Artificial muscle collaborative mirror rehabilitation system and control method based on ultrasonic detection control

Abstract

The present invention discloses an artificial muscle collaborative mirror rehabilitation system and control method based on ultrasonic detection control. The rehabilitation system includes an ultrasonic detection group, a pneumatic muscle group, and an electric control group. The ultrasonic detection group performs high-density array ultrasonic detection on the muscles of the user's healthy limb to obtain the muscle state of the healthy limb, and analyzes the muscle state information through the electric control group to resolve the limb collaborative movement information. The pneumatic muscle group simulates the layout and attachment method of human limb skeletal muscles on the bones, and is controlled by the electric control group for collaborative drive to achieve mirror movement of the healthy limb. The present invention adopts ultrasonic detection and pneumatic muscle simulation of mirror movement, which can detect the movement pattern of the healthy limb (including movement amplitude, movement angle, movement speed and strength, etc.) at the muscle level, and collaboratively control the affected limb in real time to completely imitate the movement of the healthy limb, so that the rehabilitation equipment can assist the patient to achieve the mirror movement effect and improve the rehabilitation efficiency.

Description

Technical Field

[0001] The present invention relates to an artificial muscle collaborative mirror rehabilitation system and a control method based on ultrasonic detection control, and belongs to the field of wearable rehabilitation equipment. Background Art

[0002] Motor dysfunction, commonly known as hemiplegia, can have a serious impact on people's work and daily lives. Currently, a considerable number of people in my country suffer from motor dysfunction, primarily due to illnesses such as stroke and accidents. Whether motor dysfunction is caused by illness or limb injury, effective limb rehabilitation training can improve limb mobility and promote neurological remodeling.

[0003] In 1960, orthopedic surgeon Robert proposed continuous passive motion (CPM) as a rehabilitation training method for motor dysfunction. By using external force to drive the affected limb through a full range of passive motion for a long time, this prevents joint stiffness and promotes blood circulation in the affected limb, accelerating recovery. CPM therapy has been proven to be effective, but due to its high intensity and high repetitiveness, it places high demands on the physical therapist's workload. To address these issues, wearable rehabilitation devices and related technologies have been applied to limb rehabilitation. The limb rehabilitation system replaces the physical therapist, driving the affected limb to perform passive movements and promote limb function recovery.

[0004] Mirror therapy (MT) has also been widely used in motor rehabilitation, achieving excellent results. Mirror therapy, also known as mirror visual feedback (MVF), utilizes the principle of plane mirror imaging to replicate images of the healthy side's movements on the affected side, encouraging the patient to imagine the affected side's movements. This treatment approach, combining visual illusions, visual feedback, and virtual reality with rehabilitation training programs, can be extended to include exoskeleton-assisted therapy, where the healthy side controls the affected side, upgrading virtual mirror therapy to real-life mirror movement therapy.

[0005] However, the current method of mirror-type passive rehabilitation training using simple rehabilitation training equipment has many disadvantages. For example, when driving the movement of the affected limb, the movement pattern of the healthy limb (including movement amplitude, movement angle, movement speed and strength, etc.) cannot be captured, resulting in the affected limb being unable to completely imitate the movement of the healthy limb, making it impossible for the rehabilitation equipment to assist the patient in achieving the mirror movement effect, thereby reducing the efficiency of rehabilitation.

[0006] To address the above issues, it is necessary to design a rehabilitation device that can detect the movement information of the healthy limb in real time (including movement amplitude, movement angle, movement speed and strength, etc.) and drive the affected limb to complete mirror movements in the same movement pattern in real time based on the above information. Summary of the Invention

[0007] In order to solve the above technical problems, the present invention provides an artificial muscle collaborative mirror rehabilitation system and control method based on ultrasonic detection control, which can detect the movement information of the healthy side limb (including movement amplitude, movement angle, movement speed and strength, etc.) in real time, and drive the affected side limb in real time to complete the mirror movement in the same movement pattern according to the above information.

