Wrist rehabilitation training system based on muscle synergy and variable stiffness impedance control
Through a wrist rehabilitation training system based on muscle coordination and variable stiffness impedance control, using surface electromyography signals and virtual reality technology, the problem of hand muscle strength recovery in rehabilitation training for stroke patients is solved, achieving efficient rehabilitation training results and patients Autonomy.
Patent Information
- Application Number
- CN202210443610.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-04-26
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2042-04-26
AI Technical Summary
In rehabilitation training, stroke patients cannot obtain timely and effective rehabilitation training due to the imbalance of the number of patients and the number of medical staff. Traditional rehabilitation training methods are difficult to ensure the intensity and time of training, especially the recovery of hand muscle strength, and traditional methods lack patients. Autonomy and enthusiasm.
Adopt a wrist rehabilitation training system based on muscle coordination and variable stiffness impedance control, and establish joint angle and stiffness mapping models through surface electromyography signal acquisition and preprocessing, combine with serious virtual reality games, and use impedance control methods to rehabilitate in a virtual environment. Training, adjust the difficulty of the game to improve the training effect.
It improves the accuracy and efficiency of wrist rehabilitation training, enhances the patient's autonomy and enthusiasm, effectively improves the recovery effect of hand muscle strength, and enhances the stability and accuracy of interaction with the environment.
Smart Images

Figure CN114822761B_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the field of rehabilitation training and motor learning, and in particular relates to a wrist rehabilitation training system based on muscle synergy and variable stiffness impedance control. Background Art
[0002] Stroke is an acute cerebrovascular disease, a disease in which blood cannot flow into the brain due to sudden rupture or blockage of blood vessels in the brain, thus causing brain tissue damage. The "China Stroke Prevention and Treatment Report 2019" outlines that stroke is a major chronic disease that seriously endangers the health of the Chinese people. Stroke has a high incidence and disability rate. Among them, 70%-80% of patients will experience symptoms of hemiplegia, which makes patients lose the ability to carry out independent daily life activities, seriously affecting the living standards of patients, and may also cause various physiological and psychological diseases, and even endanger their lives. And with the continuous acceleration of the aging of the population and the urbanization process, as well as the unhealthy lifestyle of residents, the prevalence of stroke has continued to increase, bringing a serious burden to families and society.
[0003] At present, the plasticity of the human central nervous system has been verified. Scientific and regular rehabilitation training can help patients improve their damaged functions and restore the functions of their limbs. For stroke patients, motor function recovery usually requires long-term guidance and help from professional medical staff to complete rehabilitation training. However, due to the serious imbalance between the number of patients and the number of medical staff, a large number of patients cannot receive timely rehabilitation training, which affects the effect and timing of patient recovery. In addition, traditional rehabilitation training methods are mainly carried out by medical staff manually or with the help of simple machinery to drive the patient's limbs for treatment, and it is difficult to ensure the intensity and time of training. In recent years, the rapid development of sensor technology, human-machine interface technology, virtual reality technology and robotics technology has made up for the shortcomings of traditional rehabilitation training, and rehabilitation training can be carried out in a more precise and intelligent way. Exoskeleton equipment and robots use human motion signals (position signals, force signals, force-position mixed signals) or physiological signals (electromyographic signals, electroencephalographic signals, and electrooculographic signals) as interactive information, and conduct rehabilitation training on patients' limbs through various control algorithms. It records relevant information during the training process, which can be analyzed by professional therapists and timely adjust the rehabilitation training methods.
[0004] Surface electromyography is a bioelectric signal generated by the nerve center to control muscle contraction. Because the action information it contains is ahead of the corresponding human action and can better reflect the human action intention, it is often used as an interactive signal in various rehabilitation training systems. The neuromuscular synergy control theory explains the process from the central nervous system to the muscle. The central nervous system generates a small amount of signals, which are distributed to the corresponding muscles through the network. The muscles contract to generate force, thereby driving the movement of the limbs. Therefore, muscle synergy is used in training and rehabilitation to convert electromyography signals into motion signals.
