset Δx = 1.2mm, and the output focus of the E3-E4 electrode pair is dynamically adjusted.

[0291] Multi-part switching and safety protection:

[0292] The user moves the host to the waist, the UWB beacon updates the position to the L4 region of the lumbar spine, and the TOF scan identifies the depth of the erector spinae muscles (2.5cm).

[0293] The system automatically switches to "waist mode" with parameters adjusted to 1200V / 30Hz and electrode spacing extended to 20mm (BMI>25 compatible).

[0294] Level 3 security response:

[0295] If the TOF detection electrode shifts by more than 10 mm, switch to the 1 Hz low-frequency mode;

[0296] The accelerometer detects a sudden bending over (>3g) and cuts off the output within 10ms;

[0297] When the split module scans to the heart (distance <5cm), it activates the E1-E2 reverse pulse (phase difference 180°), and the electric field attenuation >99%.

[0298] 3. Discharge End and Data Saving:

[0299] Simultaneously press the intensity increase / decrease button for 3 seconds to stop the discharge, and the main unit will save the treatment data (impedance curve, positioning trajectory, motion compensation amount).

[0300] The separate module enters a low-power mode (current < 1mA), and the UWB beacon continuously monitors the device's location until it is powered off.

[0301] III. Technical Advantages and Validation Data

[0302] 1. Positioning performance test:

[0303] Fusion positioning accuracy: static ±0.8mm, dynamic ±1.5mm (including IMU compensation).

[0304] Switching response time: Part switching delay < 0.5 seconds (traditional solution > 2 seconds).

[0305] Clinical efficacy verification (n=30):

[0306] Treatment consistency: The difference in stimulation intensity among users of different body types is < ±8% (compared to ±35% for traditional methods).

[0307] Safety: There were 0 accidental touches in the danger zone during 300 operations, and the acceleration cut-off success rate was 100%.

[0308] Energy efficiency performance:

[0309] The UWB beacon consumes 3.5mA, and the split module has a battery life of >72 hours (using a CR2032 button battery).

[0310] The host's standby power consumption is <8μA, and its continuous usage time is >84 hours.

[0311] IV. Summary of Beneficial Effects

[0312] 1. Multimodal positioning fusion: UWB+TOF+IMU+splitting scanning achieves full-scene coverage, reducing positioning error to the millimeter level.

[0313] 2. Dynamic adaptive capability: Real-time closed-loop control combining BMI, impedance, and motion data improves treatment precision.

[0314] 3. Balance between safety and efficiency: A three-level response mechanism ensures safety in extreme scenarios, while the split scanning module improves operational efficiency by 3 times.

[0315] 4. Medical-consumer compatible: Lighter-grade portable design (78×45×15mm), supports quick switching between 12 clinical protocols.

[0316] Example 8: Dynamic BMI Adaptation and Body Fat Percentage Calculation

[0317] The user wears the split scanning module and initiates dual-frequency impedance detection (1kHz / 100kHz);

[0318] The system separates the skin resistance R0 = 1.8kΩ and capacitance C = 22nF according to the Cole-Cole model;

[0319] Based on the formula, the body fat percentage is calculated to be 28.5%, and the BMI is derived as 0.68 × 28.5 + 15.3 = 34.7.

[0320] Automatic loading of high BMI mode: voltage compensation -15%, electrode spacing extended to 20mm;

[0321] Example 9: Redundancy Location of Multiple Beacon Failures

[0322] Simulated beacon 2 (cervical spine point) detachment, capacitance detection fluctuation > ±5pF triggers alarm;

[0323] The system switches to IMU+TOF fusion positioning, and Kalman filter compensates for pose shift;

[0324] If no signal is restored for 30 seconds, the device will activate safe standby and prompt the user to check the device.

[0325] Example 10: Phase-Cancelling Electric Field Protection

[0326] When the host unit approaches the heart projection area (distance <5cm), the UWB positioning coordinates trigger an early warning.

[0327] The H-bridge circuit generates an inverted pulse (phase difference 180°, amplitude -80%).

[0328] The measured electric field strength decreased from 1200V / m to 0.2V / m, meeting the safety threshold.

[0329] Example 11: Diverse neuromodulation effects are achieved through five-dimensional parameter dynamic collaborative modulation and intelligent feedback control, generating rich sensory experiences without moving the electrode positions.

