Self-centering knee joint testing and training system and method

Abstract

The invention provides a self-centering knee joint measuring and training system and method. The self-centering knee joint measuring and training system comprises a rotating shaft aligning mechanism, a torque output mechanism, a torque measuring mechanism, a bending and stretching mechanism and a data processing module which are arranged on a seat. The rotating shaft aligning mechanism comprises a linear guide rail; the torque output mechanism comprises a motor and a harmonic reducer; the torque measuring mechanism comprises a potentiometer and a static torque sensor which are arranged in the motor; the flexion and extension mechanism comprises a guide rail support, a linear bearing, an optical axis in sliding connection with the linear bearing and a shank bandage arranged on the optical axis and used for fixing the shank of a user. The rotating center of the device is dynamically adjusted in a passive mode, so that the rotating shaft of the device is automatically aligned with the rotating shaft of the knee joint, and measuring errors caused by deviation of the rotating shafts are eliminated; meanwhile, a testing method and two training methods are provided, testing and training integration is achieved, and the requirements of various users can be met.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent medical rehabilitation devices, and particularly relates to a self-aligning knee joint measurement and training system and method. Background Art

[0002] The knee joint is an important part of the human body, playing the role of bearing body weight and transmitting loads. There are common knee joint movement disorders caused by traumatic diseases such as meniscus injury, ligament strain or rupture, and tendonitis. Usually, surgical treatment is combined with postoperative rehabilitation training to promote the recovery of knee joint muscle strength. The human knee joint allows the femur and tibia to have a relatively large range of motion in two degrees of freedom, namely flexion and extension and axial rotation, and can be approximately regarded as a hinge structure with one degree of rotational freedom in anatomy.

[0003] In recent years, with the development of sports medicine and rehabilitation technology, knee joint measurement and training systems have been widely used in the fields of medical rehabilitation and sports training. However, there are still many limitations in the design and function of existing knee joint measurement and training devices, which are mainly reflected in the following aspects: Most existing knee joint measurement and training devices adopt a fixed rotation axis design and cannot be dynamically adjusted according to the actual rotation axis of the user's knee joint. Since the relative movement of the femur and tibia during knee joint flexion and extension causes the rotation axis to change continuously, this fixed design will lead to a mismatch between the device and the actual movement of the knee joint, resulting in measurement errors and affecting the training and rehabilitation effects. The accurate measurement of knee joint torque is crucial for evaluating knee joint function, formulating rehabilitation plans, and optimizing training programs. However, due to problems such as misalignment of the rotation axis and unreasonable sensor arrangement in existing devices, it is difficult to achieve high-precision torque measurement, which limits its application in clinical and scientific research. Summary of the Invention

[0004] The purpose of the present invention is to provide a self-aligning knee joint measurement and training system and method to solve the above-mentioned deficiencies of the prior art.

[0005] To achieve the above purpose, the present invention adopts the following technical solutions:

[0006] A self-aligning knee joint measurement and training system, comprising a seat, a thigh strap arranged on the seat for fixing the user's thigh, a rotation axis alignment mechanism, a torque output mechanism, a torque measurement mechanism, a flexion and extension mechanism and a data processing module arranged on the seat; the rotation axis alignment mechanism comprises linear guide rails arranged in parallel on the left and right sides of the seat; the torque output mechanism comprises a motor slidably connected to the linear guide rail and a harmonic reducer coaxially connected to the motor shaft of the motor, the motor shaft of the motor is perpendicular to the linear guide rail, and the motor dynamically adjusts the output torque through a PID controller; the torque measurement mechanism comprises a potentiometer built in the motor for detecting the flexion and extension angle of the user's knee joint and a static torque sensor coaxially connected to the harmonic reducer for measuring the torque of the user's knee joint; the flexion and extension mechanism comprises a guide rail bracket connected to the static torque sensor, a linear bearing arranged on the guide rail bracket, an optical axis with one end arranged in the linear bearing and capable of slidingly connecting with the linear bearing, and a calf strap arranged at the other end of the optical axis for fixing the user's calf; the data processing module adopts an embedded processor, can control the output torque of the torque output mechanism based on a biomechanical model or an inverse dynamics model combined with a fuzzy PID control algorithm, and can process and analyze the data detected by the torque measurement mechanism based on a machine learning algorithm.

[0007] Furthermore, it further comprises a user interaction module and a power supply module; the user interaction module is connected to the external interface of the data processing module, and comprises an industrial control screen and a speaker for information transmission and user interaction; the power supply module is used for power supply of the system and provides circuit protection; the data processing module presets corresponding execution programs according to the measurement or training requirements of the user, and is configured with a biomechanical model, an inverse dynamics model, a muscle-skeleton model, a fatigue monitoring algorithm and a reinforcement learning algorithm.