[0005] As the most flexible part of the human body, the hand has the greatest impact on people's daily lives and is also the most vulnerable part. Among stroke patients, most patients will experience abnormal hand muscle strength and hand dysfunction. It is extremely important for the recovery of patients' hand functions. The rehabilitation training needs of the hands are diverse and the rehabilitation difficulty is relatively high. At present, passive rehabilitation training is mostly used, and the robot drives the patient to complete the preset trajectory, which cannot give play to the patient's initiative and enthusiasm. In addition, most rehabilitation training only focuses on the recovery of the patient's hand and wrist movements, but does not pay attention to the recovery of hand muscle strength. When interacting with the physical environment, it is necessary to effectively change the size of the muscle force to improve the stability and accuracy of the interaction with the environment. Summary of the invention
[0006] To solve the above problems, the present invention discloses a wrist rehabilitation training system based on muscle synergy and variable stiffness impedance control, which extracts joint angle and stiffness information from surface electromyography signals and applies them to rehabilitation training games, so that subjects can complete the training games, thereby achieving rehabilitation training effects on the wrist.
[0007] To achieve the above object, the technical solution of the present invention is as follows:
[0008] The wrist rehabilitation training system based on muscle synergy and variable stiffness impedance control includes:
[0009] The EMG signal acquisition and preprocessing module collects the surface EMG signals during exercise through the surface EMG signal sensor and performs preprocessing such as filtering;
[0010] The muscle synergy decomposition and mapping model acquisition module establishes the surface electromyography signal and joint angle mapping model and the surface electromyography signal and human body stiffness mapping model through the muscle synergy decomposition method and regression network model;
[0011] A human-computer interaction control module controls the agent in the virtual environment by means of impedance control, wherein the stiffness of the impedance control varies with the stiffness of the human arm;
[0012] The virtual reality serious game module generates serious game difficulty and virtual interference suitable for users' rehabilitation training and provides visual feedback.
[0013] Furthermore, the electromyographic signal acquisition and preprocessing module includes:
[0014] The surface electromyography signals of the human body's radial wrist flexor, radial wrist extensor, ulnar wrist flexor, ulnar wrist extensor, digital flexor and digital extensor were collected using Delsys.
[0015] The preprocessing of surface electromyography signal includes six steps: full-wave rectification, low-pass filtering, normalization, neural activation model extraction and muscle activation model extraction.
[0016] Furthermore, the muscle coordination decomposition and mapping model acquisition module includes:
[0017] Muscle synergy decomposition: The preprocessed signal is decomposed by non-negative matrix decomposition to obtain its spatial synergy and spatial synergy; the non-negative matrix decomposition method is as follows:
[0018] Z(t)=W·C(t)
[0019] Where Z(t) is the preprocessed surface electromyography signal, W is the spatial synergy, and C(t) is the spatial synergy. All elements in the non-negative matrix factorization method are non-negative values.
[0020] Obtain the joint angle mapping model, use the time domain collaborative data and motion label data to train the multi-layer perception model, and obtain the joint angle mapping model:
[0021] x(t)=MLP(W + ·Z(t)
[0022] Where x(t) is the joint angle, W + is the inverse matrix of time domain coordination, Z(t) is the preprocessed surface electromyography signal, and MLP(·) is the trained multi-layer perception model.
[0023] Obtain the human body stiffness mapping model, and use the synergy effect model to project the preprocessed surface electromyographic signals to the synergies in different directions with different degrees of freedom. These synergies are the contributions of muscle contraction in each direction, and then the model of surface electromyographic signals to human body stiffness is obtained by the co-contraction principle:
[0024] K i (t) = min(W i+ T Z(t), W i- T ·Z(t))
[0025] Where Ki (t) is the stiffness value of each degree of freedom, W i+ T , W i- T is the synergy coefficient matrix in the two directions of the degree of freedom, and Z(t) is the preprocessed surface electromyography signal.
