PID (Proportion Integration Differentiation) electrical stimulation balance rehabilitation method based on injury synergistic effect

Abstract

The invention discloses a PID (Proportion Integration Differentiation) electrical stimulation balance rehabilitation method based on a damage synergistic effect. According to the method, firstly, myoelectricity and IMU joint movement data of a healthy subject on a disturbance platform are collected, non-negative matrix factorization (NMF) is conducted on the myoelectricity data, and the muscle synergistic effect of the healthy subject under a disturbance normal form is solved; myoelectricity and IMU signals of stroke patients in the same disturbance normal form are collected, nonnegative matrix factorization is carried out on the myoelectricity of the patients, and similarity analysis is carried out on the health synergistic effect and the patient synergistic effect to determine the injury synergistic effect. The injury synergistic effect is applied to adjustment of electrical stimulation pulse width parameters and used for conducting electrical stimulation on lower limb muscles, joint angle data are collected in real time, PID electrical stimulation control based on the injury synergistic effect is achieved, and therefore a patient can track a joint angle track of a healthy subject under a disturbance normal form. The patient is assisted to recover the injury synergistic effect under the disturbance normal form, the balance capacity is improved, and a new path is developed for rehabilitation treatment of the stroke patient.

Description

Technical Field

[0001] The present invention relates to the technical field of motion analysis and rehabilitation engineering, and relates to a stroke lower limb perturbation rehabilitation training method based on multi-channel electromyography and functional electrical stimulation of the lower limbs. Specifically, it relates to a PID electrical stimulation balance rehabilitation method based on injury synergy. Background Art

[0002] Stroke, also known as apoplexy, is a sudden cerebrovascular disease. After a stroke, patients usually experience varying degrees of motor impairment, resulting in the brain's inability to normally control the muscle movement functions of the limbs. This damage interrupts the neural connections between the brain and the limbs, seriously affecting the patient's daily activities. Hemiplegia is one of the most common functional impairments after a stroke, especially prominent in the lower limbs. Common symptoms include weakened muscle strength, muscle control disorders, poor balance stability, and a particularly high risk of falls. The balance ability of the lower limbs is of fundamental significance for humans to complete activities such as social interaction and self-care. Limited balance not only reduces the patient's quality of life but also becomes a major obstacle to returning to social and family life. Therefore, restoring the balance ability of the lower limbs is one of the important goals of stroke rehabilitation and plays a crucial role in the reconstruction of the patient's social and family life.

[0003] Muscle Synergy refers to the cooperation of multiple muscle groups to complete specific movement tasks. This coordination is controlled by the nervous system with the aim of simplifying the complexity of movement control and achieving efficient and smooth movements. In human movement, a single movement is usually not completed by a single muscle but by the joint action of multiple muscles. Muscle synergy can reduce the brain's individual control of each muscle, making it easier to coordinate complex movement tasks. For example, during walking, multiple muscles in the thigh, hip, calf, etc. need to cooperate precisely to maintain body balance and propulsion. These muscles complete a series of fine movements, such as stepping, pushing off the ground, and supporting, through synergy at different stages of movement.

[0004] The study of muscle synergy is of great significance in the field of motor control and rehabilitation. Through the analysis of synergy, researchers can better understand how the brain achieves precise movement output by simplifying motor control. In the process of motor control, it is not necessary to control each muscle individually, but to efficiently coordinate the actions of multiple muscle groups by activating synergy patterns. This mechanism not only reduces the burden on the brain, but also improves the efficiency and stability of movement. In neurorehabilitation, the study of muscle synergy helps to reveal the pathological mechanism of movement disorders after central nervous system damage (such as stroke). Patients may not be able to activate muscle synergy patterns normally after injury, resulting in movement disorders or uncoordinated movements. Therefore, through rehabilitation training, rebuilding or optimizing muscle synergy can help patients restore normal motor function, especially playing a key role in daily activities such as walking and grasping.

[0005] Functional electrical stimulation (FES) is a technology that uses electrical stimulation of peripheral nerves or muscles to restore or enhance impaired motor function. FES is primarily used to help patients with motor dysfunction caused by central nervous system damage. It uses electrical current to directly stimulate paralyzed or weakened muscles, causing them to contract functionally, thereby restoring specific movements such as walking, grasping, and standing. FES works by placing electrodes near muscles or nerves to deliver appropriate electrical stimulation signals to the nerves or muscles, simulating the natural electrical signals sent by the brain to the muscles. This stimulation triggers muscle contraction, causing them to perform certain movements. Its goal is not only to achieve short-term motor recovery, but also to help patients reshape the neuromuscular connection through repeated electrical stimulation training, thereby achieving long-term functional improvement. FES not only compensates for lost muscle motor function but also stimulates neuroplasticity. In rehabilitation training, FES can be used in conjunction with traditional physical therapy to help patients recover faster and better.