[0008] Furthermore, the linear guide rail comprises a circular rail fixedly connected to the outer side surface of the seat and a slider sleeved on the circular rail; the motor is slidably connected to the slider through a frame, and a Hall sensor is arranged on the frame as an emergency stop switch of the motor, and the Hall sensor can send a motor stop command to the PID controller when the surrounding magnetic flux density reaches a preset threshold.

[0009] Furthermore, a permanent magnet is arranged on the guide rail bracket and can cooperate with the Hall sensor for emergency stop control of the motor.

[0010] A self-aligning knee joint measurement and training method is realized based on the above self-aligning knee joint measurement and training system, and the measurement and training method comprises an isokinetic motion knee joint torque measurement method, which specifically comprises the following steps:

[0011] S1. The user is seated on the seat and fixes the thighs to the seat through thigh straps, and fixes the calves to the optical axis through calf straps.

[0012] S2. The user selects a corresponding preset execution program for isokinetic knee joint torque measurement through the user interaction module.

[0013] S3. Start the motor, drive the optical axis to perform a pendulum motion at a preset angular velocity through a harmonic reducer, drive the user's calves to flex and extend, and drive the optical axis to slide and extend on the linear bearing. At the same time, the torque output measurement module slides on the circular track under the pulling action of the calf extension action, so that the rotation axis of the torque output measurement module is always aligned with the rotation axis of the user's knee joint.

[0014] S4. The knee joint flexion and extension angle of the user is detected in real time through a potentiometer and transmitted to the data processing module, and the lower limb parameters of the user are collected in real time through the data processing module.

[0015] The lower limb parameters include the mass, length, moment of inertia, and centroid position of the user's thighs, calves, and feet.

[0016] S5. The data processing module controls the output torque of the torque output module based on the inverse dynamics model combined with the fuzzy PID control algorithm.

[0017] The inverse dynamics model can calculate the torque required by the knee joint based on the following formula (1) according to the motion state of the user's knee joint:

[0018]

[0019] In formula (1): τ represents the torque required by the knee joint; M(q) is the mass matrix, representing the inertial characteristics of the system; represents the Coriolis force and centrifugal force matrix; G(q) is the gravity matrix; q represents the knee joint flexion and extension angle; represents the knee joint angular velocity; represents the knee joint angular acceleration; the and are obtained through the first-order and second-order derivatives of q respectively;

[0020] S6. The torque data of the user's knee joint is detected in real time through a static torque sensor and transmitted to the data processing module.

[0021] S7. The data processing module receives the torque data of the user's knee joint, processes and analyzes it, generates a torque measurement report, and at the same time performs data display and interaction through the user interaction module.

[0022] Furthermore, the measurement and training method also includes a rehabilitation training method for knee joint injury patients, specifically including the following steps:

[0023] S1. The user is seated on the seat and fixes the thighs to the seat through the thigh straps and fixes the calves to the optical axis through the calf straps.

[0024] S2. The user selects a corresponding preset execution program for the rehabilitation training of knee joint injury patients through the user interaction module.

[0025] S3. Start the motor, drive the optical axis to perform a pendulum motion at a preset angular velocity through the harmonic reducer, drive the user's calf to flex and extend and drive the optical axis to slide and extend on the linear bearing. At the same time, the torque output measurement module slides on the circular track under the pulling action of the calf extension action, so that the rotation axis of the torque output measurement module is always aligned with the rotation axis of the user's knee joint.

[0026] S4. Real-time detect the flexion and extension angle of the user's knee joint through the potentiometer and transmit it to the data processing module, and the data processing module collects the lower limb parameters of the user in real time.

[0027] S5. Control the output torque of the torque output module through the data processing module based on the biomechanical model combined with the fuzzy PID control algorithm.

[0028] The said biomechanical model can calculate the torque required by the knee joint based on the kinematic and mechanical characteristics of the user's knee joint according to the following formula (2):

[0029] τ = τ inertial +τ coriolis +τ gravity (2);

[0030] Where:

[0031]

[0032] τ gravity = mglsinθ (5);

[0033] In formulas (2)-(5): τ represents the torque required by the knee joint; τ inertoal represents the inertial force caused by the angular acceleration of the knee joint; τ coriolis represents the Coriolis force and centrifugal force caused by the knee joint angle; τ gravity represents the torque caused by gravity; θ represents the flexion and extension angle of the knee joint; represents the angular velocity of the knee joint; represents the angular acceleration; the said and are obtained through the first and second derivatives of θ respectively; I represents the moment of inertia; represents the Coriolis force and centrifugal force matrix; l represents the distance from the center of mass to the rotation axis; m represents the mass; g represents the acceleration due to gravity;

[0034] S6. Real-time collect the motion state data of the user's knee joint through the data processing module, and optimize the output torque of the torque output module based on the reinforcement learning algorithm according to the reward mechanism;

[0035] The reward mechanism is as follows: Positive rewards appear when the user completes the training goal, and negative rewards appear when the user is uncomfortable or the training effect is poor; The reward mechanism is constructed based on the reinforcement learning algorithm, and the Actor-Criti method is used to combine the policy function and the value function to evaluate the user's current motion state. The policy function and the value function are respectively represented by the following formulas (6) and (7):