[0026] Furthermore, the human-computer interaction control module uses a variable stiffness impedance control method to control the agent in the virtual environment. The virtual impedance control is a damping-spring-mass system, and the expression method of the system is:
[0027]
[0028] Where M is the mass of the object, B is the damping coefficient, K is the spring (stiffness) coefficient, x(t), They are the current position, current speed, current acceleration, F ext is the additional interference force. In this method, x(t) can be obtained from the model of the relationship between surface electromyographic signal and joint angle. They are the first-order derivative and second-order derivative of x(t), respectively; K can be obtained from the model of surface electromyographic signal to human body stiffness. The relationship between the damping coefficient and the spring (stiffness) coefficient of the system is: The mass of the object M is the mass of the target preset in the training game, and the additional interference force F ext A variable virtual force is set in the training game.
[0029] Further, the VR serious gaming module includes;
[0030] The serious game generation module is based on the pygame platform and generates different serious games for rehabilitation training for different training stages;
[0031] The feedback module is used to provide patients with game information such as interference in the virtual environment and the movement position of the virtual agent; the serious game adjustment module is used to adjust the difficulty of the serious game according to the effect of each round of rehabilitation training to increase the challenge of rehabilitation training.
[0032] The beneficial effects of the present invention are:
[0033] (1) The system uses a muscle-synergistic joint angle estimation method, which more intuitively reflects the relationship between joint movement and human body stiffness and muscle activation, and the mapping from electromyography to joint angle and stiffness is more accurate.
[0034] (2) The system adopts the virtual impedance control method to apply virtual force in the virtual scene, which can directly train muscle strength and achieve more efficient rehabilitation training.
[0035] (3) The rehabilitation training system adjusts the scene and complexity of the training game according to the training performance of the subject, which increases the fun of the rehabilitation training and can effectively improve the effect of the rehabilitation training. BRIEF DESCRIPTION OF THE DRAWINGS
[0036] Figure 1 It is a system framework diagram of the present invention. DETAILED DESCRIPTION
[0037] The present invention will be further explained below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the following specific embodiments are only used to illustrate the present invention and are not used to limit the scope of the present invention.
[0038] The rehabilitation training tasks in the present invention usually refer to the free movement of wrists and fingers in daily life, such as complex tasks such as picking up things and doing housework. Here, the therapist / technician tells the game goal, and the subject can control the movement of the target through finger and wrist movements to complete the task.
[0039] like Figure 1 As shown, the wrist rehabilitation training system based on muscle synergy and variable stiffness impedance control provided by the example of the present invention includes the following steps:
[0040] (1) Acquisition and preprocessing of surface electromyographic signals during wrist and finger flexion and extension
[0041] As the most commonly used interactive information in rehabilitation training systems, surface electromyography signals contain a large amount of action and stiffness information. This information can be decomposed through specific algorithms and used to control prostheses, robots or targets in virtual scenes. The Delsys wireless surface electromyography acquisition system is used to collect bioelectric signals generated during muscle activity. The subjects need to flex and extend the wrist and fingers respectively, and record the surface electromyography signals of the radial wrist flexor, radial wrist extensor, ulnar wrist flexor, ulnar wrist extensor, digital flexor and digital extensor. The collected surface electromyography signals are subjected to full-wave rectification and low-pass filtering to obtain the envelope of the electromyography signals, and then normalized so that the processed data is within the range of 0-1. Then the neural activation degree is obtained through the neural activation model, and finally the muscle activation degree is obtained through the muscle activation model.
[0042] (2) Determine the relationship model between surface electromyography signals and joint angles
[0043] According to the principle of muscle synergy decomposition, the preprocessed electromyographic signal is decomposed by non-negative matrix decomposition. The signals of wrist flexion and extension and finger flexion and extension are decomposed by muscle synergy respectively.
