A hip joint traction planning system and method based on digital twinning and multi-source perception

Abstract

The application discloses a hip joint traction planning method and system based on digital twinning and multi-source sensing, which is used in cooperation with a pure manual hip joint traction frame and comprises the following steps: a control terminal acquires multi-source sensing data through a traction force sensing bandage bound to the ankle of a patient, a surface electromyography sensor, an inertial measurement unit and a wireless inclination sensor; a digital twinning three-dimensional model is constructed based on the geometric dimensions of each component of the traction frame, kinematics mapping and envelope body interference detection are performed; an optimal manual operation sequence is solved; after confirmation through digital twinning virtual preview, the handle sleeve built-in tactile vibration motor and nine-axis inertial sensor are used to guide and calculate the operation amount in real time. The application upgrades the experience-dependent manual traction operation to a data-driven intelligent traction mode, and has the flexibility of manual operation and the ability of intelligent safety guarantee and data management.

Description

Technical Field

[0001] This invention relates to the field of hip joint traction planning system technology, and in particular to a hip joint traction planning system and method based on digital twin and multi-source sensing. Background Technology

[0002] The hip joint, formed by the femoral head and acetabulum, is a ball-and-socket joint and one of the largest and most heavily loaded joints in the human body. In postoperative or post-injury rehabilitation of the hip joint, traction therapy is a crucial method for restoring joint mobility, relieving muscle spasms, and promoting articular cartilage repair. The core of traction therapy lies in applying a continuous and controllable axial tensile load to the hip joint with appropriate traction force and angle, thereby gradually widening the joint space and loosening adhesions.

[0003] Currently, hip traction devices used clinically are mainly divided into two categories. One category is electric traction beds or computer-controlled traction systems, which can automatically control traction force and traction angle. However, these devices are expensive to purchase, placing a significant financial burden on medical institutions that already have manual traction frames. The other category is purely manual traction frames, which allow for manual adjustment of the traction posture. These frames are simple in structure, low in cost, and flexible in operation, making them the most commonly used type of traction device in medical institutions at all levels.

[0004] However, purely manual traction frames have the following prominent problems in practical use. The traction force relies entirely on the operator's sense of touch for estimation. Without mechanical sensors or physiological monitoring during traction, the operator cannot know the actual traction force transmitted to the patient's limbs in real time, nor can they perceive whether the patient's muscle tissue is in a state of latent stress due to overstretching. The appropriateness of the traction amount can only be judged based on the patient's reported pain, posing a risk of tissue damage from overstretching or affecting treatment efficacy from insufficient traction. Furthermore, multi-degree-of-freedom manual adjustment lacks a predictive mechanism; spatial interference may occur between moving parts during adjustment, requiring repeated trial and error by the operator, resulting in low efficiency. In addition, manual traction frames lack data recording capabilities; the operational parameters and patient physiological responses for each traction session cannot be archived, hindering continuous optimization of rehabilitation programs and medical quality control. Summary of the Invention

[0005] To address the aforementioned technical problems in the existing technology, the present invention aims to provide a hip joint traction planning system and method based on digital twin and multi-source sensing. Through non-invasive external sensors and digital twin technology, it provides operators with functions such as quantitative traction parameter calculation, virtual pre-simulation verification, real-time closed-loop safety monitoring, and adaptive data management.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A hip traction planning method based on digital twin and multi-source sensing, used in conjunction with a purely manual hip traction frame including a base, lifting rod, pin, top seat, pivot, locking pin, connecting rod, cylinder, height adjustment mechanism, front and rear adjustment mechanism, and footrest, includes the following steps:

[0008] S1. Non-invasive multi-source sensing data initialization: The control terminal collects the patient's physical parameters and simultaneously acquires the attitude data of the inertial measurement unit (IMU) strapped to the patient's affected limb and the baseline signal of surface electromyography (sEMG) attached to the hip muscles on the affected side; at the same time, it acquires the initial tension data through the traction sensing straps attached to both sides of the foot placement frame in a non-invasive manner, and acquires the initial support tilt angle data of the cylinder through the wireless tilt sensor magnetically attached to the outer wall of the cylinder or the side plate surface of the height adjustment mechanism.

[0009] S2. Kinematic mapping and envelope interference detection based on digital twin: The control terminal constructs a digital twin three-dimensional physical model based on the geometric dimensions of each physical component of the manual hip traction frame, and maps the multi-source sensing data to the joint space kinematic matrix according to the target traction posture; three-dimensional geometric envelope models are established for the lifting rod, cylinder, height adjustment mechanism and front and rear adjustment mechanism respectively, and the spatial geometric position of each mechanical component in the expected motion trajectory is calculated by envelope intersection, and motion paths that cause mechanical collision or spatial interference are actively eliminated;

[0010] S3. Generation of the optimal standardized manual operation path sequence: Based on the set of safe and feasible paths after interference removal, the control terminal uses the combination of maximizing the smoothness of the patient's traction posture and minimizing the mechanical adjustment variation as the composite optimization objective function to solve and generate the optimal standardized manual operation sequence, which includes the target height value of the lifting rod, the target rotation angle value of the rotating shaft, the target rotation number of the first crank handle, the target rotation number of the second crank handle, and the target tilt angle adjustment value of the cylinder.

[0011] S4. Digital Twin Virtual Pre-show and Two-way Tactile Interaction Guidance: The control terminal renders on the screen and dynamically pre-shows the expected trajectory of the traction frame and the affected limb under the operation sequence in three-dimensional space; after confirmation of execution, the external tactile feedback handle sleeve fitted on the first and second rocker arms outputs vibration guidance commands through interference fit; at the same time, the nine-axis inertial sensor built into the handle sleeve collects and calculates the actual number of rotations and physical displacement of the rocker arms in real time, and outputs commands to the operator for adjusting the height of the lifting rod, unlocking and adjusting the rotating shaft, adjusting the cylinder tilt angle, rotating the first rocker arm, and rotating the second rocker arm.

[0012] S5. Dual closed-loop safety control based on force-electromyography cross-diagnosis and cumulative fatigue: During manual adjustment, the control terminal monitors the tension data and sEMG signal of the traction force sensing strap in real time, and performs cross-logic verification to diagnose and locate mechanical jamming or muscle spasm faults; at the same time, it calculates the time-domain integral work of the sEMG signal in real time. When the instantaneous abnormal threshold is triggered, a secondary muscle safety warning is triggered, and the built-in vibration motor of the corresponding handle sleeve is controlled to output the highest frequency pulse vibration to give the operator a tactile prompt to stop rotating the handle immediately.

[0013] S6. Parameter Adaptive Dynamic Correction: After the traction training is completed, the control terminal extracts the actual mechanical displacement operation amount and the patient's IMU posture response amount recorded in this training, and uses machine learning algorithms to correct the parameters of the patient's specific kinematic mapping model.

[0014] Furthermore, in step S2, the construction of the outer envelope model and interference detection specifically include:

[0015] The control terminal generates polyhedral convex hull shells based on the actual boundary dimensions of the mounting bracket, extension plate, side plate, and slider in the height adjustment mechanism, as well as the actual boundary dimensions of the toothed plate and moving block in the front and rear adjustment mechanisms. During the virtual space calculation, the intersection of the meshes of the convex hull shells of each moving component in three-dimensional Euclidean space is discretized according to the time step. If the volume intersection of the convex hull shells of any two non-directly hinged components is greater than zero, it is determined that there is collision interference at the spatial node, and the system performs anti-interference backtracking to re-plan the operation order of the first and second rockers.

[0016] Furthermore, in step S3, the solution process for the composite optimization objective function is as follows:

[0017] The control terminal establishes the objective function. ,in This is a penalty term for the abrupt change rate of the spatial angle of the affected limb in the pre-simulated trajectory. and These represent the planned total number of rotations of the first and second rocker handles required to achieve the target, respectively. The tilt angle adjustment amount of the cylinder. , , Weighting coefficients are set for the system; a heuristic search algorithm is used to find the most feasible path in the set of safe and feasible paths. The combination of variables that yields the minimum value is used to generate the optimal operation sequence.

