A dual closed-loop control system for hand function rehabilitation integrating robotic hand and functional electrical stimulation

Abstract

The present invention relates to a dual closed-loop control system for hand function rehabilitation that integrates a manipulator and functional electrical stimulation, and belongs to the field of rehabilitation medical technology, including a myoelectric-inertial signal synchronous acquisition and data preprocessing module, a hand fine movement feature extraction and intention analysis module, a hand movement function monitoring and intelligent evaluation module, a wearable manipulator adaptive control module, a functional electrical stimulation dynamic control module, an information transmission and database module, and a wearable rehabilitation hand hardware. The present invention fully considers the subjective intention and motor ability of the patient, integrates the passive rehabilitation of the rehabilitation robot and the active rehabilitation advantages of neural regulation, solves the problems of the existing hand rehabilitation strategies and systems, such as simple training mode, poor human-computer interaction and individual adaptability, and unsatisfactory rehabilitation training effect, and helps to accelerate the recovery of neuroplasticity function and the rehabilitation process.

Description

Technical Field

[0001] The invention relates to a double closed-loop control system for hand function rehabilitation integrating a manipulator and functional electrical stimulation, and belongs to the technical field of rehabilitation medicine. Background Art

[0002] Stroke, with its high morbidity, high mortality, and high disability rate, seriously threatens human health and life. According to statistics, 30%-60% of stroke patients have motor dysfunction in the upper limbs, especially the hands, which seriously affects the patients' rehabilitation process and quality of life. In addition, since their rehabilitation training techniques and methods are relatively backward, research on advanced, convenient, and efficient hand function rehabilitation treatment techniques and equipment is of great research value and practical significance.

[0003] Functional rehabilitation of hand after stroke is an important link that hinders the overall rehabilitation process of patients, and it is also a pain point and difficulty in the field of rehabilitation. In recent years, with the increasing number of stroke patients and the increasing demand for daily living ability, fine functional rehabilitation of hand after stroke is imperative. The existing clinical rehabilitation methods for hand after stroke are mainly traditional rehabilitation treatments such as neuromuscular proprioception facilitation technology, occupational therapy, and traditional Chinese medicine. Its relatively mature technology has limitations such as physician dependence and rehabilitation targeting. In recent years, rehabilitation treatment technologies based on physical stimulation such as electricity, magnetism, and ultrasound (such as functional electrical stimulation, transcranial electrical stimulation, transcranial direct current stimulation, etc.) have been continuously developed. It has the characteristics of long-lasting efficacy and strong targeting, but it is relatively lacking in effective regulation effect evaluation and regulation strategy guidance.

[0004] Although foreign technologies and products related to hand rehabilitation have made some progress, there are relatively few rehabilitation systems for patients in the early flaccid and spastic stages, and there are still limitations in achieving full-cycle rehabilitation training for the hand and meeting the rehabilitation needs of different levels and treatment stages. The gap between China and foreign countries in hand rehabilitation research is not large. Driven by the good industry-university-research model jointly built by research institutions, hospitals and enterprises, some technologies have reached the international leading level. However, the industrial application of existing hand rehabilitation robots still has the following shortcomings: (1) Existing systems are mostly based on passive motor rehabilitation methods, and patients do not actively participate in rehabilitation, which is limited in actively inducing neural remodeling; (2) It is difficult to achieve individual adaptive rehabilitation training and adaptive control of rehabilitation modes for patients with different rehabilitation stages and degrees of hand motor dysfunction; (3) Rehabilitation robot technology based on new technologies such as brain-computer interface and biofeedback still has problems such as high cost and large volume, which limits its popularity in home and community use.

[0005] To this end, we have combined advanced technologies such as robots, neural regulation, virtual reality, and fully considered the patient's subjective intentions and motor abilities. We have integrated the passive rehabilitation advantages of rehabilitation robots and the active rehabilitation advantages of neural regulation to develop a wearable hand function rehabilitation system that integrates motion-physiological information feedback and functional electrical stimulation closed-loop regulation. This is of great significance for realizing intelligent, miniaturized, and personalized needs, accelerating the process of neuroplasticity rehabilitation, restoring the daily living ability of stroke patients, and promoting the development of intelligent rehabilitation technology. Summary of the invention

[0006] The purpose of the present invention is to provide a dual closed-loop control system for hand function rehabilitation that integrates a manipulator and functional electrical stimulation, fully considering the patient's subjective intention and motor ability, integrating the passive rehabilitation advantages of a rehabilitation robot and the active rehabilitation advantages of neural regulation, and solving the problems of existing hand rehabilitation strategies and systems, such as simple training modes, poor human-computer interaction and individual adaptability, and unsatisfactory rehabilitation training effects, which helps to accelerate the recovery of neuroplasticity functions and the rehabilitation process.