[0006] Research on functional electrical stimulation (FES) based on muscle synergy has become a hot topic in the field of neurorehabilitation in recent years, particularly in gait recovery and cycling training paradigms. Traditional FES technology typically uses electrodes to stimulate a single muscle or muscle group to achieve a specific movement. However, this model does not fully reflect the complex movement control process of the human body. Muscle synergy theory proposes that the central nervous system achieves efficient movement control by activating a group of muscles working together. Summary of the Invention

[0007] Based on the technical problems involved in the background art, the present invention proposes a PID electrical stimulation balance rehabilitation strategy based on the synergistic effect of injury. The present invention first collects the electromyography and IMU joint motion data of healthy subjects and patients on a perturbation platform, solves the muscle synergy through non-negative matrix factorization (NMF), and determines the injury synergy by performing similarity analysis on the healthy synergy and the patient synergy. The injury synergy is applied to the electrical stimulation parameters to electrically stimulate the lower limb muscles, and the joint angle data is collected in real time to implement the PID electrical stimulation online control strategy based on the injury synergy.

[0008] The present invention is implemented by adopting the following technical solutions:

[0009] A PID electrical stimulation rehabilitation training method based on the injury synergy, based on the following hardware devices:

[0010] Rehabilitation training device: A balance board for performing lower limb perturbation training on the subject;

[0011] IMU acquisition module: Used to collect the motion data of the subject during the perturbation rehabilitation training;

[0012] Surface electromyography acquisition module: Used to collect the multi-channel electromyography signals (EMG) of the lower limbs of the subject during the rehabilitation training;

[0013] Muscle synergy analysis module: Used to perform muscle synergy analysis on the multi-channel electromyography signals (EMG) of the lower limbs, and further evaluate the balance function status of stroke patients;

[0014] PID control module: Perform PID electrical stimulation control based on the synergistic effect of injury.

[0015] The PID electrical stimulation rehabilitation training strategy based on the synergistic injury under the condition of balance perturbation includes the following specific steps:

[0016] (1) Template construction: The healthy subject stands on the balance board and performs stable and regular perturbations in the front-back direction. The surface electromyography measurement unit is respectively placed on multiple lower limb muscles to measure the surface electromyography signals of the multi-channel muscles of the lower limbs under the perturbation condition. The inertial measurement unit is respectively fixed on the hip joint, knee joint, and ankle joint of the healthy subject to measure the joint angle data of the limb under the perturbation condition.

[0017] For the lower limb surface electromyography data of the healthy subject, the baseline drift is removed, rectified, filtered, and downsampled from the multi-channel electromyography signals to obtain the electromyography signal envelope of the multi-channel electromyography signals of the lower limbs, and the normal synergy of the healthy subject is solved through NMF non-negative matrix factorization; for the lower limb joint angle data of the healthy subject, the baseline drift is removed, filtered, and root mean square processed to obtain the lower limb joint angle motion curve template;

[0018] (2) Offline data processing and injury synergy determination

[0019] The stroke patient stands on a balance board and undergoes the same perturbation training paradigm as healthy subjects, and the surface electromyogram signals of multiple lower limb muscles are measured under perturbation conditions.

[0020] For the lower limb surface electromyogram data of stroke patients, the baseline drift, rectification, filtering, and downsampling are respectively performed on the multiple lower limb electromyogram signals to obtain the electromyogram signal envelope of the multiple lower limb electromyogram signals. The abnormal synergy of stroke patients is solved by non - negative matrix factorization (NMF), and the synergy between healthy subjects and patients is compared by cosine similarity.

[0021] Injury synergy is defined as the synergy that is not manifested in patients in healthy synergy. That is, for each normal synergy of healthy subjects, there is no synergy in the patient's synergy with a cosine similarity value greater than 0.8. The muscle synergies of healthy subjects are W1, W2, W3, W4, and the muscle synergies of patients are W1, W2, W3. The cosine similarity values between the muscle synergies W1, W2 of healthy subjects and the muscle synergies W1, W2 of patients are greater than 0.8, the cosine similarity value between the healthy synergy W3 and the patient's synergy W3 is less than 0.8, and the healthy synergy W4 is not manifested in patients. Those used for rehabilitation strategies by functional electrical stimulation are Synergy3 and Synergy4, that is

[0022] X = W1×H1 + W2×H2 + W3×H3 + W4×H4

[0023] Synergy3 ≈ W3×H3

[0024] Synergy4 ≈ W4×H4

[0025] Among them, X is the envelope matrix of the original electromyogram signal, and Synergy3 and Synergy4 are injury synergies.