[0018] Further, in step S4, the real-time acquisition and calculation of the actual number of rotations and physical displacement of the crank handle using the nine-axis inertial sensor built into the handle sleeve specifically includes:

[0019] When medical staff manually rotate the first and second cranks, the accelerometer inside the nine-axis inertial sensor continuously collects the downward vector of Earth's gravity in three-dimensional space.

[0020] Without an external mechanical stationary reference stator, the periodic sine wave signal formed by the projection of the gravity vector onto the sensor's own rotating coordinate system is extracted. By analyzing the complete peak period number and current phase angle of the sine wave signal in real time, the actual total number of rotations of the crank and the instantaneous angular displacement are calculated.

[0021] By combining the fixed pitch lead of the screw, the actual total number of rotations of the first crank is mapped into the actual lifting displacement of the slider; by combining the fixed transmission ratio parameters of the gear and rack, the actual total number of rotations of the second crank is mapped into the actual horizontal displacement of the moving block.

[0022] Furthermore, the mechanism for localization based on force-electromyography cross-diagnosis in step S5 is specifically as follows:

[0023] If the actual tension detected by the traction force sensing strap exceeds the physical safety threshold, and the fluctuation amplitude of the synchronously acquired sEMG signal is less than the preset muscle stress threshold, the control terminal determines that the fault source is a pure mechanical transmission jam inside the traction frame and issues a first-level mechanical fault warning.

[0024] If the actual tension detected by the traction force sensing strap is within the normal range, but the energy or amplitude of the high-frequency component of the sEMG signal surges and exceeds the muscle stress threshold, the control terminal determines that the fault source is that the patient has a latent muscle spasm or is about to suffer traction injury, triggers a secondary muscle safety warning, and controls the built-in vibration motor of the operating handle sleeve to output the highest frequency pulse vibration.

[0025] Furthermore, in step S5, the monitoring strategy based on cumulative fatigue specifically includes:

[0026] The control terminal uses an integral function. Real-time calculation of the cumulative work performed by the affected muscle group within a single traction cycle; when When the patient's personal health record shows that the extreme fatigue threshold has been reached, the system also determines that the current rehabilitation process has reached the physiological tolerance limit and issues visual, auditory and tactile alarms through the control terminal, prompting the operator to immediately stop the adjustment and remove the traction force.

[0027] Furthermore, during the execution of steps S4 and S5, a hazard verification step based on inverse kinematics inversion is also included:

[0028] The control terminal inputs the actual spatial posture angle variable of the patient's thigh, fed back by the IMU sensor attached to the patient's limb, into the digital twin model of the system in real time. The control terminal uses the inverse kinematics solver to inversely derive and calculate the theoretical values ​​of the slider displacement and the moving block displacement that theoretically need to match the posture variable. The difference is compared with the actual physical displacement calculated by the handle sleeve in step S4. If the absolute error exceeds the system tolerance threshold, it is determined that the limb has slipped relative to the foot placement frame or that the screw has abnormal slippage. An emergency braking warning is then triggered and the operator is notified to stop all adjustment operations immediately.

[0029] This invention also proposes a hip joint traction planning system based on digital twin and multi-source sensing, comprising:

[0030] Main mechanical actuator: a movable hip joint traction frame, comprising a base, lifting rod, pin, top seat, rotating shaft, locking pin, connecting rod, cylinder, height adjustment mechanism, front and rear adjustment mechanism, and foot placement bracket; the height adjustment mechanism includes a mounting bracket, screw, and slider, the first rocker arm meshes with the side teeth at the end of the screw through a side tooth fixed on the connecting shaft, driving the slider to move up and down along the screw; the front and rear adjustment mechanism has a second rocker arm, the second rocker arm meshes with the rack on the gear plate through a gear fixed on the gear shaft, driving the moving block to move horizontally along the direction of the gear plate;

[0031] External sensing and acquisition hardware group: including an IMU attitude sensor for wearing on the affected limb, an sEMG sensor for attaching to the skin surface of the affected side, traction sensing straps set on both sides of the foot placement frame, and a wireless tilt sensor set on the outer wall of the cylinder or the side plate surface of the height adjustment mechanism.

[0032] Intelligent interactive hardware group: tactile feedback handle sleeves respectively fitted onto the outside of the first and second rocker handles, containing a nine-axis inertial acquisition circuit without a stator structure and a vibration motor.

[0033] Digital twin computing main control unit: Equipped with memory and processor, and containing software modules for executing digital twin pre-simulation, interference detection and force-electromyography cross-analysis programs.

[0034] Furthermore, the external sensing and acquisition hardware group and the intelligent interaction hardware group are both configured as non-intrusive modification components that do not change the original drawing structure of the main mechanical actuator.

[0035] The traction sensing strap includes a medical-grade elastic webbing wrapped around the front of the patient's ankle joint. High-density polyester tension yarns are woven along the length of the elastic webbing to serve as the main tensile force bearing structure. Each end of the elastic webbing is connected to a miniature tension sensor on either side of the foot support frame. The miniature tension sensor is connected in series in the force transmission path between the elastic webbing and the frame, directly measuring the axial tension of the elastic webbing to obtain the actual traction force borne by the patient's ankle. The wireless tilt sensor integrates a neodymium iron boron magnet, which adheres to the cylinder (metal outer wall or side plate surface) without damage due to magnetic attraction. The tactile feedback handle sleeve is a hollow tube made of thermoplastic polyurethane elastomer or silicone rubber, and is tightly fixed to the outer ring of the first and second metal crank handles using interference static friction.

[0036] The present invention also provides a computer-readable storage medium having a set of program code instructions that are executed by a processor, which, when executed on a device, cause the device to complete the control flow of the method described thereon.

[0037] The beneficial effects of this invention are as follows:

[0038] This invention upgrades traditional experience-dependent traction operations to a data-driven intelligent traction mode through non-invasive external sensors and digital twin technology. A traction force-sensing strap, attached to the patient's ankle, directly measures the actual traction force borne by the limb. Combined with surface electromyography (EMG) signals, it monitors the patient's muscle physiological state in real time, capturing latent signs of injury before the patient subjectively perceives pain, thus achieving safe closed-loop control of the traction process. Digital twin virtual simulation and envelope interference detection pre-verify the feasibility and safety of the adjustment scheme before operation, eliminating the trial-and-error risks in multi-axis linkage adjustments. During operation, tactile vibrations from the handle guide the operator to precisely execute the optimal operation sequence in real time, and inverse kinematics inversion verifies the integrity of the transmission chain, promptly detecting mechanism abnormalities. All operational parameters and physiological response data throughout the traction process are automatically recorded and archived, supporting continuous optimization of personalized rehabilitation plans based on historical data. This allows traction therapy to move from extensive experience-based management to a precise, traceable, and quantifiable intelligent management stage. Attached Figure Description

[0039] Figure 1 This is a schematic diagram of the overall structure of the mobile hip traction frame used in this invention.

[0040] Figure 2 for Figure 1 A schematic diagram of the planar structure of a mobile hip traction frame.

[0041] Figure 3 for Figure 1 A structural schematic diagram of a mobile hip traction frame from another perspective.

[0042] Figure 4 for Figure 1 A schematic diagram of the structure combining the height adjustment mechanism and the front-to-back adjustment mechanism of the mobile hip traction frame.

[0043] Figure 5 for Figure 1 A plan view showing the combination of the height adjustment mechanism and the front-to-back adjustment mechanism of the mobile hip traction frame.

[0044] Figure 6 This is a schematic diagram of the overall architecture of the traction planning system of the present invention.

[0045] Figure 7 This is a flowchart of the traction planning method of the present invention.