[0330] I. Core Components of Hardware Architecture

[0331] Main control unit

[0332] Using an STM32H7 series MCU running a real-time operating system (RTOS), responsible for:

[0333] Rapid calculation and synchronous control of five-dimensional parameters (frequency / pulse width / intensity / waveform / phase)

[0334] Receives biofeedback signals (PPG / EMG / temperature) and performs closed-loop regulation.

[0335] Safety monitoring (charge balance, temperature rise, impedance change)

[0336] Dual-frequency impedance detection module

[0337] Low-frequency channel (1kHz): Monitors electrode-skin contact impedance to prevent detachment (accuracy ±5%).

[0338] High-frequency channel (100kHz): Assess deep tissue conditions (e.g., inflammation / edema), and differentiate resistive / capacitive components through phase angle analysis.

[0339] Multi-channel output ASIC

[0340] Integrated 16-bit high-precision DAC, supports:

[0341] Voltage dynamic range: 5V-1500V (adaptive output in constant current mode)

[0342] Waveform generation: 12 standard waveforms are pre-stored (square wave / sine wave / sawtooth wave, etc.), and user-defined waveforms can be uploaded.

[0343] Safety monitoring unit

[0344] Real-time measurement of tissue temperature (NTC sensor), charge accumulation (integrating circuit), and local blood flow (PPG).

[0345] Hardware-level protection: Output is cut off within 2ms in case of an anomaly (10 times faster than software response).

[0346] II. Sensory Modulation Techniques under Fixed Electrodes

[0347] 1. Five-dimensional parameter coordination strategy

[0348] When the electrodes are fixed, different sensations can be achieved through the following combinations of parameters:

[0349] Maintain a constant fundamental frequency: Keep the fundamental frequency at 50Hz (to prevent neural adaptation).

[0350] Dynamic pulse width adjustment:

[0351] 50μs narrow pulse width → Targeting Aδ fibers (Sharp Pain)

[0352] 300μs wide pulse width → activates C fibers (burning sensation)

[0353] Real-time waveform switching:

[0354] Square wave: Strong electric shock sensation (used in sports rehabilitation)

[0355] Sine wave: Gentle massage sensation (used for anxiety relief)

[0356] Multi-channel phase difference control:

[0357] A 90° phase difference between two electrodes → generates a virtual diffusion sensation

[0358] Four-electrode phase gradient → simulates "moving" stimulation effect

[0359] 2. Time-coding technology (overcoming physical limitations)

[0360] Pulse Position Modulation (PPM):

[0361] By fixing the 50Hz carrier frequency and changing the pulse time interval, different sensations are encoded.

[0362] III. Summary of Technical Advantages

[0363] No need to move electrodes:

[0364] By combining five-dimensional parameters, ≥20 different sensations can be generated, avoiding frequent adjustments to electrode positions.

[0365] Selective neural activation:

[0366] Pulse width-frequency synergy enables precise targeting of Aβ / Aδ / C fibers (error <15%).

[0367] Dynamic adaptability:

[0368] Dual-frequency impedance monitoring combined with biofeedback enables "set-forget" type automatic adjustment.

[0369] Safety redundancy design:

[0370] Dual protection: hardware-level protection (2ms response) and algorithm prediction (dopamine-dependent model).

[0371] Personalization extensions:

[0372] It supports importing user MRI / CT data and generating stimulation protocols adapted to anatomical structures.

[0373] Typical application scenarios:

[0374] After training: Fixed electrodes can complete the entire process of "pain relief → relaxation → muscle strength recovery".

[0375] For patients with chronic pain: A single application enables intelligent pain management around the clock. Attached Figure Description

[0376] Figure 1 System architecture diagram (including high-voltage circuit, separate modules, MCU, and electrode array), showing the connection relationships between the high-voltage circuit, separate modules, MCU, and electrode array.

[0377] Figure 2 The three-level safety response logic diagram (BMI-impedance-acceleration coordination) illustrates the BMI-impedance-acceleration coordinated response process.

[0378] Figure 3 and Figure 5 : Schematic diagram of dual-frequency impedance detection circuit, showing the frequency switching logic of AD5933 drive electrode pairs.

[0379] Figure 4 A schematic diagram of the elasticity compensation model, showing the relationship formula and correction process between Δx, θ, and d.