[0026] For two n - dimensional vectors a and b, the cosine similarity between the two variables is defined as the cosine value of the angle between the two variables.

[0027] a=(x 11 ,x 12 ......,x 1n )

[0028] b=(x 21 ,x 22 ......,x 2n )

[0029]

[0030] Where a·b is the inner product of variables a and b, and |a| and |b| are the norms of variables a and b.

[0031] For the lower limb joint angle data of healthy subjects, baseline drift removal, filtering, and root mean square processing are performed to obtain the lower limb joint angle motion curve template.

[0032] (3) Patient perturbation training and PID electrical stimulation online control adjustment

[0033] Based on the patient injury synergy determined in step (2), the injury synergy effect is applied to the PID electrical stimulation control strategy of the lower limb muscles.

[0034] The stroke patient still stands on the balance board and performs the same perturbation training paradigm. The lower limb joint angle data is transmitted to the upper computer in real time. Real-time data preprocessing is performed in the upper computer. Based on the comparison between the patient's real-time joint angle data and the joint angle template of healthy subjects, the real-time angle deviation is obtained. Based on the injury synergy, the PID electrical stimulation control is completed in real time on the upper computer. When the real-time joint angle deviates positively from the angle template, the injury synergy Synergy3 is used as the basic unit for dynamic adjustment on the time axis to change the electrical stimulation intensity. When the real-time joint angle deviates negatively from the angle template, the injury synergy Synergy4 is used as the basic unit for dynamic adjustment on the time axis to change the electrical stimulation intensity. The mapping strategy is as Figure 2 shown. It can achieve PID control with the injury synergy combination as the basic unit according to the real-time joint angle deviation, and can enable the real-time joint angle of the patient to track the trajectory of the joint angle template of healthy subjects.

[0035] Ei = 1,2,...n = W3×H3 + W4×H4

[0036]

[0037] Where E is the envelope matrix determined according to the injury synergy, i refers to the muscle, n is the number of muscles, PW i (t) is the pulse width of the electrical stimulation of the i-th muscle at time t, and E i (t) is the size of the muscle envelope of the i-th muscle at time t, and MaxE i is the maximum value of the i-th muscle envelope.

[0038] The PID electrical stimulation strategy modulates the electrical stimulation pulse width between 0 and a maximum value of 400 μs. During the initial calibration process, a pulse width of 400 μs was used to individually determine the electrical stimulation amplitude for each muscle to produce a visibly obvious contraction without causing discomfort. Based on the determination of the electrical stimulation frequency and amplitude, the maximum pulse width of 400 μs is linearly mapped to the maximum pulse width parameter limit of the lower limb muscle electrical stimulation parameters in the injury coordination time series, and PID electrical stimulation control is carried out with injury coordination as the basic unit to dynamically adjust the injury coordination.

[0039] The core principle is: based on the injury muscle coordination and the joint angle template of healthy subjects, and by real-time reading the joint angle, PID electrical stimulation control is carried out with injury coordination as the unit. By minimizing the control error, strictly controlling the overshoot, and shortening the system stabilization time, more precise and efficient functional electrical stimulation is achieved.

[0040] PID control, that is, proportional-integral-derivative control, is a feedback-based control method. It forms a control deviation e(t) = r(t) - y(t) according to the joint angle template r(t) and the actual joint angle y(t), and calculates the control quantity u(k) using the proportional, integral, and derivative three links to control the electrical stimulation parameters. Its control law formula is:

[0041]

[0042] where K p is the proportionality coefficient, K i is the integral coefficient, K d is the derivative coefficient. The functions of each link are as follows:

[0043] Proportional (P) control: The output of the controller is in a certain proportional relationship with the input error signal, that is, u p (k) = K p e(t). Once the deviation occurs, the controller will immediately have a control effect. The larger the proportionality coefficient K p , the faster the response speed.

[0044] Integral (I) control: The output of the controller is proportional to the integral of the input error signal, that is, The integral term will accumulate over time. When there is a steady-state error after the system enters the steady state, the integral term can push the output of the controller to increase and further reduce the steady-state error until the steady-state error is eliminated.