[0046] The reference numerals and components involved in the accompanying drawings are shown below:

[0047] 1. Base; 2. Lifting rod; 3. Pin; 4. Top seat; 5. Rotating shaft; 6. Locking pin; 7. Cylinder; 8. Connecting rod; 9. Height adjustment mechanism; 10. Front and rear adjustment mechanism; 11. Foot placement bracket; 12. Front support rod; 13. Rear support rod; 14. Roller; 15. Side support rod; 16. Stabilizing plate; 17. Height adjustment wheel; 18. Rotating rod; 19. Mounting bracket; 20. Screw; 21. Side plate; 22. First crank handle; 23. Connecting shaft; 24. Slider; 25. Extension plate; 26. Screw hole; 27. Gear plate; 28. Moving block; 29. ​​Second crank handle; 30. Gear shaft; 31. Rack. Detailed Implementation

[0048] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0049] I. Overall Structure of the Traction Frame

[0050] The structure of the movable hip traction frame used in this embodiment is as follows: Figures 1 to 5 As shown. The traction planning system of this invention realizes quantitative planning and safety monitoring of the traction process through external sensors and intelligent terminals.

[0051] Specifically, refer to Figure 1 and Figure 2The mobile hip traction frame includes a base 1, a lifting rod 2, a pin 3, a top seat 4, a rotating shaft 5, a locking pin 6, a cylinder 7, a connecting rod 8, a height adjustment mechanism 9, a front and rear adjustment mechanism 10, and a foot placement bracket 11.

[0052] The upper end of the base 1 is connected to the lifting rod 2, and the pin 3 is connected to the lifting rod 2. The other end of the lifting rod 2 is connected to the top seat 4. The extension and retraction of the lifting rod 2 is used to adjust the overall height of the traction frame, and the selected height is locked by inserting the pin 3 into the positioning hole on the lifting rod. A height adjustment wheel 17 is installed on the side of the lifting rod 2. The operator can drive the gear and rack mechanism inside the lifting rod 2 to achieve lifting by rotating the height adjustment wheel 17.

[0053] One end of the rotating shaft 5 is rotatably inserted into the front of the top seat 4. A ring of pin holes is provided on the side of the rotating shaft 5. A locking pin 6 that mates with the pin holes is inserted into the upper end of the top seat 4. The rotating shaft 5 is rotated by the rotating rod 18, thereby driving the height adjustment mechanism 9, the front and rear adjustment mechanism 10 and the foot placement bracket 11 to rotate as a whole. After rotating to the correct position, the rotation position is locked by inserting the locking pin 6 into the corresponding pin hole.

[0054] The height adjustment mechanism 9 is hinged to the other end of the rotating shaft 5. The front-to-back adjustment mechanism 10 is vertically connected to the height adjustment mechanism 9, and the footrest 11 is connected to the other end of the front-to-back adjustment mechanism 10. A connecting rod 8 is also connected to the rotating shaft 5. The other end of the connecting rod 8 is hinged to a cylinder 7, and the other end of the cylinder 7 is hinged to the bottom surface of the height adjustment mechanism 9. The extension and retraction of the cylinder 7, through the connecting rod 8, causes a change in the relative angle between the rotating shaft 5 and the height adjustment mechanism 9, thereby changing the support angle of the footrest 11.

[0055] Furthermore, referring to Figure 2 Two front support rods 12 are connected to the front of the base 1, and two rear support rods 13 are connected to the rear of the base 1. Rollers 14 are connected to the bottom of both the front support rods 12 and the rear support rods 13, allowing the entire traction frame to move. Side support rods 15 are connected to both the left and right sides of the base 1, and stabilizing discs 16 are connected to the bottom of the side support rods 15 to provide lateral stability support during traction.

[0056] Reference Figure 4 and Figure 5The height adjustment mechanism 9 includes a mounting frame 19, a first side tooth, a second side tooth, a screw 20, a side plate 21, a first rocker arm 22, a connecting shaft 23, and a slider 24. The cylinder 7 is hinged to the bottom of the mounting frame 19. Extension plates 25 are provided at both ends of the mounting frame 19, and two screws 20 are rotatably connected between the two extension plates 25. Two screw holes 26 are provided on the slider 24, and the screw holes 26 are connected to the screws 20. The front-to-back adjustment mechanism 10 is vertically connected to the slider 24. The side plates 21 are connected to the left and right sides of the upper end of the mounting frame 19. The connecting shaft 23 rotatably passes through and is connected to the two side plates 21. Two first side teeth are fixedly sleeved on the connecting shaft 23. Second side teeth that cooperate with the first side teeth are provided at the ends of the two screws 20. The two first side teeth mesh with the two second side teeth respectively. The first rocker arm 22 is connected to both ends of the connecting shaft 23. When the operator cranks the first crank handle 22, it drives the connecting shaft 23 to rotate. Through the meshing transmission of the first side teeth and the second side teeth, it drives the screw 20 to rotate. The rotating screw 20 drives the slider 24 to move up or down along the screw axis through the screw hole 26, thereby realizing the height adjustment of the foot placement bracket 11.

[0057] Reference Figure 4 and Figure 5 The front-to-back adjustment mechanism 10 includes a toothed plate 27, a movable block 28, a second rocker handle 29, a gear, and a gear shaft 30. One end of the toothed plate 27 is vertically connected to the slider 24. The movable block 28 is slidably sleeved on the toothed plate 27. The gear shaft 30 is rotatably connected through the movable block 28. The gear is fixedly sleeved on the gear shaft 30. A rack 31 is provided on the top of the toothed plate 27. The rack 31 meshes with the gear. The second rocker handle 29 is connected to both ends of the gear shaft 30. When the operator shakes the second rocker handle 29, it drives the gear shaft 30 to rotate. Through the meshing transmission between the gear and the fixed rack 31, the movable block 28 is driven to move forward or backward along the direction of the toothed plate 27, thereby realizing the adjustment of the front-to-back distance of the footrest 11.

[0058] In summary, the traction frame in this embodiment has the following five independently adjustable mechanical degrees of freedom: the first is the overall lifting of the lifting rod 2, which is operated by the height adjustment wheel 17 and locked by the pin 3; the second is the overall rotation of the rotating shaft 5, which is operated by the rotating rod 18 and locked by the locking pin 6; the third is the support tilt angle of the cylinder 7, which is achieved by adjusting the cylinder extension and retraction; the fourth is the height of the foot placement frame 11, which is achieved by shaking the first rocker handle 22 to drive the screw 20 and the slider 24, with transmission self-locking; the fifth is the front and rear distance of the foot placement frame 11, which is achieved by shaking the second rocker handle 29 to drive the gear rack and the moving block 28, with transmission self-locking.

[0059] II. System Overall Architecture

[0060] The traction planning system architecture of the present invention is as follows: Figure 6 As shown, the whole is divided into four modules: the main mechanical execution end, the external sensing and acquisition hardware group, the intelligent interaction hardware group, and the digital twin computing main control machine.

[0061] (a) Main mechanical actuator

[0062] This refers to the aforementioned mobile hip traction frame.

[0063] (II) External sensing and data acquisition hardware group

[0064] In this embodiment, the external sensing and acquisition hardware group includes the following four types of sensor components:

[0065] First, the IMU attitude sensor uses a miniature nine-axis MEMS inertial measurement unit, measuring approximately 20mm × 15mm × 5mm and weighing no more than 5g. It is fixed to the proximal lateral thigh of the affected side of the patient (near the greater trochanter of the femur) using a medical elastic bandage, with the sensor's X-axis aligned with the longitudinal axis of the thigh. The sensor is powered by a built-in lithium battery, providing at least 8 hours of battery life. It communicates with the control terminal via Bluetooth Low Energy 5.0, with a preferred sampling rate of 100Hz.

[0066] Second, the sEMG surface electromyography sensor uses disposable Ag / AgCl surface electrode patches, which are attached to the surface of the muscle belly of the muscles surrounding the hip joint on the affected side of the patient. In this embodiment, a dual-lead configuration is preferred: the first lead is attached to the surface projection area of ​​the iliopsoas muscle (approximately 3 cm below the inguinal ligament), and the second lead is attached to the surface projection area of ​​the gluteus maximus muscle (at the midpoint of the line connecting the posterior inferior iliac spine and the ischial tuberosity), with an electrode spacing of 20 mm. After differential amplification, the signal is transmitted to the control terminal via a wireless module, with a preferred sampling rate of 1000 Hz and a bandpass filter range of 20 Hz to 500 Hz.