[0008] In order to solve the above technical problems, the present invention adopts the following technical solution:

[0009] An artificial muscle collaborative mirror rehabilitation system based on ultrasonic detection control, comprising an electronic control unit, an ultrasonic detection unit worn on the user's healthy limb to obtain the muscle status of the healthy limb, and a pneumatic muscle unit worn on the user's affected limb to drive the affected limb to perform mirror rehabilitation movements;

[0010] The electronic control group includes a microcontroller, a posture sensor for detecting the posture of the healthy side and the posture of the affected side under the drive of the pneumatic muscle group in real time, and an air pump driving board for driving the pneumatic muscle group to move;

[0011] The microcontroller receives the muscle status signal of the healthy side limb transmitted by the ultrasonic detection group, and outputs the control signal to the air pump driving board, and the air pump driving board drives the pneumatic muscle group to control the affected side limb to perform mirror rehabilitation movements; at the same time, the microcontroller receives the posture information of the healthy side limb and the affected side limb detected by the posture sensor, and compares the posture information of the affected side limb with the posture information of the healthy side limb, performs control correction through the algorithm, and outputs the corrected information to the air pump driving board to correct the movement of the pneumatic muscle group controlling the affected side limb.

[0012] A further improvement of the technical solution of the present invention is that the muscle state obtained by the ultrasonic detection group includes the muscle's force intensity, force speed and force direction.

[0013] A further improvement of the technical solution of the present invention is that: the ultrasonic detection group includes a beamformer, a pulse generator and an ultrasonic probe array board, and the ultrasonic probe array board is composed of multiple ultrasonic probes integrated on the array board and arranged in a matrix;

[0014] The microcontroller initializes the beamformer and generates a beam signal to act on the corresponding channel of the pulse generator, thereby generating a corresponding high-voltage pulse signal. The high-voltage pulse signal is output to the ultrasonic probe of the corresponding channel, stimulating the ultrasonic probe to emit an ultrasonic excitation signal and receive an echo signal, and at the same time transmitting the echo signal to the microcontroller.

[0015] A further improvement of the technical solution of the present invention is that all the ultrasonic probes on the ultrasonic probe array plate are numbered, and individually transmit ultrasonic excitation signals and receive corresponding echo signals.

[0016] A further improvement of the technical solution of the present invention is that the array plate of the ultrasound probe array plate is configured to be made of a flexible material that can fit closely to a limb.

[0017] A further improvement of the technical solution of the present invention is that: the pneumatic muscle group includes a wearable outer garment worn on the user's limbs and multiple groups of muscle units hinged on the wearable outer garment, and the arrangement and position of the muscle units on the wearable outer garment are the same as the attachment method and position of the corresponding human skeletal muscles on the bones; one end of the muscle unit is airtightly connected to the array air pump through a flexible air tube; the array air pump includes multiple air pumps arranged in a matrix structure, each air pump can independently control the air volume through an air pump drive plate, and each group of muscle units is driven by a corresponding air pump to undergo morphological changes.

[0018] A further improvement of the technical solution of the present invention is that the executive body of the muscle unit is a pneumatic muscle structure, which mainly includes an outer woven mesh and an inner elastic rubber tube, has a highly nonlinear characteristic, and its elongation is proportional to the interval pressure, simulating the expansion and contraction deformation of human muscles to perform driving operations; ball joints are respectively provided at both ends of one side of the muscle unit, which are hinged to the ball joint seat on the wearable outer garment.

[0019] A further improvement of the technical solution of the present invention is that the wearable outer garment is made of tough fabric that is not easy to stretch and is extremely unlikely to undergo elastic deformation, and is fixed to the limbs by straps.

[0020] A further improvement of the technical solution of the present invention is that the posture sensors are arranged at the joint ends of the healthy side and the affected side, and the detected posture information includes the movement amplitude, movement angle, movement speed and strength of the limbs.

[0021] In order to solve the above technical problems, the second technical solution adopted by the present invention is:

[0022] Step 1: Identify the patient's limb that requires rehabilitation, determine the agonist and antagonist muscles that drive the limb's movements based on anatomical theory, locate the muscle bellies of the aforementioned muscles, and determine the muscle sites to be examined by ultrasound.