[0036]

[0037] V(s)←V(s)+α[r+γV(s′)-V(s)] (7);

[0038] Where:

[0039]

[0040] In formulas (6)-(10): θ represents the parameter of the policy function; J(θ) represents the objective function of the policy; α represents the learning rate; s represents the current state; s′ represents the next state; a represents the action taken in state s; a′ represents the action taken in state s′; V(s) represents the value of state s; V(s′) represents the value of state s′; Q(s,a) represents the value of taking action a in state s; Q(s′,a′) represents the value of taking action a′ in state s′; represents the expectation; r represents the current reward; γ represents the discount factor, which is used to balance the current reward and the future reward, and its value range is 0≤γ≤10≤γ≤1; represents the gradient of the policy function; r+γV(s′)-V(s) represents the advantage function, which represents the quality of the current action; represents the objective function after updating with the policy gradient;

[0041] S7. Real-time detect the torque data of the user's knee joint through the static torque sensor and transmit it to the data processing module;

[0042] S8. Receive the torque data of the user's knee joint through the data processing module, process and analyze it, generate a rehabilitation strategy, and at the same time perform data display and interaction through the user interaction module.

[0043] Furthermore, the measurement and training method also includes a muscle training method for the legs, which specifically includes the following steps:

[0044] S1. The user is located on the seat, fixes the thigh to the seat through the thigh strap, and fixes the calf to the optical axis through the calf strap;

[0045] S2. The user selects the preset execution program corresponding to the leg muscle training through the user interaction module;

[0046] S3. Start the motor, drive the optical axis through the harmonic reducer to perform a pendulum motion at a preset angular velocity, drive the user's calf to flex and extend, and drive the optical axis to slide and extend on the linear bearing. At the same time, the torque output measurement module slides on the circular track under the pulling action of the calf extension action, so that the rotation axis of the torque output measurement module is always aligned with the rotation axis of the user's knee joint;

[0047] S4. Real-time detect the flexion and extension angle of the user's knee joint through the potentiometer and transmit it to the data processing module, and collect the user's lower limb parameters in real time through the data processing module;

[0048] S5. Control the output torque of the torque output module through the data processing module based on the biomechanical model combined with the fuzzy PID control algorithm;

[0049] S6. Real-time detect the torque data of the user's knee joint through the static torque sensor and transmit it to the data processing module. The data processing module analyzes the torque data based on the muscle-skeleton model combined with the fatigue monitoring algorithm and judges the fatigue degree of the user's muscles;

[0050] The data processing module based on the muscle-skeleton model combined with the fatigue monitoring algorithm judges the motion state by analyzing the change trend of the signal characteristics through the torque, angle, and angular velocity data of the user's knee joint. The signal characteristics include the maximum torque of the knee joint, the average torque, the angle change range, and the angular velocity change value;

[0051] S7. Optimize the output torque of the torque output module according to the fatigue degree of the user's muscles, and gradually reduce the torque output after the training reaches a certain duration;

[0052] S8. Real-time display relevant data through the user interaction module during the whole training process, and generate training suggestions after the training ends.

[0053] Further, the fuzzy PID control algorithm controls the output torque of the torque output module and is represented by the following formula (11):

[0054]

[0055] In formula (11): u(t) represents the control output; e(t) represents the error between the target torque and the actual torque; d represents the differential; t represents the time; K p 、K i 、K d 、are the proportional, integral, and differential coefficients respectively.

[0056] As can be seen from the above technical solutions, the present invention automatically aligns the rotation axis of the device with the rotation axis of the knee joint by passively adjusting the rotation center of the dynamic adjustment device, eliminating the measurement error caused by the deviation of the rotation axis. At the same time, a testing method and two training methods are proposed. In the testing method, the self-centering function is established by constructing an inverse dynamics model, and the fuzzy PID control algorithm is used to significantly improve the accuracy of torque measurement. In the two training methods, through the biomechanical model combined with the fuzzy PID control algorithm, reinforcement learning algorithm, and fatigue monitoring algorithm, the dynamic adjustment of the training plan is realized during rehabilitation training to improve the rehabilitation effect, and the dynamic adjustment of the training intensity and real-time detection of muscle fatigue degree are realized during muscle training, providing scientific training suggestions. The present invention realizes the integration of measurement and training and can meet the needs of various users. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 is a schematic diagram of the overall structure of the self-centering knee joint measurement and training system of the present invention;

[0058] Figure 2 is a schematic diagram of a partial structure of the self-centering knee joint measurement and training system of the present invention;

[0059] Figure 3 is a schematic diagram of a partial structure of the self-centering knee joint measurement and training system of the present invention;

[0060] Figure 4 is a schematic diagram of a partial structure of the self-centering knee joint measurement and training system of the present invention;