[0007] In order to achieve the above object, the technical solution adopted by the present invention is:

[0008] A dual closed-loop control system for hand function rehabilitation integrating a manipulator and functional electrical stimulation, including a myoelectric-inertial signal synchronous acquisition and data preprocessing module, a hand fine movement feature extraction and intention analysis module, a hand movement function monitoring and intelligent evaluation module, a wearable manipulator adaptive control module, a functional electrical stimulation dynamic control module, an information transmission and database module, and a wearable rehabilitation hand hardware;

[0009] The myoelectric-inertial signal synchronous acquisition and data preprocessing module is connected to the hand fine movement feature extraction and intention analysis module and the hand movement function monitoring and intelligent evaluation module through wireless communication. The data of the hand fine movement feature extraction and intention analysis module and the hand movement function monitoring and intelligent evaluation module are stored in the database through wireless communication, and are connected to the wearable manipulator adaptive control module and the functional electrical stimulation dynamic regulation module. The wearable manipulator adaptive control module and the functional electrical stimulation dynamic regulation module both control the wearable rehabilitation hand hardware through wireless communication and thus control the hand movement.

[0010] A further improvement of the technical solution of the present invention is that: the electromyographic-inertial signal synchronous acquisition and data preprocessing module is used for electromyographic and inertial data acquisition and signal preprocessing; the signal preprocessing first uses an adaptive high-pass filter and an adaptive 50Hz power frequency notch filter to remove baseline drift and power frequency interference in the signal respectively, and then uses a Butterworth third-order bandpass FIR filter for processing to select the effective frequency band characteristics of the signal.

[0011] A further improvement of the technical solution of the present invention is that: the hand fine movement feature extraction and intention analysis module is used for fine movement feature extraction and intention analysis; the fine movement feature extraction obtains the multi-domain dynamic features of electromyography and motion information based on multi-scale entropy and wavelet packet decomposition, and extracts stable feature indicators that reflect the fine intention of the hand; intention analysis obtains the user's fine movement intention through the extracted multi-domain dynamic features using a support vector machine algorithm.

[0012] A further improvement of the technical solution of the present invention is that the hand motor function monitoring and intelligent evaluation module is used for motor function feature extraction, functional status evaluation and movement completion evaluation; the motor function feature extraction combines consistency and non-negative matrix decomposition to decompose the synchronous coupling characteristics of electromyography and motion information; the functional status evaluation is based on the extracted synchronous coupling characteristics of electromyography and motion information to obtain objective quantitative indicators that reflect the hand motor function state, thereby realizing hand motor function state evaluation; the movement completion evaluation is based on the extracted synchronous coupling characteristics of electromyography and motion information to obtain objective quantitative indicators that reflect the hand movement trajectory, thereby realizing hand movement completion evaluation.

[0013] A further improvement of the technical solution of the present invention is that the wearable manipulator adaptive control module is used for rehabilitation mode selection and key parameter control.

[0014] A further improvement of the technical solution of the present invention is that the functional electrical stimulation dynamic control module is used for dynamic control and parameter selection of functional electrical stimulation, and an online learning and closed-loop adjustment strategy of FES combination parameters is constructed based on a deep learning algorithm to achieve dynamic control of key parameters such as stimulation position and intensity; parameter selection is based on the extracted features, and appropriate parameters are selected for control.

[0015] A further improvement of the technical solution of the present invention is that the information transmission and database module includes wireless communication and a database.

[0016] A further improvement of the technical solution of the present invention is that the wearable rehabilitation hand hardware includes a wearable manipulator and a functional electrical stimulation unit.