[0023] Step 2: Attach the ultrasound detection group to the determined muscle site, adhere the ultrasound probe array board to the corresponding muscle site using medical double-sided tape, compare the ultrasound information of multiple ultrasound probes on the ultrasound probe array board, and select the probe that first detects the ultrasound information change in time sequence as the probe of the muscle belly site. If multiple probes detect the ultrasound information change at the same time, select the ultrasound information with the largest amplitude as the ultrasound probe information of the muscle belly site, and complete the ultrasound detection;

[0024] Step 3: The microcontroller of the electronic control group processes the signals collected by the ultrasonic detection group, analyzes the ultrasonic signals of each muscle site separately, measures the morphological changes of each muscle and analyzes its status information, including force intensity, force speed and force direction;

[0025] Step 4: Integrate the status information of multiple muscles to perform muscle synergy analysis. The mathematical model formula is:

[0026]

[0027] Each muscle coordination pattern can be represented as a temporal module w i (t) and spatial modules w, m(t) is the double linear combination of the time module and the spatial module w, m(t) is the sEMG signal collected at time t, P and N are the number of time modules and spatial modules respectively, c ij t ij and are the scalar activation coefficients and time delays of temporal module i and spatial module j, respectively, to obtain the number of muscle synergies and the similarity of synergies, so as to calculate the muscle force of each individual muscle in the pattern;

[0028] Step 5: Based on the CNN-SVM deep learning model, a multi-map between the ultrasound signal and muscle force of a single muscle is established. The mapping includes: mapping between ultrasound amplitude and muscle force, mapping between ultrasound frequency and muscle force change, and mapping between ultrasound spatiotemporal characteristics and muscle force distribution. The muscle force information of each muscle unit involved in muscle synergy is then calculated. The main steps are:

[0029] Step 51: Convert the original multi-channel signal into two-dimensional samples as the input of CNN through preprocessing, and divide the samples into training set, validation set and test set;

[0030] Step 52: Design and optimize the CNN architecture, use the training set to train the network, automatically extract the feature vector corresponding to each training set, and use its validation set to verify the network performance;

[0031] Step 53: Use the extracted feature vector as an input sample to train the SVM;

[0032] Step 54: Replace the fully connected layer of CNN with the trained SVM, then input the test sample into the trained CNN, and finally obtain the feature vector corresponding to the test sample;

[0033] Step 55: The extracted feature vector of the test sample is used as an input sample, and the trained SVM is used to estimate muscle strength;

[0034] Step 6: The electronic control group processes the coordinated motion information of the multiple muscles and the status information of the individual muscles and controls the air pump drive board. The array air pump drives the muscle units to drive the patient's affected limb to achieve a mirror-image movement of the healthy limb.

[0035] Step 7: After the patient's affected limb produces a mirror movement, the posture sensor detects the posture information of the same joint end position of the healthy and affected limbs, and compares the postures of the healthy and affected sides. If there is a posture deviation, the deviation information is calculated and correction parameters are introduced to achieve closed-loop control. After repeated corrections, accurate mirror movements of the healthy and affected sides can be achieved.

[0036] The specific method for calculating deviation information is:

[0037] Step 71: The detected healthy side posture coordinates are expressed as x, y, z, and the detected affected side posture coordinates are expressed as x, y, z. h ,y h ,z h Indicates that Δx, Δy, and Δz are calculated respectively, where: Δx=xx h ,Δy=yy h ,Δz=zz h ;

[0038] Step 72: Set the target coordinate value of the affected limb predicted in steps 5 and 6 to x p ,y p ,z p , and use Δx, Δy, Δz as correction parameters, with x g ,y g ,z g is the target coordinate value after correction, where: x g =x p +Δx,y g =y p +Δy,z g =z p +Δz;

[0039] Step 73: Repeat steps 71 and 72 repeatedly until Δx, Δy, and Δz are within the allowable error range.