[0061] Figure 5 is a schematic diagram of a partial structure of the self-centering knee joint measurement and training system of the present invention;

[0062] Figure 6 is a schematic diagram of the mechanical principle of the self-centering knee joint measurement and training system of the present invention;

[0063] Figure 7 is a schematic diagram of the mechanical principle of the self-centering knee joint measurement and training system of the present invention;

[0064] Figure 8 is a schematic diagram of the step flow of the self-centering knee joint measurement and training method of the present invention;

[0065] Figure 9 is K p fuzzy rule table;

[0066] Figure 10 is K i fuzzy rule table;

[0067] Figure 11 is K d fuzzy rule table;

[0068] In the figure: 1. Linear guide rail; 2. Motor; 3. Harmonic reducer; 4. Static torque sensor; 5. Guide rail bracket; 6. Linear bearing; 7. Optical axis; 8. Calf strap; 9. User interaction module; 10. Circular rail; 11. Slide block; 12. Frame; 13. Hall sensor; 14. Permanent magnet. Specific implementation mode

[0069] A preferred implementation mode of the present invention will be described in detail below with reference to the accompanying drawings.

[0070] As Figures 1-3 shown, the self-aligning knee joint measurement and training system includes a seat, a thigh strap provided on the seat for fixing the user's thigh, a rotation axis alignment mechanism, a torque output mechanism, a torque measurement mechanism, a flexion and extension mechanism, and a data processing module provided on the seat; the rotation axis alignment mechanism includes linear guide rails 1 arranged in parallel on the left and right sides of the seat; the torque output mechanism includes a motor 2 slidably connected to the linear guide rail 1 and a harmonic reducer 3 coaxially connected to the motor shaft of the motor 2, the motor shaft of the motor 2 is perpendicular to the linear guide rail 1, and the motor 2 dynamically adjusts the output torque through a PID controller; the torque measurement mechanism includes a potentiometer built in the motor 2 for detecting the flexion and extension angle of the user's knee joint and a static torque sensor 4 coaxially connected to the harmonic reducer 3 for measuring the torque of the user's knee joint; the flexion and extension mechanism includes a guide rail bracket 5 connected to the static torque sensor 4, two linear bearings 6 provided on the guide rail bracket 5, two optical axes 7 with one ends respectively arranged in the linear bearings 6 and capable of slidingly connecting with the linear bearings 6, and a calf strap 8 provided at the other ends of the two optical axes 7 for fixing the user's calf.

[0071] The data processing module in this preferred embodiment uses an embedded processor, which can control the output torque of the torque output mechanism based on a biomechanical model or an inverse dynamics model combined with a fuzzy PID control algorithm, and can process and analyze the data detected by the torque measurement mechanism based on a machine learning algorithm.

[0072] The biomechanical model in this preferred embodiment is a model that describes the movement and mechanical characteristics of the human body under mechanical action through mathematical equations; the inverse dynamics model is a mathematical model that derives the required force or torque of the system from the known motion state.

[0073] The self - centering knee joint measurement and training system described in this preferred embodiment further includes a user interaction module and a power supply module; the user interaction module is connected to the external interface of the data processing module, and includes an industrial control screen and a speaker, which are used for information transmission and user interaction; the power supply module is used for system power supply and provides circuit protection; the data processing module presets corresponding execution programs according to the measurement or training needs of the user, and is configured with a biomechanical model, an inverse dynamics model, a muscle - bone model, a fatigue monitoring algorithm, and a reinforcement learning algorithm.

[0074] As Figure 4 and 5 shown, the linear guide rail 1 includes two circular rails 10 fixedly connected to the outer side surface of the seat and a slider 11 sleeved on the two circular rails; specifically, the motor 2 is slidably connected to the slider 11 through a frame 12, and a Hall sensor 13 is provided on the frame 12 as an emergency stop switch for the motor 2, which can send a motor 2 stop command to the PID controller when the magnetic flux density around the sensor reaches a preset threshold; correspondingly, a permanent magnet 14 is provided on the guide rail bracket 4, which can cooperate with the Hall sensor for the emergency stop control of the motor 2.

[0075] The Hall sensor 13 described in this preferred embodiment is a magnetic sensor based on the Hall effect, which can use the change of the magnetic field to control the on - off state of the switch. When a magnetic object such as the permanent magnet described in this preferred embodiment approaches the Hall sensor, the magnetic flux density around the Hall sensor 13 will change, thereby causing the generation of a Hall voltage, and this voltage signal can be used to control the on - off state of the switch.

[0076] In specific use, the user can select a corresponding preset execution program through the user interaction module 9 according to the measurement or training needs; as Figure 6 and 7 shown, after the motor 2 is started, the harmonic reducer 3 drives the optical axis 7 to perform a pendulum motion at a preset angular velocity, driving the user's calf to flex and extend and driving the optical axis 7 to slide and extend on the linear bearing 6. At the same time, the torque output measurement module slides on the circular rail 10 under the pulling action of the calf elongation action, so that the rotation axis of the torque output measurement module is always aligned with the knee joint rotation axis.