[0067] Third, the traction sensing bandage includes a medical-grade elastic webbing approximately 40mm wide. High-density polyester reinforcing yarns are woven along the length of the webbing as the main tensile load-bearing structure, with a total thickness not exceeding 2mm. In use, the bandage is wrapped around the front of the patient's ankle joint (at the junction of the distal tibia and the dorsum of the foot). Each end of the bandage is fastened to the side panels of the foot support bracket 11 via a miniature tension sensor. The miniature tension sensor is connected in series in the force transmission path between the bandage and the side panels, directly measuring the axial tension of the bandage. The measurement range of this miniature tension sensor is 0 to 200N, with a resolution better than 0.5N and a response time of less than 5ms. The sensor body integrates a wireless signal transmission module, transmitting data to the control terminal via Bluetooth. Because the traction force is directly measured by the miniature tension sensor connected in series in the force path, it does not depend on the contact pressure between the bandage and the skin, and is therefore unaffected by individual differences such as the patient's ankle curvature and the tightness of the bandage, ensuring accurate and reliable measurement results.

[0068] Fourth, a wireless tilt sensor is used, employing a dual-axis MEMS accelerometer-type tilt sensor with an integrated neodymium iron boron magnet, which can be directly magnetically attached to the metal structure surface of the traction frame. In this embodiment, it is preferably attached to the outer wall of the metal cylinder body of cylinder 7 to measure the tilt angle of cylinder 7 relative to the horizontal plane. The measurement range is ±90 degrees, the resolution is 0.05 degrees, and it communicates with the control terminal via Bluetooth, with a sampling rate preferably of 50Hz.

[0069] It should be noted that all four sensor components mentioned above are installed using a non-invasive method: the IMU attitude sensor is strapped to the patient's thigh with a bandage; the sEMG surface electromyography sensor is attached to the patient's skin with medical adhesive electrode patches; the traction sensing bandage is connected to the side panels of the footrest 11 via quick-release buckles and wraps around the patient's ankle; and the wireless tilt sensor is magnetically attached to the metal outer wall of the cylinder 7 (leaving no marks on the metal surface after direct removal). All external components can be quickly installed and removed, and the traction frame can still be used independently as a regular manual traction frame when the external components are not installed.

[0070] (III) Intelligent Interactive Hardware Cluster

[0071] In this embodiment, the intelligent interactive hardware group includes two haptic feedback handle sleeves, corresponding to the first joystick 22 and the second joystick 29, respectively.

[0072] Each handle sleeve is a hollow cylindrical tube, with its inner diameter matching the outer diameter of the corresponding joystick. It is securely fastened to the outer ring of the joystick using interference static friction, requiring no additional securing measures during use; it can be removed simply by pulling it out axially. The outer shell of the handle sleeve is made of thermoplastic polyurethane elastomer (TPU), with a preferred Shore A70 to A85 hardness, ensuring both grip comfort and rigid coupling between the internal sensors and the joystick. The handle sleeve integrates the following circuit modules: a nine-axis inertial sensor module for real-time acquisition of rotational motion data of the handle sleeve / joystick; a vibration motor, employing a miniature eccentric rotor vibration motor with a diameter of 6mm and a rated speed of 12000rpm, capable of generating approximately 1.2G of vibration acceleration under a 3.3V drive voltage, sufficient to transmit this acceleration to the operator's hand through the handle sleeve shell; a Bluetooth communication module for establishing a low-power Bluetooth connection with the control terminal, enabling bidirectional transmission of sensor data and vibration control commands; and a miniature lithium battery and charging circuit, charging via a USB-C interface, providing at least 20 hours of battery life on a single charge.

[0073] (iv) Digital twin computing main control unit

[0074] The control terminal that runs the traction planning software can be a tablet, laptop, or doctor's workstation. The terminal's memory pre-stores a database of three-dimensional geometric models of all moving parts of the traction frame, kinematic mapping algorithms, envelope interference detection programs, force-electromyography cross-diagnosis algorithms, cumulative fatigue calculation programs, inverse kinematics inversion verification programs, and a patient record management database. The processor executes these algorithms in real time and communicates bidirectionally with all external sensors and the handle via Bluetooth.

[0075] III. Implementation of the Traction Planning Method

[0076] The following combination Figure 7 The flowchart below provides a detailed explanation of each step in the traction planning method of this invention.

[0077] (I) Initialization of Multi-Source Sensing Data

[0078] At the start of the traction planning process, medical staff input the patient's basic physiological information into the guided interface of the control terminal, including height, weight, affected leg length (measured from the greater trochanter of the hip joint to the lateral malleolus), and the target traction angle set in the clinical orders. These parameters will serve as the basic input for subsequent kinematic mapping model calculations.

[0079] Simultaneously, the system acquires initial data from the following four sensors:

[0080] The first channel receives data from the IMU attitude sensor. The system acquires the initial three-dimensional spatial attitude angles (pitch, roll, and yaw) of the patient's thighs while in a resting state, serving as the reference zero point for subsequent angle changes. The IMU attitude sensor integrates a three-axis accelerometer, a three-axis gyroscope, and a three-axis magnetometer, outputting nine-axis raw data. The control terminal fuses this data using complementary filtering or Kalman filtering algorithms to output stable spatial attitude angles.

[0081] The second channel receives data from the sEMG surface electromyography sensor. The system acquires the patient's sEMG baseline signal in a completely relaxed, resting state for 5 to 10 seconds, and calculates the root mean square (RMS) value of the sEMG signal during this period as an individualized resting baseline threshold. This baseline threshold will be used as a reference for subsequent determination of abnormal muscle conditions.

[0082] The third channel receives data from the traction force sensing strap. The system collects the initial tension data of the strap when it is wrapped around the patient's ankle but before traction is applied, serving as the zero-point reference for subsequent changes in tension.

[0083] The fourth channel receives data from a wireless tilt sensor. The system acquires the initial value of the actual support tilt angle of cylinder 7. This sensor integrates a dual-axis MEMS accelerometer, which calculates the tilt angle by measuring the projection component of the gravity vector in the sensor coordinate system, with a measurement accuracy of ±0.1 degrees.

[0084] (II) Kinematic mapping and envelope interference detection based on digital twin

[0085] This step is one of the core algorithm stages of this invention, and its purpose is to establish a digital twin model of the traction frame and perform kinematic calculations and interference pre-detection.

[0086] The control terminal's software system pre-stores precise three-dimensional geometric models of all moving parts of the traction frame, including the overall dimensions of the base 1, lifting rod 2, and top seat 4; the spatial geometric dimensions of the mounting frame 19, extension plate 25, and side plate 21; the outlines of the slider 24, toothed plate 27, and moving block 28; the hinge positions and range of motion of the cylinder 7 and connecting rod 8; and the bearing surface dimensions of the foot placement frame 11.

[0087] Based on the patient parameters (height, leg length) and target traction angle input in the preceding steps, the system establishes the forward kinematic equation of the traction frame using the Denavit-Hartenberg (DH) parametric method. This equation incorporates the adjustment amounts for each degree of freedom—the height h of the lifting rod 2, the rotation angle of the rotating shaft 5, etc. The extension and retraction of cylinder 7 The number of rotations of the first crank 22 The number of rotations of the second crank 29 —Mapped to the position and orientation matrix of the footrest 11 in three-dimensional space The specific mapping relationship is as follows:

[0088]

[0089] Each of these Let be the homogeneous transformation matrix corresponding to the degrees of freedom. Specifically, The homogeneous transformation matrix represents the vertical translation height h of the lifting boom 2. This indicates the angle of rotation of shaft 5 around its axis. The homogeneous transformation matrix; Indicates the extension / retraction amount of cylinder 7 The homogeneous transformation matrix of the change in the tilt angle of the height adjustment mechanism 9; This indicates that the first crank 22 rotates. The homogeneous transformation matrix of the displacement of slider 24 caused by screw 20 after the turn; This indicates that the second crank 29 rotates. The homogeneous transformation matrix of the displacement of moving block 28 caused by the gear and rack transmission after the circle.