[0044] Z wf (t) = W wf ·C wf (t)
[0045] Z we (t) = W we ·C we (t)
[0046] Z ff (t) = W ff ·C ff (t)
[0047] Z fe (t) = W fe ·C fe (t)
[0048] Where Z wf (t), Z we (t), Z ff (t), Z fe (t) are the electromyographic signals when the wrist is bent, the wrist is extended, the fingers are bent, and the fingers are extended, respectively. wf , W we , W ff , W fe The spatial coordination corresponding to wrist flexion, wrist extension, finger flexion, and finger extension, C wf (t), C we (t), C ff (t), C fe (t) are the temporal coordination corresponding to wrist bending, wrist extension, finger bending, and finger extension respectively. The obtained spatial coordination can be combined by column to obtain the overall spatial coordination:
[0049] W=[W wf , W we , W ff , W fe ]
[0050] By training the multi-layer perception model regression network with the acquired time, coordination and label data, we can obtain the relationship model from electromyographic signals to joint angles:
[0051] x(t)=MLP(W + ·Z(t)
[0052] Where x(t) is the joint angle, W + is the inverse matrix of time domain coordination, Z(t) is the preprocessed surface electromyography signal, and MLP(·) is the trained multi-layer perception model.
[0053] (3) Determine the surface electromyography signal and human body stiffness model
[0054] According to the synergy effect model, the preprocessed surface electromyographic signal is projected to the synergy of different degrees of freedom and different directions. These synergies are the contributions of muscle contraction in each direction. The specific projection method is the same as the muscle synergy decomposition method in (2). Then, the model of surface electromyographic signal to human body stiffness is obtained based on the co-contraction principle:
[0055] K w (t) = min(W wf T Z(t), W we T ·Z(t))
[0056] K f (t) = min(W ff T Z(t), W fe T ·Z(t))
[0057] Where K w (t), K f (t) are the stiffness in the wrist flexion and extension direction and the stiffness in the finger flexion and extension direction, W wf T , W we T , W ff T , W fe T are the synergy coefficient matrices in the directions of wrist flexion, wrist extension, and finger flexion, respectively, and Z(t) is the preprocessed surface electromyography signal.
[0058] (4) Initialize the task control scenario and set the task complexity
[0059] In order to improve the subjects' enthusiasm for participating in rehabilitation training, different training game scenarios were used. Two games, Flying Bird and Captain Rogers, were used. The complexity of the game was changed by increasing or decreasing the number of obstacles in the game.
[0060] (5) Conduct rehabilitation training
[0061] Through a large amount of long-term training, the subjects can get effective rehabilitation. The control method of virtual impedance is adopted in the control:
[0062]
[0063] Where M is the mass of the object, B is the damping coefficient, K is the spring (stiffness) coefficient, x(t), They are the current position, current speed, current acceleration, F extThe joint angles obtained from the surface electromyography signal and the joint angle relationship model are mapped to the coordinates of the control target of the training game through projection transformation. The human body stiffness obtained from the surface electromyography signal and the human body stiffness model is used as the spring (stiffness) coefficient of the virtual impedance control method. The subjects need to control the actual output target to complete the relevant tasks.
[0064] (6) Adjust the mission scenario and mission complexity based on game performance
[0065] During rehabilitation training, the time and score of each task completed by the subject will be recorded. After each round of training, the game task scene and task complexity will be adjusted according to their performance. When the time to complete the game becomes shorter or the score becomes higher, the game difficulty will be increased, otherwise it will be reduced.
[0066] It should be noted that the above content only illustrates the technical idea of the present invention and cannot be used to limit the protection scope of the present invention. For ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications all fall within the protection scope of the claims of the present invention.