[0045] Derivative (D) control: The output of the controller is proportional to the rate of change of the input error signal, that is, The derivative link can predict the trend of deviation change, has an anticipatory control effect, can improve the stability of the system, and reduce the overshoot and oscillation.

[0046] (4) Muscle synergy control theory

[0047] The most commonly used method to solve muscle synergy is non - negative matrix factorization (NMF), that is, decomposing X into two non - negative matrices W and H.

[0048] Muscle synergy data information is obtained by non - negative matrix factorization of the electromyogram envelope signal. The weights and activation coefficients in the described multi - channel electromyogram data information are generated by the following formula:

[0049] X≈W×H

[0050] where X∈R m×T is the original data matrix of the electromyogram signal, m is the number of channels of the electromyogram signal, and T is the number of sampling points in the entire movement process. W is the muscle weight contribution matrix, and H is the weight coefficient matrix, that is, the activation coefficient matrix.

[0051]

[0052] In the muscle synergy model, the number K of muscle synergies is uncertain. The muscle synergy models X reconstructed by the matrices W and H decomposed under different numbers K of synergies m×n =W m×K H K×T have different degrees of similarity. The reconstruction similarity degree is measured by the VAF parameter, and its definition is:

[0053]

[0054] If the VAF value is close to 1, it means that the synergy pattern after NMF decomposition can well explain the original data, that is, the reconstructed matrix X r is very close to the original matrix X, indicating a high accuracy of the synergy decomposition; if the VAF value is low (close to 0), it means that there is a large difference between the reconstructed data and the original data, indicating that the decomposed synergy pattern is not sufficient to explain most of the variance in the original data, and the decomposition effect is poor.

[0055] (5) Functional electrical stimulation parameter setting

[0056] Based on the normal synergy of healthy subjects and the abnormal synergy of patients, cosine similarity comparative analysis is carried out to determine the damaged muscle synergy of each patient. The damaged muscle synergy is used for the pulse width parameter of functional electrical stimulation. The real - time deviation angle is calculated from the real - time joint angle data and the joint angle response template of healthy subjects, and used as the control quantity for PID electrical stimulation control based on the damaged synergy. The controller adjusts the pulse width of the electrical stimulation mode in real time to track the joint angle template of healthy subjects, simulate normal voluntary movement, and re - learn the control strategy under the perturbation task.

[0057] PID electrical stimulation balance rehabilitation training strategy, wherein the frequency of the functional electrical stimulation parameters is 15 - 50 Hz; the pulse width of the functional electrical stimulation parameters is 100 - 1000 us; the duty cycle of the functional electrical stimulation parameters is between 1:1 and 1:3; the rise / fall time of the functional electrical stimulation parameters is 1 - 2 s; the current intensity of the functional electrical stimulation parameters is 0 mA - 100 mA.

[0058] Beneficial effects

[0059] 1. The present invention fills the blank in the research on the lower limb balance muscle response pattern of stroke patients, innovates the functional electrical stimulation control strategy, and according to the personalized co - injury characteristics of patients, improves the balance function state of patients and re - learns the ideal balance response strategy.

[0060] 2. The movement functional electrical stimulation technology of the present invention is easy to implement. By collecting multi - channel muscle signals and movement data during the lower limb perturbation rehabilitation training of stroke patients, and using non - negative matrix factorization (NMF) analysis, the cooperation pattern between multiple lower limb muscles is determined.

[0061] 3. By comparing the muscle cooperation response patterns between healthy people and patients, the damaged cooperation of patients is determined, and the damaged muscle cooperation is reconstructed by PID - controlled functional electrical stimulation, improving the incorrect movement response pattern of patients and re - learning the correct movement response pattern.

[0062] 4. Introducing the muscle cooperation model into FES can be more in line with the natural movement control mode of the human body and improve the rehabilitation effect. Brief description of the drawings

[0063] Figure 1 Flow chart of the injury - coordinated electrical stimulation technology;

[0064] Figure 2 Schematic diagram of the injury - coordinated electrical stimulation strategy;

[0065] Figure 3 Schematic diagram of the lower limb perturbation training;

[0066] Figure 4 Schematic diagram of the construction of the electrical stimulation online system;

[0067] Figure 5 Functional electrical stimulation parameter diagram. Detailed implementation manners

[0068] The following further describes the present invention with reference to the accompanying drawings.