[0090] After establishing the kinematic mapping model, the system further constructs a three-dimensional geometric outer envelope model (i.e., a convex hull shell) for the key moving components. Specifically, the system constructs a convex hull shell for the mounting bracket 19 and the extension plate 25. Construct a convex hull shell for the side plate 21 and the connecting shaft 23 area. Construct a convex hull shell for slider 24 Construct a convex hull shell for toothed plate 27 and moving block 28. .

[0091] During the virtual space calculation process, the system discretizes and calculates the instantaneous position of each moving component in its trajectory according to a preset time step (e.g., 0.1 seconds). At each time step, the system calculates the volume intersection of the convex hull shells of any two non-directly hinged components in three-dimensional Euclidean space. Specifically, for any two non-directly hinged components A and B, the system calculates their intersection volume. ,like If a collision occurs at a spatial node within that time step, the system determines that the path is infeasible and executes an anti-interference backtracking algorithm to readjust the order or adjustment of the first crank 22 and the second crank 29 until a safe and feasible path without interference is found.

[0092] Taking this embodiment as an example, the engineering significance of interference detection is illustrated as follows: During the movement of slider 24, if slider 24 moves to a position close to the end of extension plate 25, and at this time the toothed plate 27 of the front and rear adjustment mechanism 10 is exactly at a certain tilt angle, a spatial collision may occur between the end of toothed plate 27 and extension plate 25. In traditional operation methods, doctors cannot predict such interference and can only find the jamming during actual adjustment and then reverse the adjustment. This invention constructs a precise convex hull shell for each moving component and detects the intersection volume of all non-directly hinged component pairs in virtual space at time steps, thus realizing the early detection and active avoidance of interference. When interference is detected, the system automatically backtracks and adjusts the operation sequence, for example, replacing the scheme of "adjusting the first crank 22 first and then the cylinder 7" with "adjusting the cylinder 7 first and then the first crank 22", thereby finding an optimal operation path without collision throughout the entire process without changing the final target posture.

[0093] (III) Generation of the optimal standardized manual operation path sequence

[0094] Based on the set of safe and feasible paths selected in the preceding steps, the system further solves for the optimal sequence of operations. The control terminal constructs the following composite optimization objective function:

[0095]

[0096] The meanings of each item are as follows: This is a penalty term for the abrupt change rate of the patient's affected limb's spatial angle in the pre-simulated trajectory, reflecting the smoothness of the patient's posture changes during the operation. The smaller the value, the gentler and more responsive the procedure is to the patient. This value is obtained by taking the first difference of the angle change at each time step of the kinematic mapping equation and taking the maximum value. and These represent the planned total number of rotations of the first rocker arm 22 and the second rocker arm 29 required to achieve the target posture, respectively, reflecting the complexity of the mechanical operation. The fewer the rotations, the simpler the operation. This represents the adjustment amount (in degrees) required for cylinder 7 to adjust from the current initial tilt angle to the target tilt angle, reflecting the range of cylinder adjustment. , , The preset weighting coefficients are used to balance the priorities among the three optimization objectives.

[0097] The three terms of the objective function J correspond to three core clinical concerns. The first term... To ensure patient comfort, because sudden changes in hip joint angle during traction therapy can lead to pain reflexes and protective muscle spasms, the system, by penalizing the rate of angle abrupt changes, tends to generate gentle manipulation sequences where each adjustment causes only small-amplitude angle changes. (Second item) To ensure operational efficiency, excessive crank rotations mean longer operation times, requiring patients to wait longer in non-target positions. (Third item) To ensure the economic efficiency of cylinder 7 adjustment, since the adjustment of cylinder 7 requires the operator to bend over or walk around to the side of the traction frame, which is not very convenient, this incentive system minimizes the adjustment range of cylinder 7 and allocates more adjustment range to the first crank handle 22 and the second crank handle 29, which are more convenient to operate.

[0098] Weighting coefficient , , The specific value can be flexibly set according to the clinical scenario. In this embodiment, the preferred value is [value]. , , Prioritizing the smoothness of patient posture changes. For vulnerable patients in the early postoperative period, this can be increased. The weighting; for patients in the middle and late stages of recovery, it can be appropriately increased. The system assigns weights to improve operational efficiency. The system provides a manual adjustment function for the weight coefficients in the control terminal's settings interface, allowing attending physicians to flexibly configure them according to the patient's specific situation.

[0099] The system employs a heuristic search algorithm, prioritizing either a genetic algorithm or a particle swarm optimization algorithm, to find the combination of variables that minimizes the objective function J within a set of safe and feasible paths. .

[0100] After the solution is obtained, the system serializes the optimal parameter combination into a standardized manual operation command sequence according to the causal logic of mechanical transmission. The output order of the operation commands follows these principles: unlock first, then adjust; large-range coarse adjustment first, then small-range fine adjustment; height direction first, then forward / backward direction. The specific operation sequence is as follows: Operation command A, pull out pin 3, rotate the hand-cranked height adjustment wheel 17, and adjust the lifting rod 2 to the target height value. Once in position, insert pin 3 into the corresponding positioning hole to lock it; operate command B, pull out locking pin 6, rotate rotating rod 18, and rotate shaft 5 to the target angle value. Once in position, insert locking pin 6 into the corresponding pin hole to lock it; operate command C to adjust the extension and retraction of cylinder 7 to the target support tilt angle value. Operation command D: Rotate the first crank 22 clockwise or counterclockwise for a total of [number missing]. The drive screw 20 moves the slider 24 to a specified height position; operation command E, rotates the second crank 29 clockwise or counterclockwise for a total of The drive gear rolls along the rack 31, causing the moving block 28 to reach the specified front or rear position.

[0101] (iv) Digital twin virtual pre-visualization and two-way haptic interaction guidance

[0102] This step includes two core functions: virtual preview and haptic interaction guidance.

[0103] During the virtual rehearsal phase, before the operator makes any manual adjustments, the control terminal dynamically plays the complete execution process of the optimal operation sequence generated by the aforementioned steps in a 3D visualization on the screen. During the rehearsal, the digital twin model on the screen synchronously displays the real-time motion of each moving component of the traction frame (lifting rod 2, mounting bracket 19, slider 24, moving block 28, footrest bracket 11, etc.), the expected posture changes of the patient's thigh after each operation step, the current parameter values ​​for each operation step (such as current height, current angle, number of rotations, etc.), and the interference safety status indicators throughout the rehearsal (green indicates safety, red indicates interference). After the medical staff confirms the expected effect of the entire operation sequence on the screen, they click the "Confirm Execution" button, and the system enters the actual operation guidance mode.

[0104] During the haptic interaction guidance phase, after confirmation of execution, the control terminal displays text descriptions of the current step on the screen in the order of operation instructions A to E, while simultaneously providing real-time guidance to the operator through vibration feedback from the haptic feedback handle sleeves. One handle sleeve is fitted onto the first crank 22 to guide its operation, and the other handle sleeve is fitted onto the second crank 29 to guide its operation.

[0105] Taking the operation guidance of the first crank 22 as an example, the system calculates the remaining deviation of the crank from the target number of revolutions in real time based on the current operation progress. The vibration motor is controlled to output different vibration signals in the following modes: when When rotating the handle, the handle sleeve remains stationary and without vibration, prompting the operator to continue rotating the handle; when During the circle, the handle sleeve outputs low-frequency intermittent vibration (preferably 2Hz, lasting 200ms with a 300ms interval) to prompt the operator to slow down when approaching the target; when When the handle sleeve outputs a short, double-tap vibration (two 100ms pulses, 150ms interval), it prompts the operator to stop upon reaching the target; when At this time, the handle sleeve outputs high-frequency continuous vibration (preferably 10Hz continuous output) to prompt the operator to reverse and return if the target is exceeded. The operation guidance of the second joystick 29 adopts the same vibration mode logic.