[0077] As Figure 8 shown, the self - centering knee joint measurement and training method can perform isokinetic knee joint torque measurement, rehabilitation training for knee - injured users, and leg muscle training based on the above - mentioned self - centering knee joint measurement and training system.

[0078] Example 1. Isokinetic knee joint torque measurement

[0079] Specifically, it includes the following steps:

[0080] S1. The user is seated on the seat and fixes the thighs to the seat through thigh straps, and fixes the calves to the optical axis through calf straps.

[0081] S2. The user selects the corresponding preset execution program for isokinetic knee joint torque measurement through the user interaction module.

[0082] S3. Start the motor, drive the optical axis to perform a pendulum motion at a preset angular velocity through a harmonic reducer, drive the user's calves to flex and extend, and drive the optical axis to slide and extend on the linear bearing. At the same time, the torque output measurement module slides on the circular rail under the pulling action of the calf extension action, so that the rotation axis of the torque output measurement module is always aligned with the rotation axis of the user's knee joint.

[0083] S4. Real-time detect the flexion and extension angle of the user's knee joint through a potentiometer and transmit it to the data processing module, and the data processing module real-time collects the user's lower limb parameters.

[0084] The lower limb parameters include the mass, length, moment of inertia, and centroid position of the user's thighs, calves, and feet.

[0085] S5. The data processing module controls the output torque of the torque output module based on the inverse dynamics model combined with the fuzzy PID control algorithm.

[0086] In this step, the inverse dynamics model simplifies the knee joint into a single-degree-of-freedom hinge structure, where the femur and tibia are connected by the knee joint, and the lower limb is simplified into a rigid body link system, including the thigh, calf, and foot; by inputting the knee joint angle, knee joint angular velocity, and knee joint angular acceleration and performing kinematic analysis, the knee joint torque is obtained. The inverse dynamics model can calculate the required torque of the knee joint based on the following formula (1) according to the motion state of the user's knee joint:

[0087]

[0088] In formula (1): τ represents the required torque of the knee joint; M(q) is the mass matrix, representing the inertial characteristics of the system; represents the Coriolis force and centrifugal force matrix; G(q) is the gravity matrix; q represents the flexion and extension angle of the knee joint; represents the knee joint angular velocity; represents the knee joint angular acceleration; the and are obtained through the first and second derivatives of q respectively.

[0089] S6. Real-time detect the torque data of the user's knee joint through a static torque sensor and transmit it to the data processing module.

[0090] S7. Receive the torque data of the user's knee joint through the data processing module, process and analyze it to generate a torque measurement report, and at the same time, perform data display and interaction through the user interaction module.

[0091] Embodiment 2. Rehabilitation training for users with knee joint injuries

[0092] Specifically, it includes the following steps:

[0093] S1. The user is located on the seat, fixes the thigh to the seat through the thigh strap, and fixes the calf to the optical axis through the calf strap;

[0094] S2. The user selects the corresponding preset execution program for the rehabilitation training of knee joint injury patients through the user interaction module;

[0095] S3. Start the motor, drive the optical axis to perform a pendulum motion at a preset angular velocity through the harmonic reducer, drive the user's calf to flex and extend, and drive the optical axis to slide and extend on the linear bearing. At the same time, the torque output measurement module slides on the circular rail under the pulling action of the calf extension action, so that the rotation axis of the torque output measurement module is always aligned with the rotation axis of the user's knee joint;

[0096] S4. Real-time detect the flexion and extension angle of the user's knee joint through the potentiometer and transmit it to the data processing module, and the data processing module real-time collects the lower limb parameters of the user;

[0097] S5. Based on the biomechanical model and combined with the fuzzy PID control algorithm, the data processing module controls the output torque of the torque output module;

[0098] Specifically, the biomechanical model can calculate the torque required by the knee joint based on the following formula (2) according to the kinematic and mechanical characteristics of the user's knee joint:

[0099] τ = τ inertial +τ coriolis +τ gravity (2);

[0100] Where:

[0101]

[0102] τ gravity = mglsinθ (5);

[0103] In formulas (2)-(5): τ represents the torque required by the knee joint; τ inertial represents the inertial force caused by the angular acceleration of the knee joint; τ coriolis represents the Coriolis force and centrifugal force caused by the knee joint angle; τ gravity represents the torque caused by gravity; θ represents the flexion and extension angle of the knee joint; represents the angular velocity of the knee joint; represents angular acceleration; the and are obtained through the first and second derivatives of θ respectively; I represents the moment of inertia; represents the Coriolis force and centrifugal force matrix; l represents the distance from the center of mass to the axis of rotation; m represents the mass; g represents the acceleration due to gravity;

[0104] S6. Real - time collect the motion state data of the user's knee joint through the data processing module, and optimize the output torque of the torque output module based on the reinforcement learning algorithm according to the reward mechanism;