[0040] Due to the adoption of the above technical solution, the technical advancements achieved by the present invention are:

[0041] The present invention adopts ultrasonic detection and pneumatic muscle simulation of mirror motion. When driving the affected limb to move, the movement pattern of the healthy limb (including movement amplitude, movement angle, movement speed and strength, etc.) can be detected at the muscle level. At the same time, the posture of the healthy side and the affected side can be compared. If there is a posture deviation, the deviation information is calculated and correction parameters are introduced to realize closed-loop control. After repeated corrections, accurate mirror movements of the healthy side and the affected side are realized, and the affected limb is coordinated and controlled in real time to completely imitate the movement of the healthy side limb, so that the rehabilitation equipment can assist the patient to achieve a complete mirror motion effect and improve the rehabilitation efficiency.

[0042] The ultrasound probe array board of the present invention is composed of multiple ultrasound probes integrated on the array board and arranged in a matrix. The ultrasound information of the multiple ultrasound probes of the ultrasound probe array board is compared, and the probe that first detects the change in ultrasound information is selected from the time sequence as the probe of the muscle belly site. If it is detected that multiple probes first detect the change in ultrasound information at the same time, the ultrasound information with the largest amplitude is selected from the amplitude as the ultrasound probe information of the muscle belly site. This can more accurately locate the ultrasound probe information of the muscle belly site, making data collection more precise. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a schematic diagram of the overall structure of the present invention;

[0044] Figure 2 It is a structural diagram of the ultrasonic detection group and the electric control group of the present invention;

[0045] Figure 3 It is a schematic structural diagram of the pneumatic muscle group of the present invention;

[0046] Figure 4 Schematic diagram of the work flow of the control method of the present invention;

[0047] Among them, 1. Ultrasonic detection group, 11. Beamformer, 12. Pulse generator, 13. Pulse generator, 2. Pneumatic muscle group, 21. Wearable outer garment, 211. Ball joint seat, 22. Muscle unit, 221. Ball joint head, 23. Flexible air tube, 24. Array air pump, 3. Electronic control group, 31. Microcontroller, 32. Posture sensor, 33. Air pump driver board. DETAILED DESCRIPTION

[0048] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments are some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0049] The present invention provides an artificial muscle collaborative mirror rehabilitation system and control method based on ultrasonic detection control, the specific structure is as follows Figure 1 As shown, it includes an ultrasonic detection group 1, a pneumatic muscle group 2, and an electronic control group 3.

[0050] Specifically, the following uses the rehabilitation scenario of patients with upper limb motor dysfunction as a specific example. The structure and principle of the artificial muscle collaborative mirror rehabilitation system based on ultrasonic detection control are described in conjunction with the accompanying drawings. The specific working mode of the system in the mirror rehabilitation exercise of the upper and lower arms around the shoulder and elbow joints is analyzed:

[0051] Figure 2 What is shown is the ultrasonic detection group 1 and the electronic control group 3. The ultrasonic detection group 1 includes: a beam former 11, a pulse generator 12, and an ultrasonic probe array board 13. The microcontroller 31 initializes the beam former 11 and generates a beam signal to act on the corresponding channel of the pulse generator 12, thereby generating a corresponding high-voltage pulse signal. The high-voltage pulse signal is output to the ultrasonic probe of the corresponding channel, stimulating the probe to generate an ultrasonic signal. In order to facilitate wearing, multiple ultrasonic probes are integrated on the ultrasonic probe array board 13. The ultrasonic probe array board 13 is flexible and can fit tightly with the limbs. It is fixed and adhered to the limbs through medical double-sided tape. All probes on the ultrasonic probe array board 13 are numbered and transmit ultrasonic excitation signals and echo signals separately. The electronic control group 3 includes a microcontroller 31, a posture sensor 32, and an air pump drive board 33. The microcontroller 31 controls the generation of the excitation wave and the processing and analysis of the echo, receives information from the posture sensor 32, and outputs control information to the air pump drive board 33. The posture sensor 32 is composed of an MPU6050 nine-axis sensor, a wireless transceiver, and a lithium battery. It can be attached to the joint end position of the wearable outer garment 21 and the corresponding healthy limb position, and detects the posture of the healthy limb and the limb posture achieved by the affected limb under the drive of the pneumatic muscle group 2 in real time. The posture sensor performs posture detection in the following way: the gyroscope and accelerometer of the MPU6050 are used to collect the angular velocity signal and acceleration signal of the object's rotation, respectively, and convert them into digital signals through an ADC (analog-to-digital converter). These two signals are then processed and integrated to respectively calculate the position change and posture change of the corresponding joint end, thereby obtaining the posture information of the same joint end position of the current healthy and affected limbs.