[0106] It should be noted that during the execution of operation commands A to C (lifting rod height adjustment, shaft rotation unlocking, cylinder tilt angle adjustment), since these three degrees of freedom are not currently equipped with external encoders, the system guides users through a visual progress bar and text prompts on the control terminal screen. The guidance method is visual and auditory interaction.

[0107] During operation guidance, the nine-axis inertial sensor module built into the handle sleeve collects the rotational motion data of the crank handle in real time and calculates the actual number of rotations and physical displacement. The specific calculation principle is as follows: When medical staff manually rotate the first crank handle 22 and the second crank handle 29, the accelerometer inside the nine-axis inertial sensor module continuously collects the downward vector of Earth's gravity in three-dimensional space. Under the premise of no external mechanical stationary reference stator, the system extracts the periodic sine wave signal formed by the projection of this gravity vector onto the sensor's own rotating coordinate system. Each complete rotation of the handle completes one full sine wave cycle, and the projection component of the gravity vector in the sensor's rotation plane completes one full sine wave cycle. Specifically, let the instantaneous angular velocity of the handle rotating about its axis be... The initial phase angle is The two components of the accelerometer in the plane perpendicular to the rotation axis then exhibit the following sinusoidal waveforms:

[0108]

[0109]

[0110] in This is the constant of gravitational acceleration (approximately 9.8 m / s²). The system calculates the total number of rotations of the crank by tracking the complete cycles of this sinusoidal signal in real time (one complete sinusoidal cycle corresponds to a 360-degree rotation of the crank, which is one revolution), and uses the current phase angle... The instantaneous value is used to accurately obtain the angle of the last incomplete loop.

[0111] After obtaining the actual number of rotations, the system performs displacement calculations based on the mechanical transmission parameters of the traction frame. Let the pitch lead of screw 20 be... (That is, for every revolution of the screw, the nut, i.e., the slider 24, moves axially) (millimeters), then the actual lifting displacement of slider 24 is:

[0112]

[0113] Let the pitch circle radius of the gear be... Then the linear displacement of rack 31, that is, the horizontal displacement of moving block 28, is:

[0114]

[0115] The above transmission parameters and These are all inherent structural constants of the traction frame, which are input once during system initialization and do not need to be measured repeatedly thereafter.

[0116] (v) Dual closed-loop safety control based on force-electromyography cross-diagnosis and cumulative fatigue

[0117] This step is the core element ensuring the safety of this invention and is integral to the entire manual adjustment process. This step comprises three parallel sub-modules: routine kinematic monitoring, force-electromyography cross-diagnosis, and cumulative fatigue monitoring.

[0118] In the conventional kinematics monitoring submodule, the system compares the real-time calculated displacement with the target displacement calculated in the previous steps, and displays the percentage of adjustment progress of each degree of freedom in real time on the control terminal screen in the form of a progress bar.

[0119] In the force-electromyography cross-diagnostic submodule, the control terminal synchronously receives data from the traction force sensing strap and the sEMG surface electromyography sensor, and performs cross-logic verification. The core idea of ​​this cross-diagnostic mechanism is that abnormal combinations of force sensor data and sEMG signals can distinguish between two fundamentally different fault sources.

[0120] During hip traction, two fundamentally different types of abnormalities may occur. The first type is mechanical abnormalities, such as abnormal frictional resistance generated by insufficient lubrication or foreign objects in the screw 20, resulting in an abnormally high actual traction force transmitted to the patient's foot even when the operator is turning the handle with normal force. The second type is physiological abnormalities, such as protective spasms occurring in the muscles around the patient's hip joint during continuous traction. In this case, even if the traction force remains within a safe range, the muscle tissue is already in a dangerously overactivated state. Traditional single-force sensor monitoring can only detect the first type of abnormality and cannot detect the second type. This invention achieves accurate differentiation between the two types of abnormalities by cross-analyzing force sensor data and sEMG signals.

[0121] The specific cross-diagnostic logic is as follows: If the actual traction force detected by the traction sensor strap is... Exceeding the system's preset physical security threshold (This threshold is set based on the patient's weight and clinical orders, preferably 5% to 10% of the patient's weight), and the fluctuation amplitude of the synchronously acquired sEMG signal is less than the preset muscle stress threshold. If the control terminal determines that the fault source is a purely mechanical transmission jam inside the traction frame (for example, an abnormal increase in the frictional resistance between the screw 20 and the screw hole 26, or a foreign object blocking the gear and rack 31), the system will issue a first-level mechanical fault warning and display "Mechanical transmission abnormality, please check each adjustment mechanism" in a red flashing pop-up window on the control terminal screen, and at the same time issue a buzzer prompt.

[0122] If the actual traction force detected by the traction sensor strap Within the normal range (i.e.) However, the high-frequency components of the sEMG signal show a surge in energy or amplitude that exceeds the muscle stress threshold. If the system detects an overstretched muscle spasm or an impending traction injury, the control terminal determines that the fault is caused by the patient experiencing latent muscle spasm or impending traction injury. The physiological basis for this determination is that when muscles are overstretched, muscle fibers generate abnormal high-frequency discharge activity before significant joint displacement occurs. This physiological stress signal appears earlier than the pain signal perceived by the patient. At this time, the system triggers a secondary muscle safety warning, displaying a red pop-up message on the control terminal screen: "Patient muscle condition is abnormal, please stop adjusting immediately." Simultaneously, the system controls the built-in vibration motor of the corresponding handle sleeve (the handle sleeve currently being rotated) to output the highest frequency pulse vibration (preferably a 10Hz, 50% duty cycle square wave drive signal) to provide the operator with a strong tactile prompt to immediately stop rotating the handle. Since the operator's hands are gripping the handle, this vibration signal is directly transmitted to the operator's hands through the skin's tactile channels, ensuring that even if the operator's attention is focused on the screen or the patient's face and they do not see the warning text on the screen immediately, they can still perceive the emergency stop tactile prompt within milliseconds.

[0123] In addition, when both abnormalities occur simultaneously (i.e., the tensile force exceeds the standard and the sEMG signal also exceeds the standard), the system will trigger both Level 1 and Level 2 warnings at the same time.

[0124] In the cumulative fatigue monitoring submodule, the control terminal performs time-series cumulative fatigue monitoring. Muscle damage can be caused not only by a single excessive stretch but also by prolonged low-intensity continuous traction. Even if the instantaneous sEMG signal and traction force data are within safe limits, if the patient's muscles remain in a state of mild stress for an extended period, micro-damage within the muscle fibers will gradually accumulate, eventually manifesting as muscle tearing or severe spasm at some point. The absolute integral of the sEMG signal is widely used in exercise physiology as an objective quantitative indicator of muscle work and fatigue level. Its physical meaning is that the amplitude of the electromyographic signal is proportional to the muscle contractile force, and integrating the absolute value of the amplitude over time yields the cumulative work done by the muscle during that time period. Specifically, the control terminal uses the following integral function to calculate the cumulative work done by the patient's affected muscle group in real time during the current traction training cycle:

[0125]

[0126] in This marks the start of the traction training. For the current moment, for The amplitude of the sEMG signal at time 1.

[0127] When the accumulated iEMG value reaches the pre-registered extreme fatigue threshold in the patient's personal health record... At a certain time (this threshold is set by the attending physician in the system based on the patient's physical condition and rehabilitation stage), even if neither the current instantaneous traction force nor the instantaneous sEMG signal triggers the abnormal threshold, the system still determines that the current rehabilitation process has reached the patient's physiological tolerance limit. At this time, the system issues a visual alarm (the screen displays "Accumulated fatigue has reached the upper limit, it is recommended to end this training" with an orange background), an auditory alarm (the terminal emits a soft but clear prompt tone), and a tactile alarm (the handle sleeve outputs a low-frequency long pulse vibration mode, such as 1Hz frequency for 1 second, which is different from the high-frequency short pulse mode of muscle spasm warning), prompting the operator to stop the current adjustment operation and gradually remove the traction force.