[0105] The reward mechanism is as follows: positive rewards occur when the user completes the training goal, and negative rewards occur when the user is uncomfortable or the training effect is poor; the reward mechanism is constructed based on the reinforcement learning algorithm, and the Actor - Criti method is used to evaluate the user's current motion state by combining the policy function and the value function. The policy function and the value function are represented by the following formulas (6) and (7) respectively:

[0106]

[0107] V(s)←V(s)+α[r + γV(s′)-V(s)] (7);

[0108] Where:

[0109]

[0110] In formulas (6)-(10): θ represents the parameter of the policy function; J(θ) represents the objective function of the policy; α represents the learning rate; s represents the current state; s′ represents the next state; a represents the action taken in state s; a′ represents the action taken in state s′; V(s) represents the value of state s; V(s′) represents the value of state s′; Q(s,a) represents the value of taking action a in state s; Q(s′,a′) represents the value of taking action a′ in state s′; represents the expectation; r represents the current reward; γ represents the discount factor, which is used to balance the current reward and the future reward, and its value range is 0≤γ≤10≤γ≤1; represents the gradient of the policy function; r + γV(s′)-V(s) represents the advantage function, which represents the quality of the current action; represents the objective function after updating using the policy gradient;

[0111] S7. Real - time detect the torque data of the user's knee joint through the static torque sensor and transmit it to the data processing module;

[0112] S8. Receive the torque data of the user's knee joint through the data processing module, process and analyze it to generate a rehabilitation strategy, and at the same time, display and interact with the data through the user interaction module.

[0113] Embodiment 3. Muscle training of the leg

[0114] S1. The user is located on the seat, fixes the thigh to the seat through the thigh strap, and fixes the calf to the optical axis through the calf strap;

[0115] S2. The user selects the corresponding preset execution program for the muscle training of the leg through the user interaction module;

[0116] S3. Start the motor, drive the optical axis to perform a pendulum motion at a preset angular velocity through the harmonic reducer, drive the user's calf to flex and extend, and drive the optical axis to slide and extend on the linear bearing. At the same time, the torque output measurement module slides on the circular track under the pulling action of the calf extension action, so that the rotation axis of the torque output measurement module is always aligned with the rotation axis of the user's knee joint;

[0117] S4. Real-time detect the flexion and extension angle of the user's knee joint through the potentiometer and transmit it to the data processing module, and the data processing module real-time collects the user's lower limb parameters;

[0118] S5. The data processing module controls the output torque of the torque output module based on the biomechanical model combined with the fuzzy PID control algorithm;

[0119] S6. Real-time detect the torque data of the user's knee joint through the static torque sensor and transmit it to the data processing module. The data processing module analyzes the torque data based on the muscle-skeleton model combined with the fatigue monitoring algorithm and judges the fatigue degree of the user's muscles;

[0120] The data processing module based on the muscle-skeleton model combined with the fatigue monitoring algorithm judges the motion state by analyzing the change trend of the signal characteristics through the torque, angle, and angular velocity data of the user's knee joint. The signal characteristics include the maximum torque of the knee joint, the average torque, the angle change range, and the angular velocity change value;

[0121] S7. Optimize the output torque of the torque output module according to the fatigue degree of the user's muscles, and gradually reduce the torque output after the training reaches a certain duration;

[0122] S8. Through the user interaction module, display the relevant data in real time during the whole training process, and generate training suggestions after the training ends.

[0123] In the above embodiment, the fuzzy PID control algorithm controls the output torque of the torque output module and is represented by the following formula (11):

[0124]

[0125] In formula (11): u(t) represents the control output; e(t) represents the error between the target torque and the actual torque; d represents the differential; t represents time; K p , K i , K d , are the proportional, integral, and differential coefficients respectively; the K p , K i , K d 's fuzzy rule tables are respectively as Figure 9 , 10 and 11 show.

[0126] The described fuzzy PID control algorithm is applicable to the control of nonlinear systems, has strong robustness to system parameter changes and external disturbances, and can easily establish and adjust a fuzzy rule base based on expert experience and experimental data, ensuring the scientificity and effectiveness of device torque measurement, rehabilitation training, and muscle training.

[0127] The above-described embodiments are only descriptions of the preferred embodiments of the present invention, and do not limit the scope of the present invention. Without departing from the design spirit of the present invention, various deformations and improvements made by those of ordinary skill in the art to the technical solutions of the present invention shall fall within the protection scope determined by the claims of the present invention.