[0052] After receiving the information from the posture sensor 32, the microcontroller 31 compares and analyzes it with the posture information of the healthy limb, controls and corrects it through an algorithm, and then outputs the corrected information to the air pump drive board 33. After accumulating multiple corrections, the posture error between the healthy side and the affected side can be controlled within the allowable range.

[0053] Figure 3The specific structure of the pneumatic muscle group 2 is shown. The pneumatic muscle group 2 includes: a wearable outer garment 21, a muscle unit 22, a flexible air tube 23 and an array air pump 24. The wearable outer garment 21 is similar in appearance to ordinary clothing and can be worn on the corresponding limb of the user. A ball joint seat 211 is fixed on the wearable outer garment 21, and is hinged and fixed to the ball joint head 221 on the muscle unit 22 through a ball joint. The muscle unit 22 is airtightly connected to the array air pump 24 through a flexible air tube 23, and the array air pump 24 drives the muscle unit 22 to change its shape. The wearable outer garment 21 is made of a tough fabric that is not easy to stretch and is extremely unlikely to undergo elastic deformation. It is firmly fixed to the limb by straps. The executive body of the muscle unit 22 is a conventional pneumatic muscle The muscle unit 22 has a meat structure, and the main body is mainly composed of an outer woven mesh and an inner elastic rubber tube. It has a highly nonlinear characteristic, and its elongation is proportional to the interval pressure. It can simulate the expansion and contraction of human muscles to drive operations. Both ends of the muscle unit 22 are respectively provided with a ball joint head 221, which is used to be fixed to the ball joint seat 211 on the wearable outer garment 21, and is connected to the flexible air pipe 23 at one end as the input port of the air source. The arrangement of the muscle unit 22 on the wearable outer garment 21 is the same as the arrangement of the corresponding human skeletal muscle on the skeleton. The array air pump 24 is a plurality of air pumps combined into a whole, and each air pump corresponds to an air source output port. Each air source output port can control the air volume individually and is airtightly connected to the flexible air pipe 23.

[0054] Figure 4 The demonstration shows the control method and workflow of an artificial muscle collaborative mirror rehabilitation system based on ultrasonic detection control, using mirror rehabilitation exercises for the upper limbs as an example:

[0055] First, the patient's limb to be rehabilitated is determined, namely the upper limb, including the upper and lower arms, and the shoulder and elbow joints. Based on anatomical theory, the skeletal muscles that drive the movement of this limb are determined to be the biceps and triceps (upper arm), and the brachioradialis, extensor carpi radialis longus, and flexor carpi ulnaris (lower arm). The muscle bellies of the above muscles are located in the patient, and the muscle sites to be tested by ultrasound are determined.