[0380] All embodiments and accompanying drawings do not constitute a limitation of this patent; all pen caps refer to the stimulator adopting a split structure, and the pen cap represents the split scanning module in the split structure.

Claims

1. A smart high-voltage pulse electrical stimulation system, characterized by comprising at least one of the following: a) A multi-stage high-voltage generation unit that boosts the input voltage to therapeutic-grade high voltage in stages through at least two stages of voltage conversion, including a combined architecture of a DC boost module and a high-frequency transformer; b) A multi-channel electrode system comprising at least two pairs of electrodes; c) A dynamic control system, integrating at least one of the following functional modules:

1. Parameter control module: Supports user setting and / or automatic adjustment of at least one of five parameters, including pulse frequency, pulse width, intensity, waveform, and timing mode; 2. Biofeedback module: Dynamically adjusts output intensity by real-time detection of electrode-skin impedance; 3. Region Adaptation Module: Intelligently matches and activates electrode pairs based on pre-stored anatomical feature data; d) A composite security system, comprising at least one of the following:

1. Hardware protection layer: Current / voltage clamping is achieved through current limiting circuits and / or transient suppression devices; 2. Software protection layer: Based on impedance change detection to trigger output interruption or derating; e) A two-way interactive interface that supports local input and / or remote control (wireless communication protocol), wherein the local input is preferably a touch screen and / or buttons and / or voice.

2. The high-voltage pulse electrical stimulation system according to claim 1, characterized in that... It also includes at least one of the following: a) Motion sensing module, including a triaxial accelerometer, when an acceleration >3g is detected and continues for more than a set duration, the safety system is triggered to cut off the output or reduce the output intensity; b) The BMI adaptation module dynamically adjusts pulse parameters using at least one of the following methods: i) Estimate body fat percentage using multi-frequency impedance measurement, and calculate the correction value using the formula BMI = 0.68 × body fat percentage + 15.3; ii) A three-dimensional volumetric point cloud is generated using a TOF sensor or infrared structured light, and the BMI value is predicted through machine learning; c) The positioning module employs at least one of the following technologies: i) Infrared structured light triangulation positioning, which calculates skin deformation through coded gratings and CMOS sensors; ii) TOF-UWB fusion positioning, combined with IMU data to compensate for motion offset; iii) The TOF sensor measures the distance between the electrode and the skin, and the wearing position is determined by a gradient algorithm; iv) A beacon positioning system comprising at least two wireless beacons deployed on the body surface, supporting UWB, BLE, ultrasonic or infrared coding types, and achieving three-dimensional positioning through TDOA / TOF / optical flow tracing algorithms; d) Bridge-driven topology, generating alternating positive and negative pulses, supporting multi-electrode collaborative or independent working modes; e) Modular interface, supporting quick replacement of carbon fiber, silicone or gel electrode sheets or electrodes; f) Power module: battery, charging circuit, charging interface, and / or power management module and / or thermal management module and / or waterproof structure; g) The high-voltage pulse electrical stimulation system uses an independent power supply for the high-voltage generation unit: the boost circuit is isolated from the MCU power domain and / or the energy storage capacitor is connected in parallel with the discharge circuit to ensure that the voltage drops to a safe value after power failure; h) Phase cancellation circuit: The H-bridge driver chip works in conjunction with the phase detection module, with the reverse pulse voltage ratio preferably being -80%; i) A discharge scheme matching system based on anatomical feature encoding, comprising at least one of the following:

1. Anatomical structure dynamic mapping subsystem, including: - A split-type 3D scanning module, integrating an infrared structured light emitter (wavelength 940±10nm) and a TOF sensor array (8×8 pixels), configured to acquire full-body point cloud data (accuracy ±0.5mm) within 15 seconds; - Medical imaging database, storing images grouped by BMI (intervals of 2.5 kg / m²). 2 A human anatomical STL model, including safety buffer parameters for the heart and nerve plexus; - Dynamic registration processor, configured to execute: i) Achieve non-rigid registration between user point cloud and STL template by improving the ICP algorithm, with the objective function E = 0.7 * E_distance + 0.3 * E_stiffness; ii) The Lie group SE(3) was used to update the model to compensate for changes in body position and maintain the registration error <1.2mm; 2. An adaptive pulse generation subsystem, comprising: - A two-stage boost circuit boosts the 3.7V input to an adjustable 1200V output (step accuracy ±5V); - A phase-controllable H-bridge circuit that supports the generation of reverse cancellation pulses with a phase difference of 180±2°; 3. Multimodal security protection subsystem, integrating: - Three-level response mechanism: When an impedance change ΔZ / Δt > 30% / s is detected, the first-level derating is activated; when the electrode displacement > 10mm, the second-level low-frequency mode is activated; when the acceleration > 3g, the output is cut off within 10ms. - An electric field cancellation module is configured to attenuate the electric field strength in the target area to <5V / m (attenuation efficiency >99.9%) when the heart area is detected to be approaching.