Claims

1. A self-aligning knee joint testing and training system, comprising a seat and a thigh strap arranged on the seat for fixing the thigh of a user, characterized in that: It also includes a rotation axis alignment mechanism, a torque output mechanism, a torque measurement mechanism, a flexion and extension mechanism, and a data processing module arranged on the seat; The rotation axis alignment mechanism includes linear guide rails arranged in parallel on the left and right sides of the seat; The torque output mechanism comprises a motor slidably connected to the linear guide rail and a harmonic reducer coaxially connected to the motor shaft of the motor, the motor shaft of the motor is perpendicular to the linear guide rail, and the motor dynamically adjusts the output torque through a PID controller; The torque measurement mechanism includes a potentiometer built into the motor for detecting the flexion and extension angle of the user's knee joint and a static torque sensor coaxially connected to the harmonic reducer for measuring the user's knee joint torque; The flexion and extension mechanism includes a guide rail bracket connected to the static torque sensor, a linear bearing arranged on the guide rail bracket, an optical axis at one end of which is arranged in the linear bearing and can be slidably connected to the linear bearing, and a calf strap arranged at the other end of the optical axis for fixing the user's calf; The data processing module uses an embedded processor, which can control the output torque of the torque output mechanism based on a biomechanical model or an inverse dynamics model combined with a fuzzy PID control algorithm, and can process and analyze the data detected by the torque measurement mechanism based on a machine learning algorithm.

2. A self-aligning knee joint testing and training system according to claim 1, characterized in that: Also includes a user interaction module and a power module; The user interaction module is connected to the external interface of the data processing module, and includes an industrial control screen and a speaker for information transmission and user interaction; The power module is used to supply power to the system and provide circuit protection; The data processing module is preset with a corresponding execution program according to the user's measurement or training requirements, and is configured with a biomechanical model, an inverse dynamics model, a muscle-skeletal model, a fatigue monitoring algorithm and a reinforcement learning algorithm.

3. A self-aligning knee joint testing and training system according to claim 1, characterized in that: The linear guide includes a circular rail fixedly connected to the outer side surface of the seat and a slider sleeved on the circular rail; the motor is slidably connected to the slider through a frame, and a Hall sensor is provided on the frame as an emergency stop switch of the motor. The Hall sensor can send a motor stop command to the PID controller when the surrounding magnetic flux density reaches a preset threshold.

4. A self-aligning knee joint testing and training system according to claim 1, characterized in that: The guide rail bracket is provided with a permanent magnet, which can be used in conjunction with a Hall sensor for emergency stop control of the motor.

5. A self-aligning knee joint training method, which is implemented based on the self-aligning knee joint training system described in any one of claims 1 to 4, characterized in that: The testing and training method includes an isokinetic knee joint torque measurement method, which specifically includes the following steps: S1. The user is sitting on the seat and fixes the thigh to the seat through the thigh strap, and fixes the calf to the optical axis through the calf strap; S2, the user selects the corresponding preset execution program for isokinetic knee joint torque measurement through the user interaction module; S3, start the motor, drive the optical axis to perform a single pendulum motion at a preset angular velocity through the harmonic reducer, drive the user's calf to flex and extend, and drive the optical axis to slide and extend on the linear bearing. At the same time, the torque output measurement module slides on the circular rail under the pulling effect of the calf extension action, so that the rotation axis of the torque output measurement module is always aligned with the rotation axis of the user's knee joint; S4, detecting the user's knee flexion and extension angle in real time through a potentiometer and transmitting the angle to a data processing module, and collecting the user's lower limb parameters in real time through the data processing module; The lower limb parameters include the mass, length, moment of inertia, and center of mass position of the user's thigh, calf, and foot; S5, controlling the output torque of the torque output module through the data processing module based on the inverse dynamics model combined with the fuzzy PID control algorithm; The inverse dynamics model can calculate the torque required for the knee joint according to the motion state of the user's knee joint based on the following formula (1): In formula (1), τ represents the torque required by the knee joint; M(q) is the mass matrix, which represents the inertial characteristics of the system; represents the Coriolis force and centrifugal force matrix; G(q) gravity matrix; q represents the knee flexion and extension angle; represents the angular velocity of the knee joint; represents the angular acceleration of the knee joint; and Obtained by the first and second order derivatives of q respectively; S6, detecting the torque data of the user's knee joint in real time through a static torque sensor and transmitting the data to a data processing module; S7. Receive the torque data of the user's knee joint through the data processing module, process and analyze it, generate a torque measurement report, and display and interact with the data through the user interaction module.