[0056] Then, the ultrasonic detection group 1 is attached to the determined muscle site, and the ultrasonic probe array plate 13 is attached to the corresponding muscle site with medical double-sided tape to perform ultrasonic detection. The ultrasonic information of multiple ultrasonic probes on the ultrasonic probe array plate is compared, and the ultrasonic information of multiple ultrasonic probes on the ultrasonic probe array plate is compared. The probe that first detects the change in ultrasonic information is selected from the time sequence as the probe of the muscle belly site. If it is detected that multiple probes first detect the change in ultrasonic information at the same time, the ultrasonic information with the largest amplitude is selected from the amplitude as the ultrasonic probe information of the muscle belly site to complete the ultrasonic detection;

[0057] The microcontroller 31 of the electronic control group 3 processes the signals collected by the ultrasonic detection group 1, analyzes the ultrasonic signals of the individual muscle sites, measures the morphological changes of the muscles and analyzes the status information of the individual muscles, including the force intensity, force speed and force direction;

[0058] Based on the status information of a single muscle, the status information of multiple muscles is integrated to perform collaborative analysis and analyze the upper limb movement information driven by the coordinated force of multiple muscles, including movement amplitude, movement angle, movement speed and strength. The mathematical model formula is:

[0059]

[0060] Each muscle coordination pattern can be represented as a temporal module w i (t) and spatial modules w, m(t) is the double linear combination of the time module and the spatial module w, m(t) is the sEMG signal collected at time t, P and N are the number of time modules and spatial modules respectively, c ij t ij and are the scalar activation coefficients and time delays of temporal module i and spatial module j, respectively, so as to obtain the number of muscle synergies and the similarity of synergies, so as to calculate the muscle strength of each individual muscle in the pattern.

[0061] Establish a multiple mapping between ultrasound signals and muscle force of a single muscle. The mapping includes mapping between ultrasound amplitude and muscle force, mapping between ultrasound frequency and muscle force change, and mapping between ultrasound spatiotemporal characteristics and muscle force distribution. Then, calculate the muscle force information of each muscle unit involved in muscle synergy. The main steps are:

[0062] Step 51: Convert the original multi-channel signal into two-dimensional samples as the input of CNN through preprocessing, and divide the samples into training set, validation set and test set;

[0063] Step 52: Design and optimize the CNN architecture, use the training set to train the network, automatically extract the feature vector corresponding to each training set, and use its validation set to verify the network performance;

[0064] Step 53: Use the extracted feature vector as an input sample to train the SVM;

[0065] Step 54: Replace the fully connected layer of CNN with the trained SVM, then input the test sample into the trained CNN, and finally obtain the feature vector corresponding to the test sample;

[0066] Step 55: The extracted feature vector of the test sample is used as an input sample, and the trained SVM is used to estimate muscle strength.

[0067] The electronic control group 3 processes the coordinated motion information of the above-mentioned multiple muscles and the status information of the single muscle, and then controls the air pump drive board 33. The array air pump 24 drives the muscle unit 22. Since there are a total of 7 muscles in the healthy upper limb, 7 muscle units 22 are set on the affected side, which are respectively arranged at the positions corresponding to the biceps brachii, triceps brachii (upper arm), brachioradialis, extensor carpi radialis longus, and flexor carpi ulnaris (forearm), and the hinge positions at both ends of the muscle unit 22 are also the same as the attachment positions of the above-mentioned skeletal muscles on the bones, thereby simulating the human body muscles to drive the patient's affected limb to achieve the mirror image upper limb movement of the healthy limb.

[0068] After the patient's affected limb produces a mirror movement, the posture sensor 32 detects the posture information of the same joint end position of the healthy and affected limbs, and compares the postures of the healthy and affected sides. If there is a posture deviation, the deviation information is calculated and correction parameters are introduced to achieve closed-loop control. After repeated corrections, accurate mirror movements of the healthy and affected sides can be achieved.

[0069] The specific method for calculating deviation information is:

[0070] Step 71: The detected healthy side posture coordinates are expressed as x, y, z, and the detected affected side posture coordinates are expressed as x, y, z. h ,y h ,z h Indicates that Δx, Δy, and Δz are calculated respectively, where: Δx=xx h ,Δy=yy h ,Δz=zz h ;

[0071] Step 72: Set the target coordinate value of the affected limb predicted in steps 5 and 6 to x p ,y p ,z p , and use Δx, Δy, Δz as correction parameters, with x g ,y g ,z g is the target coordinate value after correction, where: x g =x p +Δx,y g =y p +Δy,z g =z p +Δz;

[0072] Step 73: Repeat steps 71 and 72 repeatedly until Δx, Δy, and Δz are within the allowable error range.