[0069] Existing functional electrical stimulation strategies are mainly used to improve motor function under specific tasks and cannot re - learn the correct neural control mechanism. The PID electrical stimulation balance rehabilitation strategy based on injury - coordinated action, such as Figure 1The figure shows the technical flowchart of injury synergistic electrical stimulation. First, the present invention determines the synergy between healthy subjects and patients on the perturbation platform, and conducts similarity analysis between the two to determine the injury synergy. Based on the injury synergy, electrical stimulation parameters are applied to electrically stimulate the lower limb muscles, and joint angle data is collected in real time to achieve PID electrical stimulation control, so as to enable the patient to track the joint angle trajectory of the healthy subject under the perturbation paradigm, help the patient recover the injury synergy under the perturbation paradigm, and relearn the muscle control strategy under the perturbation task.

[0070] In view of the problems existing in the prior art, the present invention adopts the following technical solutions for implementation:

[0071] A PID electrical stimulation control strategy based on injury synergy, the method is based on the following modules:

[0072] Rehabilitation training equipment: A balance board for performing lower limb perturbation training on the subject;

[0073] IMU acquisition module: Used to acquire motion data during the perturbation rehabilitation training of the subject;

[0074] Surface electromyogram acquisition module: Used to acquire multi-channel electromyogram signals EMG of the lower limbs during the rehabilitation training of the subject;

[0075] PID control module: Based on the synergy of injury, perform PID electrical stimulation control.

[0076] Implement an electrical stimulation strategy based on collaborative injury under balanced perturbation, including the following specific steps:

[0077] (1) Template construction

[0078] After cleaning the skin surface, the healthy subject stands on the balance board and is perturbed stably and regularly in the front-back direction. The subject needs to maintain his own balance, as Figure 3 shown. The surface electromyogram measurement unit is respectively placed on multiple lower limb muscles to measure the surface electromyogram signals of multiple lower limb muscles under perturbation conditions, and fixed to ensure accurate data acquisition; the inertial measurement unit is respectively fixed on the hip joint, knee joint, and ankle joint of the healthy subject to measure the joint angle data of the limb under perturbation conditions.

[0079] For the lower limb surface electromyogram data of the healthy subject, the baseline drift, rectification, filtering, and downsampling are respectively performed on the multi-channel electromyogram signals to obtain the electromyogram signal envelope of the multi-channel electromyogram signals of the lower limbs, and the normal synergy of the healthy subject is solved by NMF non-negative matrix factorization; for the lower limb joint angle data of the healthy subject, baseline drift removal, filtering, and root mean square processing are performed to obtain the lower limb joint angle motion curve;

[0080] (2) Offline Data Processing and Injury Synergy Determination

[0081] The stroke patients stand on a balance board and perform the same perturbation training paradigm as healthy subjects, and the surface electromyogram signals of multiple lower limb muscles are measured under perturbation conditions.

[0082] For the surface electromyogram data of the lower limbs of stroke patients, the baseline drift, rectification, filtering, and downsampling are respectively performed on the surface electromyogram signals of multiple lower limb muscles to obtain the electromyogram signal envelope of the surface electromyogram signals of multiple lower limb muscles. The abnormal synergy of stroke patients is solved by non-negative matrix factorization (NMF), and the synergy between healthy subjects and patients is compared by cosine similarity.

[0083] Injury synergy is defined as the synergy that is not manifested in patients in healthy synergy, that is, for each healthy synergy, there is no synergy in the patient synergy with a cosine similarity value greater than 0.8. The muscle synergies of healthy subjects are W1, W2, W3, W4, and the muscle synergies of patients are W1, W2, W3. The cosine similarity values between the muscle synergies W1, W2 of healthy subjects and the muscle synergies W1, W2 of patients are greater than 0.8, the cosine similarity value between the healthy synergy W3 and the patient synergy W3 is less than 0.8, and the healthy synergy W4 is not manifested in patients. Those used for functional electrical stimulation in rehabilitation strategies are Synergy3 and Synergy4, that is

[0084] X = W1×H1 + W2×H2 + W3×H3 + W4×H4

[0085] Synergy3 ≈ W3×H3

[0086] Synergy4 ≈ W4×H4

[0087] Among them, X is the envelope matrix of the original electromyogram signal, and Synergy3 and Synergy4 are injury synergies.

[0088] For two n-dimensional vectors a and b, the cosine similarity between the two variables is defined as the cosine value of the angle between the two variables.

[0089] a = (x 11 , x 12 ......, x 1n )

[0090] b = (x 21 , x 22 ......, x 2n )

[0091]

[0092] where a·b is the inner product of variables a and b, and |a| and |b| are the norms of variables a and b.