[0128] (vi) Hazard verification based on inverse kinematics inversion

[0129] This step serves as a second layer of safety assurance, in addition to the aforementioned force-electromyography cross-diagnosis, and is executed in parallel with steps (iv) and (v). Its core idea is to detect potential problems in the system by ensuring logical consistency between the "input" (the actual mechanical displacement measured by the handle sleeve) and the "output" (the actual changes in the patient's posture measured by the IMU).

[0130] Under normal circumstances, the displacement of slider 24 measured by the handle sleeve can be used to deduce the corresponding change in patient posture through the forward kinematics equation. This deduced value should closely match the change in posture measured by the IMU posture sensor. Specifically, the control terminal inputs the actual spatial posture angle variable of the patient's thigh, fed back by the IMU posture sensor attached to the patient's limb, into the system's digital twin model in real time. The inverse kinematics solver is then used to inversely deduce and calculate the theoretical values ​​of slider 24 displacement and moving block 28 displacement that theoretically require matching to cause this posture variable.

[0131] The system compares the theoretical displacement with the actual physical displacement calculated from the handle sleeve. If there is a significant deviation, it indicates an abnormality in some part of the system: if the actual displacement of slider 24 is greater than the displacement calculated by the IMU attitude sensor, it indicates that the machine is moving but the patient's posture has not changed accordingly, possibly because the patient's foot has slipped out of the footrest 11; if the actual displacement of slider 24 is less than the displacement calculated by the IMU attitude sensor, it indicates that the change in the patient's posture is greater than the mechanical input, possibly because the screw 20 is slipping or the transmission chain is elastically deformed; if the deviation is non-linear (the error increases rapidly with the increase of displacement), it may be due to mechanical wear or structural loosening.

[0132] The system's preset tolerance threshold is preferably 5% to 10% of the theoretical displacement. When the absolute error exceeds this threshold, the system immediately displays a red warning box on the control terminal screen saying "Transmission chain abnormality - please stop operation immediately and check the patient's foot fixation status and the connection status of each mechanism." At the same time, all handle sleeves output the preset emergency stop vibration mode (a special mode of alternating high-frequency continuous vibration and intermittent pulses, which is different from the regular warning mode), reminding the operator to immediately stop all current adjustment operations.

[0133] (vii) Adaptive Dynamic Correction of Parameters

[0134] After the active adjustment phase of this traction training ends (including the maintenance phase after the patient maintains traction in the target posture), the system automatically performs adaptive dynamic correction of parameters.

[0135] The control terminal extracted the following key data from the entire training process recording: the total number of actual rotations of the first joystick 22 recorded by the sensor built into the handle sleeve. The second rocker arm 29 actually rotated a total of 29 times. The actual lifting displacement of the slider 24 is calculated based on the fixed pitch lead of the screw 20. The actual horizontal displacement of the moving block 28 is calculated based on the fixed transmission ratio of the gear and rack 31. The IMU posture sensor recorded the actual spatial posture angle changes of the patient's thigh throughout the training process. Traction force time history curve recorded by traction force sensing strap .

[0136] The system compares and analyzes the actual operational quantities and physiological response quantities with the theoretical predictions of the kinematic mapping model in the preceding steps, and uses machine learning algorithms (preferably recursive least squares or Bayesian parameter estimation) to correct the parameters of the patient-specific kinematic mapping model. Specifically, the corrected parameters include two aspects:

[0137] The first aspect is the mechanical transmission clearance compensation coefficient. By comparing the displacement calculated from the actual number of rotations recorded by the handle sleeve with the displacement calculated from the actual attitude change recorded by the IMU attitude sensor, the transmission error coefficient caused by machining clearance and wear in the screw 20 transmission chain and the gear and rack 31 transmission chain is corrected.

[0138] The second aspect is the patient's soft tissue elasticity coefficient and damping coefficient. This is achieved through analysis of traction force. The temporal response relationship (e.g., response delay, overshoot, etc.) between the joint angle changes recorded by the IMU attitude sensor and the equivalent elastic stiffness coefficient and viscous damping coefficient of the soft tissue around the patient's hip joint were fitted and corrected.

[0139] The revised model parameters are updated and stored in the patient's personal rehabilitation record database. When the patient receives traction training again, the system will directly call the updated personalized model parameters to perform kinematic mapping calculations and generate operation sequences, thereby achieving an adaptive optimization effect of "becoming more accurate with training".

Claims

1. A hip traction planning method based on digital twin and multi-source sensing, used in conjunction with a purely manual hip traction frame comprising a base (1), a lifting rod (2), a pin (3), a top seat (4), a rotating shaft (5), a locking pin (6), a connecting rod (8), a cylinder (7), a height adjustment mechanism (9), a front and rear adjustment mechanism (10), and a footrest (11), characterized in that, Includes the following steps: S1. Non-invasive multi-source sensing data initialization: The control terminal collects the patient's physical parameters and simultaneously acquires the attitude data of the inertial measurement unit (IMU) attached to the patient's affected limb and the baseline signal of the surface electromyography (sEMG) attached to the hip muscles on the affected side; at the same time, the initial tension data is acquired through the traction sensing straps attached to both sides of the foot placement frame (11) in a non-invasive manner, and the initial support tilt angle data of the cylinder (7) is acquired through the wireless tilt sensor magnetically attached to the outer wall of the cylinder (7) or the side plate (21) of the height adjustment mechanism (9); S2. Kinematic mapping and envelope interference detection based on digital twin: The control terminal constructs a digital twin three-dimensional physical model based on the geometric dimensions of each physical component of the manual hip traction frame, and maps the multi-source sensing data to the joint space kinematic matrix according to the target traction posture; establishes three-dimensional geometric envelope models for the lifting rod (2), cylinder (7), height adjustment mechanism (9) and front and rear adjustment mechanism (10), performs envelope intersection calculation on the spatial geometric position of each mechanical component in the expected motion trajectory, and actively eliminates motion paths that cause mechanical collisions or spatial interference; S3. Generation of the optimal standardized manual operation path sequence: Based on the set of safe and feasible paths after interference removal, the control terminal uses the combination of maximizing the smoothness of the patient's traction posture and minimizing the mechanical adjustment variation as the composite optimization objective function to solve and generate the optimal standardized manual operation sequence, which includes the target height value of the lifting rod (2), the target rotation angle value of the rotating shaft (5), the target rotation number of the first rocker (22), the target rotation number of the second rocker (29), and the target tilt angle adjustment value of the cylinder (7). S4. Digital twin virtual pre-play and two-way tactile interaction guidance: The control terminal renders on the screen and dynamically pre-plays the expected trajectory of the traction frame and the affected limb under the operation sequence in three-dimensional space; after confirmation of execution, the external tactile feedback handle sleeve fitted on the first rocker (22) and the second rocker (29) through interference fit outputs vibration guidance command, and at the same time uses the nine-axis inertial sensor built into the handle sleeve to collect and calculate the actual number of rotations and physical displacement of the rocker in real time, and outputs the lifting rod (2) height adjustment command, the rotating shaft (5) rotation unlock and adjustment command, the cylinder (7) tilt angle adjustment command, the first rocker (22) rotation command and the second rocker (29) rotation command to the operator; S5. Dual closed-loop safety control based on force-electromyography cross-diagnosis and cumulative fatigue: During manual adjustment, the control terminal monitors the tension data and sEMG signal of the traction force sensing strap in real time, and performs cross-logic verification to diagnose and locate mechanical jamming or muscle spasm faults; at the same time, it calculates the time-domain integral work of the sEMG signal in real time. When the instantaneous abnormal threshold is triggered, a secondary muscle safety warning is triggered, and the built-in vibration motor of the corresponding handle sleeve is controlled to output the highest frequency pulse vibration to give the operator a tactile prompt to stop rotating the handle immediately. S6. Parameter Adaptive Dynamic Correction: After the traction training is completed, the control terminal extracts the actual mechanical displacement operation amount and the patient's IMU posture response amount recorded in this training, and uses machine learning algorithms to correct the parameters of the patient's specific kinematic mapping model.