[0073] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the present invention.

Claims

1. An artificial muscle collaborative mirror rehabilitation system based on ultrasonic detection and control, characterized by: The device comprises an electric control group (3), an ultrasonic detection group (1) worn on the healthy limb of the user for obtaining the muscle status of the healthy limb, and a pneumatic muscle group (2) worn on the affected limb of the user for driving the affected limb to perform mirror rehabilitation movements; The electric control group (3) includes a microcontroller (31), a posture sensor (32) for detecting the posture of the healthy side and the posture of the affected side under the drive of the pneumatic muscle group (2) in real time, and an air pump driving board (33) for driving the pneumatic muscle group (2) to move; The microcontroller (31) receives the muscle status signal of the healthy limb transmitted by the ultrasonic detection group (1), and outputs the control signal to the air pump driving board (33), and the air pump driving board (33) drives the pneumatic muscle group (2) to control the affected limb to perform mirror rehabilitation movements; at the same time, the microcontroller (31) receives the posture information of the healthy limb and the affected limb detected by the posture sensor (32), and compares the posture information of the affected limb with the posture information of the healthy limb, performs control correction through an algorithm, and outputs the corrected information to the air pump driving board (33) to correct the movement of the pneumatic muscle group (2) to control the affected limb; The ultrasonic detection group (1) comprises a beam former (11), a pulse generator (12) and an ultrasonic probe array plate (13), wherein the ultrasonic probe array plate (13) comprises a plurality of ultrasonic probes integrated on the array plate and arranged in a matrix; The microcontroller (31) initializes the beam former (11) and generates a beam signal to act on the corresponding channel of the pulse generator (12), thereby generating a corresponding high-voltage pulse signal. The high-voltage pulse signal is output to the ultrasonic probe of the corresponding channel, stimulating the ultrasonic probe to emit an ultrasonic excitation signal and receive an echo signal, and at the same time transmitting the echo signal to the microcontroller (31); all ultrasonic probes on the ultrasonic probe array board (13) are numbered, and individually transmit ultrasonic excitation signals and receive corresponding echo signals; The muscle belly sites are selected as follows: Comparing the ultrasonic information of the multiple ultrasonic probes of the ultrasonic probe array plate (13), selecting the probe that first detects the change in ultrasonic information as the probe at the muscle belly site in terms of time sequence, and if it is detected that multiple probes first detect the change in ultrasonic information at the same time, selecting the ultrasonic information with the largest amplitude as the ultrasonic probe information at the muscle belly site in terms of amplitude, and completing the ultrasonic detection; The microcontroller (31) of the electronic control group (3) processes the signals collected by the ultrasonic detection group (1), analyzes the ultrasonic signals of each muscle site respectively, measures the morphological changes of each muscle and analyzes its state information, including the force intensity, force speed and force direction; The state information of multiple muscles is integrated to perform muscle synergy analysis. The mathematical model formula is: Each muscle coordination pattern can be represented as a temporal module w i (t) and spatial module w j The double linear combination of m(t) is the sEMG signal collected at time t, P and N are the number of time modules and space modules respectively, c ij t ij and are the scalar activation coefficients and time delays of temporal module i and spatial module j, respectively, to obtain the number of muscle synergies and the similarity of synergies, so as to calculate the muscle force of each individual muscle in the pattern; Based on the CNN-SVM deep learning model, a multi-level mapping between ultrasound signals and muscle force of a single muscle is established. The mapping includes: mapping between ultrasound amplitude and muscle force, mapping between ultrasound frequency and muscle force change, and mapping between ultrasound spatiotemporal characteristics and muscle force distribution. The muscle force information of each muscle unit involved in muscle synergy is then calculated. The main steps are: Through preprocessing, the original multi-channel signal is converted into two-dimensional samples as the input of CNN, and the samples are divided into training set, validation set and test set; Design and optimize the CNN architecture, train the network using the training set, automatically extract the feature vectors corresponding to each training set, and verify the network performance using its validation set; The extracted feature vectors are used as input samples to train the SVM; Replace the fully connected layer of CNN with the trained SVM, then input the test sample into the trained CNN, and finally obtain the feature vector corresponding to the test sample; The extracted feature vector of the test sample is used as the input sample, and the trained SVM is used to estimate the muscle strength; The electric control group (3) processes the coordinated motion information of the multiple muscles and the state information of the individual muscles and controls the air pump driving board (33), and the array air pump (24) drives the muscle units to drive the affected limb to achieve the mirror image movement of the healthy limb; After the affected limb produces a mirror-image movement, the posture sensor (32) detects the posture information of the same joint end position of the healthy and affected limbs, and compares the postures of the healthy and affected sides. If there is a posture deviation, the deviation information is calculated and a correction parameter is introduced to realize closed-loop control. After repeated corrections, accurate mirror-image movements of the healthy and affected sides are realized. The specific method for calculating deviation information is: The detected healthy side posture coordinates are expressed as x, y, z, and the detected affected side posture coordinates are expressed as x h ,y h ,z h Indicates that Δx, Δy, and Δz are calculated respectively, where: Δx=xx h ,Δy=yy h ,Δz=zz h ; Let the predicted target coordinate value of the affected limb be x p ,y p ,z p , and use Δx, Δy, Δz as correction parameters, with x g ,y g ,z g is the target coordinate value after correction, where: x g =x p +Δx,y g =y p +Δy,z g =z p +Δz; Repeat the first two steps of the deviation information calculation method repeatedly until Δx, Δy, and Δz are controlled within the allowable error range.