[0093] For the lower limb joint angle data of healthy subjects, baseline drift removal, filtering, and root mean square processing are performed to obtain the lower limb joint angle motion curve;

[0094] (3) Patient perturbation training and PID electrical stimulation online control adjustment

[0095] Based on the patient injury synergy determined in step (2), the injury synergy effect is applied to the PID electrical stimulation control strategy of the lower limb muscles;

[0096] The stroke patient still stands on the balance board and performs the same perturbation training paradigm. The lower limb joint angle data is transmitted to the host computer in real time. Real-time data preprocessing is performed in the host computer. Based on the comparison between the patient's real-time joint angle data and the joint angle template of healthy subjects, the real-time angle deviation is obtained. The PID electrical stimulation control based on injury synergy is completed in real time by the host computer. The electrical stimulation online control is as Figure 4 shown.

[0097] Based on the determined W3 and W4 of the patient injury synergy, when the real-time joint angle is positively deviated from the angle template, the injury synergy Synergy3 is used as the basic unit for dynamic adjustment on the time axis to change the electrical stimulation intensity. When the real-time joint angle is negatively deviated from the angle template, the injury synergy Synergy4 is used as the basic unit for dynamic adjustment on the time axis to change the electrical stimulation intensity. The mapping strategy is as Figure 2 shown. It can enable the trajectory tracking of the patient's real-time joint angle and the joint angle template of healthy subjects.

[0098] Ei = 1,2,...n = W3×H3 + W4×H4

[0099]

[0100] where E is the envelope matrix determined according to the injury synergy, i refers to the muscle, n is the number of muscles, PW i (t) is the pulse width of the electrical stimulation of the i-th muscle at time t, and E i (t) is the size of the muscle envelope of the i-th muscle at time t, and MaxE i is the maximum value of the i-th muscle envelope.

[0101] The PID electrical stimulation control strategy modulates the electrical stimulation pulse width between 0 and a maximum value of 400 μs. During the initial calibration process, a pulse width of 400 μs was used to individually determine the stimulation amplitude of each muscle to produce a visibly obvious contraction without causing discomfort. On the basis of determining the electrical stimulation frequency and amplitude, the maximum pulse width of 400 μs was linearly mapped to the maximum pulse width parameter limit of the lower limb muscle electrical stimulation parameters on the injury coordination time series, and PID electrical stimulation control was carried out with injury coordination as the basic unit.

[0102] The core principle is: Based on the injury muscle coordination and the joint angle template of healthy subjects, and by real-time reading the joint angle, PID electrical stimulation control is carried out with injury coordination as the unit. By minimizing the control error, strictly controlling the overshoot, and shortening the system stabilization time, more precise and efficient functional electrical stimulation is achieved.

[0103] PID control, that is, proportional-integral-derivative control, is a feedback-based control method. It forms a control deviation e(t) = r(t) - y(t) according to the joint angle template r(t) and the actual joint angle y(t), and calculates the control quantity u(k) using the proportional, integral, and derivative three links to control the electrical stimulation parameters. Its control law formula is:

[0104]

[0105] where K p is the proportionality coefficient, K i is the integral coefficient, K d is the derivative coefficient.

[0106] The functions of each link are as follows:

[0107] Proportional (P) control: The output of the controller is in a certain proportional relationship with the input error signal, that is, u p (k) = K p e(t). Once the deviation occurs, the controller will immediately have a control effect. The larger the proportionality coefficient K p , the faster the response speed.

[0108] Integral (I) control: The output of the controller is proportional to the integral of the input error signal, that is, The integral term will accumulate over time. When there is a steady-state error after the system enters the steady state, the integral term can push the output of the controller to increase and further reduce the steady-state error until the steady-state error is eliminated.

[0109] Derivative (D) control: The output of the controller is proportional to the rate of change of the input error signal, that is, The differentiator can predict the trend of deviation changes, has an anticipatory control effect, can improve the stability of the system, and reduce the overshoot and oscillation.

[0110] (4) Muscle Synergy Control Theory

[0111] The most commonly used method to solve muscle synergy is non - negative matrix factorization (VAF), that is, decomposing X into two non - negative matrices W and H.

[0112] Obtaining muscle synergy data information through non - negative matrix factorization of the electromyogram envelope signal, the weights and activation coefficients in the described multi - channel electromyogram data information are generated by the following formula:

[0113] X≈W×H

[0114] where X∈R m×T is the original data matrix of the electromyogram signal, m is the number of channels of the electromyogram signal, and n is the number of sampling points in the entire movement process. W is the muscle weight contribution matrix, and H is the weight coefficient matrix, that is, the activation coefficient matrix.