4. Intelligent decision-making subsystem, configured as follows: - Based on the improved KNN algorithm, discharge schemes are retrieved from the historical database. The distance formula is: D=0.5*cos_sim(F1,F2)+0.5 / (1+|ΔBMI / 5|); - Dynamically update feature weights w_i = w_i + 0.1 * (actual efficacy - predicted efficacy) * F_i 5. Anatomical structure calibration is performed using at least one of the following methods: using DICOM images, preferably automatically extracting STL models of key anatomical structures (heart / nerve plexus) and mapping them to the body surface coordinate system, and / or using machine learning dynamic calibration with a 3D U-Net network architecture, the input layer being a 256×256×4 tensor (XYZ coordinates + 100kHz impedance value), the encoder containing 5 levels of downsampling, and the decoder employing deconvolution and skip connections, and / or user-defined anatomical structures are selected via touch screen, and the system records 3D bounding boxes.

3. The high-voltage pulse electrical stimulation system according to claim 1, characterized in that... A split-type wearable electrical stimulation device, comprising at least one of the following: a) Split-connection structure: i) The main unit and the split scanning module are connected via a magnetic interface. The magnet layout is an asymmetric N52 neodymium magnet, and the mechanical angle error is preferably <±2°. ii) The transmission power between the host and the split module is adaptively adjusted (0-10dBm) based on the Bluetooth RSSI signal strength, preferably to ensure communication stability within 3 meters; b) The main unit has a built-in auxiliary TOF sensor, and the separate module has a built-in battery; c) When the distance between the host and the split module causes the scanning field of view to overlap or the data integrity to be insufficient, the safety mode is triggered or the output is cut off. d) Data transmission module: After the split module scans and generates a full-body 3D point cloud model, it is transmitted to the host after encryption and / or forward error correction. Data compression adopts at least one of the following methods: i) Spatial coordinate difference prediction coding; ii) Hierarchical voxel encoding; iii) Statistical entropy coding; iv) Chunked streaming; e) Level 3 security response: i) Level 1: Reduced output strength when the distance between the split modules is abnormal and the impedance changes abruptly; ii) Level 2: Switch to low-frequency mode if abnormality persists; iii) Level 3: When communication is interrupted, the output is cut off and phase cancellation is activated; f) The split module integrates an infrared structured light emitter and / or a TOF sensor array; g) The host and / or the split scanning module have built-in UWB chips and / or synchronous positioning beacons and / or Bluetooth modules and / or auxiliary infrared light sources to assist in positioning the host and the split scanning module and / or assist in imaging; h) Bluetooth module (preferably Bluetooth 5.3 long-range mode) transmits 3D point cloud data; preferred anti-interference design with frequency hopping spread spectrum technology: including dynamically switching multiple channels in the 2.4GHz band to avoid Wi-Fi / microwave oven interference; And / or Time Division Multiplexing (TDM): The separate module synchronizes its clock with the host and alternates between transmitting and receiving according to a preset cycle to avoid signal collision; i) The infrared structured light emitter supports hardware autofocus and / or wavefront coding (WFC) technology.