2. The hip joint traction planning method based on digital twin and multi-source sensing according to claim 1, characterized in that, In step S2, the construction of the outer envelope model and interference detection specifically include: Based on the actual boundary dimensions of the mounting bracket (19), extension plate (25), side plate (21), and slider (24) in the height adjustment mechanism (9), and the actual boundary dimensions of the toothed plate (27) and moving block (28) in the front and rear adjustment mechanism (10), the control terminal generates a polyhedral convex hull shell respectively. During the virtual space calculation process, the grid intersection of the convex hull shell of each moving part in the three-dimensional Euclidean space is calculated according to the time step. If the volume intersection of the convex hull shells of any two non-directly hinged parts is greater than zero, it is determined that there is collision interference at the space node. The system performs anti-interference backtracking and replans the operation order of the first rocker (22) and the second rocker (29).

3. The hip joint traction planning method based on digital twin and multi-source sensing according to claim 1, characterized in that, In step S3, the solution process for the composite optimization objective function is as follows: The control terminal establishes the objective function. ,in This is a penalty term for the abrupt change rate of the spatial angle of the affected limb in the pre-simulated trajectory. and These represent the planned total number of rotations of the first rocker (22) and the second rocker (29) required to achieve the target, respectively. The tilt angle adjustment amount of the cylinder (7) , , Weighting coefficients are set for the system; a heuristic search algorithm is used to find the most feasible path in the set of safe and feasible paths. The combination of variables that yields the minimum value is used to generate the optimal operation sequence.

4. The hip joint traction planning method based on digital twin and multi-source sensing according to claim 1, characterized in that, In step S4, the step of using the nine-axis inertial sensor built into the handle sleeve to collect and calculate the actual number of rotations and physical displacement of the crank handle in real time specifically includes: When medical staff manually rotate the first crank (22) and the second crank (29), the accelerometer inside the nine-axis inertial sensor continuously collects the downward vector of Earth's gravity in three-dimensional space; Without an external mechanical stationary reference stator, the periodic sine wave signal formed by the projection of the gravity vector onto the sensor's own rotating coordinate system is extracted. By analyzing the complete peak period number and current phase angle of the sine wave signal in real time, the actual total number of rotations of the crank and the instantaneous angular displacement are calculated. By combining the fixed pitch lead of the screw (20), the actual total number of rotations of the first rocker (22) is mapped into the actual lifting displacement of the slider (24); by combining the fixed transmission ratio parameters of the gear and rack (31), the actual total number of rotations of the second rocker (29) is mapped into the actual horizontal displacement of the moving block (28).

5. The hip joint traction planning method based on digital twin and multi-source sensing according to claim 1, characterized in that, The mechanism for cross-diagnostic localization based on force-electromyography in step S5 is as follows: If the actual tension detected by the traction force sensing strap exceeds the physical safety threshold, and the fluctuation amplitude of the synchronously acquired sEMG signal is less than the preset muscle stress threshold, the control terminal determines that the fault source is a pure mechanical transmission jam inside the traction frame and issues a first-level mechanical fault warning. If the actual tension detected by the traction force sensing strap is within the normal range, but the energy or amplitude of the high-frequency component of the sEMG signal surges and exceeds the muscle stress threshold, the control terminal determines that the fault source is that the patient has a latent muscle spasm or is about to suffer traction injury, triggers a secondary muscle safety warning, and controls the built-in vibration motor of the operating handle sleeve to output the highest frequency pulse vibration.

6. The hip joint traction planning method based on digital twin and multi-source sensing according to claim 1, characterized in that, In step S5, the monitoring strategy based on cumulative fatigue specifically includes: The control terminal uses an integral function. Real-time calculation of the cumulative work performed by the affected muscle group within a single traction cycle; when When the patient's personal health record shows that the extreme fatigue threshold has been reached, the system also determines that the current rehabilitation process has reached the physiological tolerance limit and issues visual, auditory and tactile alarms through the control terminal, prompting the operator to immediately stop the adjustment and remove the traction force.

7. The hip joint traction planning method based on digital twin and multi-source sensing according to claim 1, characterized in that, During the execution of steps S4 and S5, a hazard verification step based on inverse kinematics inversion is also included: The control terminal inputs the actual spatial posture angle variable of the patient's thigh, fed back by the IMU sensor attached to the patient's limb, into the digital twin model of the system in real time; the control terminal uses the inverse kinematics solver to inversely deduce and calculate the theoretical displacement value of the slider (24) and the theoretical displacement value of the moving block (28) that theoretically need to match the posture variable; compares the difference with the actual physical displacement calculated by the handle sleeve in step S4. If the absolute error exceeds the system tolerance threshold, it is determined that the limb has slipped relative to the foot placement frame (11) or the screw (20) has abnormal slippage. Then, an emergency braking warning is triggered and the operator is notified to stop all adjustment operations immediately.

8. A hip joint traction planning system based on digital twin and multi-source sensing, used to implement the method according to any one of claims 1 to 7, characterized in that, include: Main mechanical actuator: a movable hip joint traction frame, including a base (1), a lifting rod (2), a pin (3), a top seat (4), a rotating shaft (5), a locking pin (6), a connecting rod (8), a cylinder (7), a height adjustment mechanism (9), a front and rear adjustment mechanism (10), and a foot placement frame (11); the height adjustment mechanism (9) includes a mounting frame (19), a screw (20), and a slider (24). The first rocker (22) meshes with the side teeth at the end of the screw (20) through the side teeth fixed on the connecting shaft (23), driving the slider (24) to move up and down along the screw (20). The front and rear adjustment mechanism (10) has a second rocker (29), which meshes with the rack (31) on the gear plate (27) through the gear fixed on the gear shaft (30), driving the moving block (28) to move horizontally along the direction of the gear plate (27). External sensing and acquisition hardware group: including an IMU posture sensor for wearing on the affected limb, an sEMG sensor for attaching to the skin surface of the affected side, traction sensing straps set on both sides of the foot placement frame (11), and a wireless tilt sensor set on the outer wall of the cylinder (7) or the side plate (21) surface of the height adjustment mechanism (9). Intelligent interactive hardware group: tactile feedback handle sleeves respectively fitted onto the outside of the first rocker (22) and the second rocker (29), containing a nine-axis inertial acquisition circuit without a stator structure and a vibration motor; Digital twin computing main control unit: Equipped with memory and processor, and containing software modules for executing digital twin pre-simulation, interference detection and force-electromyography cross-analysis programs.

9. The hip joint traction planning system based on digital twin and multi-source sensing according to claim 8, characterized in that, The external sensing and acquisition hardware group and the intelligent interaction hardware group are both non-intrusive modification components that do not change the original drawing structure of the main mechanical actuator. The traction sensing strap includes a medical-grade elastic webbing wrapped around the front of the patient's ankle joint. The elastic webbing is woven with high-density polyester tension yarns along its length as the main tensile load-bearing structure. The two ends of the elastic webbing are connected to the two side baffles of the foot placement frame (11) through a miniature tension sensor. The miniature tension sensor is connected in series in the force transmission path between the elastic webbing and the baffle to directly measure the axial tension of the elastic webbing to obtain the actual traction force borne by the patient's ankle. The wireless tilt sensor integrates a neodymium iron boron strong magnet and attaches to the metal outer wall of the cylinder (7) or the surface of the side plate (21) without damage by magnetic attraction. The tactile feedback handle sleeve is hollow tubular and made of thermoplastic polyurethane elastomer or silicone rubber. It is tightly fixed to the outer ring of the first rocker handle (22) and the second rocker handle (29) made of metal material by interference static friction.

10. A computer-readable storage medium having stored thereon a set of program code instructions executable by a processor, characterized in that, When the program code instructions are executed on the device, they cause the device to complete the method control flow as described in any one of claims 1 to 7.

Images