[0115]

[0116] In the muscle synergy model, the number K of muscle synergies is uncertain. The muscle synergy models X reconstructed by the matrices W and H decomposed under different numbers K of synergies m×n =W m×K H K×T have different degrees of similarity. The reconstruction similarity is measured by the VAF parameter, and its definition is:

[0117]

[0118] If the VAF value is close to 1, it means that the synergy pattern after NMF decomposition can well explain the original data, that is, the reconstructed matrix X r is very close to the original matrix X, indicating a high accuracy of the synergy decomposition; if the VAF value is low (close to 0), it means that the difference between the reconstructed data and the original data is large, indicating that the decomposed synergy pattern is not sufficient to explain most of the variance in the original data, and the decomposition effect is poor.

[0119] (5) Functional Electrical Stimulation Parameter Setting

[0120] Based on the similarity comparison analysis of the normal synergy of healthy subjects and the abnormal synergy of patients, the damaged muscle synergy of each patient is determined. The damaged muscle synergy is used for the pulse width parameter of functional electrical stimulation. The real-time deviation angle is calculated from the real-time joint angle data and the joint angle response template of healthy subjects and used as the control quantity to be input into the PID controller. The controller adjusts the intensity of the electrical stimulation mode in real time to track the joint angle response template, simulate normal voluntary movement, and relearn the control strategy under perturbation tasks. Some existing rehabilitation application cases are shown in Table 1.

[0121] For the PID electrical stimulation balance rehabilitation training strategy, the application frequency of the functional electrical stimulation parameters is 15 - 50 Hz; the pulse width of the functional electrical stimulation parameters is 100 - 1000 us; the duty cycle of the functional electrical stimulation parameters is between 1:1 and 1:3; the rise / fall time of the functional electrical stimulation parameters is 1 - 2 s; the current intensity of the functional electrical stimulation parameters is 0 mA - 100 mA. The electrical stimulation parameters are as Figure 5 shown.

[0122] Table 1

[0123]

[0124]

Claims

1. A PID electrical stimulation balance rehabilitation method based on damage synergy, characterized in that, It mainly includes the following steps: First, collect the electromyography and IMU joint movement data of healthy subjects and patients on the perturbation platform; Solve the muscle synergy by non - negative matrix factorization (NMF), and perform similarity analysis between the healthy synergy and the patient synergy to determine the damaged synergy; Apply the damaged synergy to the electrical stimulation parameters to electrically stimulate the lower limb muscles, and collect joint angle data in real time to achieve PID electrical stimulation control based on the damaged synergy, so as to realize the trajectory tracking of the joint angles of healthy subjects by patients under the perturbation paradigm, help patients recover the damaged synergy, and relearn the muscle control strategy under the perturbation task.