6. A self-aligning knee joint testing and training method according to claim 5, characterized in that: The testing and training method also includes a rehabilitation training method for patients with knee joint injuries, which specifically includes the following steps: S1. The user is sitting on the seat and fixes the thigh to the seat through the thigh strap, and fixes the calf to the optical axis through the calf strap; S2. The user selects a preset execution program corresponding to the rehabilitation training for the patient with knee joint injury through the user interaction module; S3, start the motor, drive the optical axis to perform a single pendulum motion at a preset angular velocity through the harmonic reducer, drive the user's calf to flex and extend, and drive the optical axis to slide and extend on the linear bearing. At the same time, the torque output measurement module slides on the circular rail under the pulling effect of the calf extension action, so that the rotation axis of the torque output measurement module is always aligned with the rotation axis of the user's knee joint; S4, detecting the user's knee flexion and extension angle in real time through a potentiometer and transmitting the angle to a data processing module, and collecting the user's lower limb parameters in real time through the data processing module; S5, controlling the output torque of the torque output module through the data processing module based on the biomechanical model combined with the fuzzy PID control algorithm; The biomechanical model can calculate the required torque of the knee joint according to the kinematic and mechanical characteristics of the user's knee joint based on the following formula (2): τ=τ inertial +t coriolis +t gravity (2); in: t gravity = mgl sinθ (5); In formulas (2)-(5): τ represents the torque required by the knee joint; τ inertial represents the inertial force caused by the angular acceleration of the knee joint; τ coriolis represents the Coriolis force and centrifugal force caused by the knee joint angle; τ gravity represents the moment caused by gravity; θ represents the flexion and extension angle of the knee joint; represents the angular velocity of the knee joint; represents angular acceleration; and They are obtained by the first and second order derivatives of θ respectively; I represents the moment of inertia; represents the Coriolis force and centrifugal force matrix; l represents the distance from the center of mass to the rotation axis; m represents the mass; g represents the gravitational acceleration; S6. The motion state data of the user's knee joint is collected in real time through the data processing module, and the output torque of the torque output module is optimized according to the reward mechanism based on the reinforcement learning algorithm; The reward mechanism is as follows: positive rewards appear when the user completes the training goal, and negative rewards appear when the user is uncomfortable or the training effect is not good. The reward mechanism is constructed based on the reinforcement learning algorithm, and the Actor-Criti method is used to combine the strategy function and the value function to evaluate the user's current motion state. The strategy function and the value function are respectively expressed by the following formulas (6) and (7): V(s)←V(s)+α[r+γV(s′)-V(s)] (7); in: In formulas (6)-(10), θ represents the parameters of the policy function; J(θ) represents the objective function of the policy; α represents the learning rate; s represents the current state; s′ represents the next state; a represents the action taken in state s; a′ represents the action taken in state s′; V(s) represents the value of state s; V(s′) represents the value of state s′; Q(s,a) represents the value of taking action a in state s; Q(s′,a′) represents the value of taking action a′ in state s′; represents expectation; r represents current reward; γ represents discount factor, which is used to balance current reward and future reward, and its value range is 0≤γ≤10≤γ≤1; represents the gradient of the policy function; r+γV(s′)-V(s) represents the advantage function, which indicates the pros and cons of the current action; Represents the objective function after updating using policy gradient; S7, detecting the torque data of the user's knee joint in real time through a static torque sensor and transmitting the data to a data processing module; S8. Receive the torque data of the user's knee joint through the data processing module, process and analyze it, generate a rehabilitation strategy, and display and interact with the data through the user interaction module.

7. A self-aligning knee joint testing and training method according to claim 5, characterized in that: The testing and training method also includes a leg muscle training method, which specifically includes the following steps: S1. The user is sitting on the seat and fixes the thigh to the seat through the thigh strap, and fixes the calf to the optical axis through the calf strap; S2, the user selects a preset execution program corresponding to the leg muscle training through the user interaction module; S3, start the motor, drive the optical axis to perform a single pendulum motion at a preset angular velocity through the harmonic reducer, drive the user's calf to flex and extend, and drive the optical axis to slide and extend on the linear bearing. At the same time, the torque output measurement module slides on the circular rail under the pulling effect of the calf extension action, so that the rotation axis of the torque output measurement module is always aligned with the rotation axis of the user's knee joint; S4, detecting the user's knee flexion and extension angle in real time through a potentiometer and transmitting the angle to a data processing module, and collecting the user's lower limb parameters in real time through the data processing module; S5, controlling the output torque of the torque output module through the data processing module based on the biomechanical model combined with the fuzzy PID control algorithm; S6. Detect the torque data of the user's knee joint in real time through a static torque sensor and transmit it to a data processing module. The data processing module analyzes the torque data based on a muscle-bone model combined with a fatigue monitoring algorithm and determines the fatigue degree of the user's muscles; The data processing module is based on the muscle-skeletal model combined with the fatigue monitoring algorithm to judge the motion state by analyzing the change trend of signal characteristics through the torque, angle and angular velocity data of the user's knee joint. The signal characteristics include the maximum torque of the knee joint, the average torque, the angle change range, and the angular velocity change value; S7. Optimize the output torque of the torque output module according to the fatigue level of the user's muscles, and gradually reduce the torque output after the training reaches a certain length of time; S8. Display relevant data in real time during the entire training process through the user interaction module, and generate training suggestions after the training is completed.

8. A self-aligning knee joint testing and training method according to any one of claims 6-7, characterized in that: The output torque of the torque output module controlled by the fuzzy PID control algorithm is expressed by the following formula (11): In formula (11), u(t) represents the control output; e(t) represents the error between the target torque and the actual torque; d represents the differential; t represents the time; K p , K i , K d , are the proportional, integral, and differential coefficients respectively.

Images