Claims
1. A wrist rehabilitation training system based on muscle synergy and variable stiffness impedance control, characterized by: include: The electromyographic signal acquisition and preprocessing module acquires the surface electromyographic signal during exercise through the surface electromyographic signal sensor and performs preprocessing; The muscle synergy decomposition and mapping model acquisition module establishes the surface electromyography signal and joint angle mapping model and the surface electromyography signal and human body stiffness mapping model through the muscle synergy decomposition method and regression network model; The muscle coordination decomposition and mapping model acquisition module includes: Muscle synergy decomposition: The preprocessed signal is decomposed by non-negative matrix decomposition to obtain its spatial synergy and spatial synergy; the non-negative matrix decomposition method is as follows: Z(t)=W·C(t) Where Z(t) is the preprocessed surface electromyography signal, W is the spatial domain synergy, and C(t) is the spatial domain synergy; all elements in the non-negative matrix decomposition method are non-negative values; Obtain the joint angle mapping model, use the time domain collaborative data and motion label data to train the multi-layer perception model, and obtain the joint angle mapping model: x(t)=MLP(W + ·Z(t)) Where x(t) is the joint angle, W + is the inverse matrix of time domain coordination, Z(t) is the preprocessed surface electromyography signal, and MLP(·) is the trained multi-layer perception model; Obtain the human body stiffness mapping model, and use the synergy effect model to project the preprocessed surface electromyographic signals to the synergies in different directions with different degrees of freedom. These synergies are the contributions of muscle contraction in each direction, and then the model of surface electromyographic signals to human body stiffness is obtained by the co-contraction principle: K i (t)=min(W i+ T ·Z(t),W i- T ·Z(t)) Where K i (t) is the stiffness value of each degree of freedom, W i+ T , W i- T is the synergy coefficient matrix in the two directions of the degree of freedom, Z(t) is the preprocessed surface electromyographic signal; A human-computer interaction control module controls the agent in the virtual environment by means of impedance control, wherein the stiffness of the impedance control varies with the stiffness of the human arm; The impedance control is a damping-spring-mass system, which can be expressed as: Where M is the mass of the object, B is the damping coefficient, K is the spring coefficient, x(t), They are the current position, current speed, current acceleration, F ext is the additional interference force; in this method, x(t) can be obtained from the model of the relationship between surface electromyographic signal and joint angle, are the first-order derivative and second-order derivative of x(t), respectively; K can be obtained from the model of surface electromyographic signal to human body stiffness, and the relationship between the damping coefficient and the spring coefficient of the system is: The mass of the object M is the mass of the target preset in the training game, and the additional interference force F ext To set a variable virtual force in a training game; The virtual reality serious game module generates serious game difficulty and virtual interference suitable for users' rehabilitation training and provides visual feedback.
2. The wrist rehabilitation training system based on muscle synergy and variable stiffness impedance control according to claim 1 is characterized in that: The electromyographic signal acquisition and preprocessing module comprises: The Delsys wireless surface electromyography acquisition device is used to collect surface electromyography signals of the radial wrist flexor, radial wrist extensor, ulnar wrist flexor, ulnar wrist extensor, digital flexor, and digital extensor of the human body; The preprocessing of surface electromyography signal includes five steps: full-wave rectification, low-pass filtering, normalization, nerve activation extraction and muscle activation extraction.
3. The wrist rehabilitation training system based on muscle synergy and variable stiffness impedance control according to claim 1, characterized in that: The virtual reality serious game module includes: The serious game generation module is based on the pygame platform and generates different serious games for rehabilitation training for different training stages; A feedback module for providing the patient with distractions in the virtual environment and game information about the movement position of the virtual agent; The serious game adjustment module is used to adjust the difficulty of the serious game according to the effect of each round of rehabilitation training to increase the challenge of rehabilitation training.
4. The method for using the wrist rehabilitation training system based on muscle synergy and variable stiffness impedance control according to claim 1, characterized in that: Firstly, the surface electromyographic signals of the patient's advantageous limb during wrist and finger flexion and extension are collected and preprocessed. Then, the relationship models between surface electromyographic signals and joint angles and surface electromyographic signals and human body stiffness are determined through the muscle synergy method. Then, the serious game scene and difficulty are set. Then, the surface electromyographic signals of the patient's unadvantaged limb are collected and preprocessed. The joint angles and human body stiffness are obtained from the relationship models between surface electromyographic signals and joint angles and surface electromyographic signals and human body stiffness. Finally, the variable stiffness impedance control method is used to control the agent in the virtual environment to complete the rehabilitation training task.