2. The PID electrical stimulation balance rehabilitation method based on injury synergy according to claim 1, wherein The method is based on the following hardware devices: Rehabilitation training equipment: A balance board used for lower limb perturbation training of subjects; IMU acquisition module: Used to collect movement data during the balance perturbation rehabilitation training of subjects; Surface electromyography acquisition module: Used to collect multi - channel electromyography signals (EMG) of the lower limbs during the rehabilitation training of subjects; Muscle synergy analysis module: Used to perform muscle synergy analysis on the multi - channel electromyography signals (EMG) of the lower limbs, and further evaluate the balance function status of stroke patients; PID control module: Based on the damaged synergy, perform PID electrical stimulation parameter control; The electrical stimulation strategy based on collaborative damage under balance perturbation conditions includes the following specific steps: (1) Template construction: The healthy subject stands on the balance board and is perturbed stably and regularly in the front - to - back direction. The surface electromyography measurement unit is placed on multiple lower limb muscles respectively to measure the surface electromyography signals of the multi - channel muscles of the lower limbs under the perturbation condition. The inertial measurement unit is fixed to the hip joint, knee joint, and ankle joint of the healthy subject respectively to measure the joint angle data of the limb under the perturbation condition; For the lower limb surface electromyography data of healthy subjects, remove the baseline drift, rectify, filter, and down - sample the multi - channel electromyography signals respectively to obtain the electromyography signal envelope of the multi - channel electromyography signals of the lower limbs, and solve the normal synergy of healthy subjects through NMF non - negative matrix factorization; perform baseline drift removal, filtering, and root - mean - square processing on the lower limb joint angle data of healthy subjects to obtain the lower limb joint angle movement curve template; (2) Off - line data processing and determination of damaged synergy The stroke patient stands on the balance board and performs the same perturbation training paradigm as the healthy subject, and measures the surface electromyography signals of the multi - channel muscles of the lower limbs under the perturbation condition; For the lower limb surface electromyography data of stroke patients, remove the baseline drift, rectify, filter, and down - sample the multi - channel electromyography signals respectively to obtain the electromyography signal envelope of the multi - channel electromyography signals of the lower limbs, solve the abnormal synergy of stroke patients through NMF non - negative matrix factorization, and compare the synergy between healthy subjects and patients through cosine similarity; The damaged synergy is defined as that for each normal synergy of a healthy subject, there is no synergy in the patient synergy with a cosine similarity greater than 0.8; (3) Patient perturbation training and online control adjustment of PID electrical stimulation Based on the patient's damaged synergy determined in step (2), use the damaged synergy as the basic unit for the electrical stimulation control strategy of the lower limb muscles; The stroke patient still stands on the balance board and performs the same perturbation training paradigm. The lower limb joint angle data is transmitted to the host computer in real time, and real-time data preprocessing is performed in the host computer. Based on the comparison between the patient's real-time joint angle data and the joint angle template of healthy subjects, the real-time angle deviation is obtained. Through the host computer, the PID electrical stimulation control based on damage coordination is completed in real time to achieve the trajectory tracking of the patient's real-time joint angle and the joint angle template of healthy subjects. (4) Functional electrical stimulation parameter setting Based on the similarity comparison analysis of the normal synergy of healthy subjects and the abnormal synergy of patients, the damaged muscle synergy of each patient is determined. The damaged muscle synergy is used for the pulse width parameter of functional electrical stimulation. The real-time deviation angle is calculated from the real-time joint angle data and the joint angle response template of healthy subjects and input into the PID controller as the control quantity. The controller adjusts the intensity of electrical stimulation in real time to track the corresponding joint angle template, simulate normal voluntary movement, and re-learn the control strategy under the perturbation task.

3. The PID electrical stimulation balance rehabilitation method based on damage synergy according to claim 1, wherein The frequency of the functional electrical stimulation parameter is 15 - 50 Hz.

4. The PID electrical stimulation balance rehabilitation method based on injury synergy according to claim 1, wherein, The pulse width of the functional electrical stimulation parameter is 100 - 1000 us.

5. The PID electrical stimulation balance rehabilitation method based on injury synergy according to claim 1, characterized in that The duty cycle of the functional electrical stimulation parameter is between 1:1 and 1:3; the rise / fall time of the functional electrical stimulation parameter is 1 - 2 s; the current intensity of the functional electrical stimulation parameter is 0 mA - 100 mA.

6. The PID electrical stimulation balance rehabilitation method based on injury synergy according to claim 1, wherein, Non-negative matrix factorization is used to solve muscle synergy, that is, X is decomposed into two non-negative matrices W and H. The non-negative matrix factorization of the electromyogram envelope signal obtains muscle synergy data information. The weights and activation coefficients in the described multi-channel electromyogram data information are generated by the following formula: X≈W×H where X ∈ R m×T is the original data matrix of the electromyogram signal, m is the number of channels of the electromyogram signal, and T is the number of sampling points in the entire movement process; W is the muscle weight contribution matrix, and H is the weight coefficient matrix, that is, the activation coefficient matrix; In the muscle synergy model, the number K of muscle synergies is uncertain, and the muscle synergy model X reconstructed by the matrices W and H decomposed under different numbers K of synergies m×n = W m×K H K×T has different degrees of similarity. The degree of reconstruction similarity is measured by the VAF parameter, which is defined as: If the VAF value is close to 1, it indicates that the collaborative patterns obtained through decomposition can well explain the original data, that is, the reconstructed matrix X r is very close to the original matrix X, indicating high accuracy of collaborative decomposition; if the VAF value is low, it means that there are significant differences between the reconstructed data and the original data, implying that the collaborative patterns obtained by decomposition are insufficient to explain most of the variances in the original data and the decomposition effect is poor.

7. The PID electrical stimulation balance rehabilitation method based on injury synergy according to claim 1, characterized in that, On the basis of determining the electrical stimulation frequency and amplitude, the maximum pulse width of 400 us is linearly mapped to the maximum pulse width parameter limit of the lower limb muscle electrical stimulation parameter on the damage coordination time series, and the PID electrical stimulation control with damage coordination as the basic unit is carried out.

Images