2. The artificial muscle collaborative mirror rehabilitation system based on ultrasonic detection control according to claim 1 is characterized in that: The muscle state acquired by the ultrasonic detection group (1) includes the muscle's force intensity, force speed and force direction.

3. The artificial muscle collaborative mirror rehabilitation system based on ultrasonic detection control according to claim 1, characterized in that: The array plate of the ultrasonic probe array plate (13) is made of a flexible material that can fit tightly against a limb.

4. The artificial muscle collaborative mirror rehabilitation system based on ultrasonic detection control according to claim 1, characterized in that: The pneumatic muscle group (2) comprises a wearable outer garment (21) worn on a user's limbs and a plurality of muscle units (22) hinged on the wearable outer garment (21); the arrangement and position of the muscle units (22) on the wearable outer garment (21) are identical to the attachment and position of the corresponding skeletal muscles on the skeleton of a human; one end of the muscle unit (22) is airtightly connected to an array air pump (24) via a flexible air tube (23); the array air pump (24) comprises a plurality of air pumps arranged in a matrix structure, each air pump being capable of individually controlling air volume via an air pump drive plate (33), and each group of muscle units (22) undergoing morphological changes driven by a corresponding air pump.

5. The artificial muscle collaborative mirror rehabilitation system based on ultrasonic detection control according to claim 4 is characterized in that: The muscle unit (22) is an actuator of a pneumatic muscle structure, comprising an outer braided mesh and an inner elastic rubber tube, and has a highly nonlinear characteristic. Its elongation is proportional to the interval pressure, simulating the expansion and contraction deformation of human muscles to perform driving operations; both ends of one side of the muscle unit (22) are respectively provided with ball joint heads (221), which are hinged to the ball joint seat (211) on the wearable outer garment (21).

6. The artificial muscle collaborative mirror rehabilitation system based on ultrasonic detection control according to claim 4, characterized in that: The wearable outer garment (21) is made of tough cloth that is not easily stretched and is extremely unlikely to undergo elastic deformation, and is fixed to the limbs by means of straps.

7. The artificial muscle collaborative mirror rehabilitation system based on ultrasonic detection control according to claim 1, characterized in that: The posture sensors (32) are arranged at the joint ends of the healthy side and the affected side, and the posture information detected includes the movement amplitude, movement angle, movement speed and strength of the limbs.

Images