4. The high-voltage pulse electrical stimulation system according to claim 1 includes a multimodal beacon positioning system, characterized in that... At least one of the following: a) Beacon deployment and data collection:

1. At least two wireless beacons are attached to the body surface, supporting at least one of the following encoding types: UWB, BLE, ultrasound, or infrared; 2. The electrostimulator has a built-in multimode receiver that acquires the signal characteristic parameters of each beacon in real time, including: Time Difference of Arrival (TDOA) and / or Received Signal Strength Indication (RSSI) and / or Channel State Information (CSI) phase offset. b) Adaptive positioning engine, dynamically selects the optimal signal source and uses a hybrid algorithm (TDOA / TOF / optical flow tracing); c) Beacon Coordination Management: Cross-protocol networking and dynamic topology calibration, updating channel model parameters; d) Enhanced security strategy: Activate redundant positioning when beacon fails; if the RSSI value of any beacon suddenly drops, react immediately, preferably by cutting off the output and triggering a system alarm; e) The dynamic calibration method for hazardous areas includes at least one of the following:

1. Impedance-temperature joint monitoring: When a specific area simultaneously meets the following conditions: When ΔZ > 30% (5-second sliding window) and ΔT > 1.5℃ Automatically marked as a temporary danger zone and generate phase cancellation command.

2. Establish a propagation model for the hazardous area, predict the electric field distribution through finite element analysis, and generate a three-dimensional shielding matrix: E_shield = E0 × exp(-k × d^n) (k = 0.35, n = 1.8 are the biological tissue attenuation coefficients, and d is the Euclidean distance to the boundary of the danger zone) 3. Danger Zone Priority Management Strategy: When multiple danger zones overlap, implement gradient protection in the order of cardiac area > nerve plexus > major blood vessels > muscle; f) Dynamic positioning algorithm:

1. The relative distance between the electrical stimulator and each beacon is calculated based on the signal propagation model, using the following formula: d_i=c·Δt_i+ε (c is the signal propagation speed, Δt_i is the time difference of arrival, and ε is the environmental noise correction term); 2. Solve for the coordinates (x, y, z) of the electrical stimulator using the least squares method. The objective function is: minΣ(d_i-\sqrt{(x-x_i)^2+(y-y_i)^2+(z-z_i)^2})^2 g) The beacon integrates at least one of the following fit status detection technologies:

1. Bioimpedance detection unit: Preferably, a 1kHz / 100kHz dual-frequency microcurrent is injected into the skin through the conductive contact ring at the bottom of the beacon to measure the contact impedance. If the impedance exceeds the threshold, it is determined to be a detachment.

2. Capacitive coupling sensor: detects the coupling capacitance between the beacon and the skin, and triggers an alarm when the fluctuation exceeds a preset value; 3. Vibration damping analysis: The vibration spectrum of the beacon is collected by a MEMS accelerometer. If the energy attenuation is less than the preset value when the skin comes into contact, it is determined that the beacon is not in contact.

4. NFC Near Field Verification: When the NFC signal strength (RSSI) between the beacon and the stimulator is less than a preset value, it is determined that the beacon is too far from the skin; 5. Temperature gradient detection: The temperature rise rate when in contact with skin is monitored by the built-in thermistor (such as NTC 10kΩ) of the beacon (normal >0.5℃ / s). An abnormal rate is marked as detachment. h) The electrical stimulator incorporates one of the following co-location modules:

1. UWB receiver chipset, including: - Decawave DW3000 chip, operating frequency band 3.5-6.5GHz, supports TDoA positioning mode; - Quadruple spiral antenna array, radiation efficiency >65%, half-power beamwidth of radiation pattern ≥120°; 2. Inertial navigation unit, comprising: -MPU6050 six-axis IMU, configured to acquire acceleration and angular velocity data at a sampling rate of 200Hz; - Extended Kalman filter, fusing IMU data with UWB positioning results, dynamically compensates for motion offset; 3. The security verification module performs at least one of the following verifications: Beacon ID whitelist verification: Only responds to the MAC address of pre-registered beacons; Signal integrity verification: Signal tampering is detected through CRC32 and parity check bits; Topology rationality judgment: When the distance between beacons changes by more than 10%, an anomaly alarm is triggered; i) A high-voltage pulse electrical stimulation system, characterized in that the collaborative localization method between the UWB receiver chipset and the IMU includes:

1. Dynamic weight allocation strategy: - When the UWB signal-to-noise ratio is >20dB, the positioning weights W_uwb = 0.8 and W_imu = 0.2; - When the signal-to-noise ratio of the UWB signal is ≤10dB, W_uwb=0.3, W_imu=0.7; 2. Motion compensation algorithm: Calculate the pose transformation matrix using IMU data: ΔT=∫(a+g)dt 2 +1 / 2∫ω×v dt (a is acceleration, g is gravitational compensation, ω is angular velocity, v is velocity) Applying ΔT to the original UWB coordinates, the corrected stimulator position is output: P_corrected=ΔT·P_uwb 3. Exception handling mechanism: -When the UWB positioning residual is greater than 3σ, the IMU autonomous navigation mode is activated for a duration of ≤30 seconds; -If the residual continues to be abnormal, a level 3 safety response will be triggered and a fault code will be recorded; j) The electrical stimulator has a built-in phase synchronization module to achieve clock calibration with the beacon, including:

1. High-precision timestamps based on the IEEE 802.15.4z standard (resolution ≤ 65ps); 2. Temperature-compensated crystal oscillator (TCXO), frequency stability ±1ppm; 3. Two-way time synchronization protocol, clock deviation correction is performed every 5 seconds to maintain time base error <100ns.

5. The high-voltage pulse electrical stimulation system according to claim 2, characterized in that... One of the following: a) The biofeedback module uses dual-frequency impedance detection (1kHz / 100kHz) to separate skin resistance (R0) and coupling capacitance (C); the dual-frequency impedance detection is based on the Cole-Cole model to separate skin resistance and capacitance, and the preferred frequencies are 1kHz and 100kHz. b) Dynamically adjust the high voltage output according to the formula V_base=500V+20V×(BMI-18.5). Adjust the output in three levels according to the BMI value, including low BMI mode (BMI<first threshold), standard mode, and high BMI mode (BMI>second threshold).

6. The high-voltage pulse electrical stimulation system according to claim 2, characterized in that... One of the following: a) Phased electric field cancellation technology: When the heart area is detected, the H-bridge driver chip is activated to generate a reverse pulse with a phase difference of ±180°±5°, and the electric field strength is attenuated to <5V / m; b) The electric field strength decays to <5V / m, with a decay efficiency >99.9%; c) The main electrode outputs a positive pulse (V1), and the canceling electrode outputs a reverse pulse (0.8V1). The electric field gradient decays with distance.

7. The high-voltage pulse electrical stimulation system according to claim 4, characterized in that... One of the following: a) Adaptive positioning engine signal priority strategy: When the UWB signal strength is > -60dBm, the TDOA algorithm is given priority; when it is < -80dBm, the TOF+IMU fusion positioning is switched. b) Kalman filtering fuses multi-source data, and IMU feedback angular velocity and acceleration; c) Dynamic channel modeling includes radio frequency fading parameters, sound speed temperature compensation, and optical background suppression.

8. The high-voltage pulse electrical stimulation system according to claim 1, characterized in that... One of the following: a) Side physical buttons and touchscreen work together: short press to adjust intensity, long press for preset duration to turn pages, double-click for emergency stop; b) Silicone anti-slip texture on the button surface; c) The touchscreen displays real-time parameters and safety status prompts.

9. The high-voltage pulse electrical stimulation system according to claim 1, characterized in that... At least one of the following: a) Flexible PCB integrated with graphene conductive layer, electrode array thickness < 0.15 mm, bending radius < 15 mm, and fit error < 1.2 mm; b) Contact resistance <200Ω, supporting multiple body parts such as wrist, waist and leg; c) Standby power consumption < 8μA, battery life > 84 hours, meets IP67 waterproof rating; d) Includes a touchscreen structure and / or a rotatable clamp.

10. The high-voltage pulse electrical stimulation system according to claim 1 includes a dynamic optimization method for the discharge scheme, characterized in that... Includes at least one of the following: a) User registration and preference management: i) Differentiate individual users through biometrics (fingerprint / face) or user ID to establish a personalized database; ii) Store user historical preference parameters, including commonly used stimulation intensities, treatment sites, and time period preferences; b) Record actual therapeutic data, including pain scores, resistance change rate, and user feedback; c) Reverse update the feature weights, with the optimized formula as follows: w_i = w_i + α(actual efficacy - predicted efficacy) F_i (α is the learning rate, and F_i is the eigenvector component); d) Personalized solution generation: Based on the user's registered preference parameters and real-time analytical features, generate the top 3 candidate solutions, with the following priority rules: i) Preference matching weight: 40% ii) Anatomical feature similarity weight: 40% iii) Security weight: 20%; e) When the change in feature weights exceeds the threshold, the scheme library is retrained.

Images