[0017] Due to the adoption of the above technical solution, the technical effects achieved by the present invention are as follows:

[0018] The present invention utilizes the characteristics and advantages of electromyographic-inertial signals and embedded systems, and realizes the recognition of the user's hand's fine intention and the evaluation of the user's motor function status and action completion by collecting electromyographic signals and inertial signals, and further realizes the adaptive selection of rehabilitation modes of the rehabilitation manipulator, the dynamic control of key parameters and the dynamic regulation of FES key parameters; the present invention integrates the passive rehabilitation advantages of the rehabilitation robot and the active rehabilitation advantages of neural regulation, solves the problems of the existing hand rehabilitation strategies and systems, such as simple training modes, poor human-computer interaction and individual adaptability, and unsatisfactory rehabilitation training effects, and helps to accelerate the functional recovery of neural plasticity and the rehabilitation process. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a structural block diagram of the present invention;

[0020] Figure 2 is a flow chart of the method of the present invention; DETAILED DESCRIPTION

[0021] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are only embodiments of a part of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work should fall within the scope of protection of the present invention.

[0022] A dual closed-loop control system for hand function rehabilitation that integrates a robotic hand and functional electrical stimulation, such as Figure 1 As shown, it includes a myoelectric-inertial signal synchronous acquisition and data preprocessing module, a hand fine movement feature extraction and intention analysis module, a hand movement function monitoring and intelligent evaluation module, a wearable manipulator adaptive control module, a functional electrical stimulation dynamic regulation module, an information transmission and database module and a wearable rehabilitation hand hardware.

[0023] The myoelectric-inertial signal synchronous acquisition and data preprocessing module is connected to the hand fine movement feature extraction and intention analysis module and the hand movement function monitoring and intelligent evaluation module through wireless communication. The data of the hand fine movement feature extraction and intention analysis module and the hand movement function monitoring and intelligent evaluation module are stored in the database through wireless communication, and are connected to the wearable manipulator adaptive control module and the functional electrical stimulation dynamic regulation module. The wearable manipulator adaptive control module and the functional electrical stimulation dynamic regulation module both control the wearable rehabilitation hand hardware through wireless communication and thus control the hand movement.

[0024] The myoelectric-inertial signal synchronous acquisition and data preprocessing module is used for myoelectric and inertial data acquisition and signal preprocessing; the hand fine movement feature extraction and intention analysis module is used for fine movement feature extraction and intention analysis; the hand motor function monitoring and intelligent evaluation module is used for motor function feature extraction, functional state evaluation and movement completion evaluation; the wearable manipulator adaptive control module is used for rehabilitation mode selection and key parameter control; the functional electrical stimulation dynamic regulation module is used for functional electrical stimulation dynamic regulation and parameter selection; the information transmission and database module includes wireless communication and database; the wearable rehabilitation hand hardware includes a wearable manipulator and a functional electrical stimulation unit.

[0025] The following is a detailed introduction to the various components of the rehabilitation manipulator control system:

[0026] 1. Myoelectric-inertial signal synchronous acquisition and data preprocessing module

[0027] The myoelectric and inertial data collection refers to collecting surface myoelectric signals and inertial signals of the user's hands.

[0028] The signal preprocessing refers to filtering the collected electromyographic signals and inertial information. First, an adaptive high-pass filter and an adaptive 50Hz power frequency notch filter are used to remove the baseline drift and power frequency interference in the signal respectively; then a Butterworth third-order bandpass FIR filter is used for processing to select the effective frequency band characteristics of the signal.

[0029] 2. Hand fine movement feature extraction and intention analysis module

[0030] This module is used for hand fine movement feature extraction and intention analysis. Fine movement feature extraction obtains multi-domain dynamic features of electromyography and motion information based on multi-scale entropy, wavelet packet decomposition and other methods. First, the signal of each frequency band is extracted by wavelet packet decomposition, and then the features of each frequency band are extracted by multi-scale entropy. Intention analysis uses the support vector machine algorithm to obtain the user's fine movement intention through the extracted multi-domain dynamic features.

[0031] Among them, multiscale entropy extends sample entropy to multiple time scales to provide additional observation perspectives when the time scale is uncertain.

[0032] The basic principle of multiscale entropy involves coarsening or downsampling the time series, mainly analyzing the time series at increasingly coarse temporal resolutions.

[0033] When scale is 1, the result of coarse-grained data is the original time series;

[0034] When scale is 2, the coarse-grained series is formed by calculating the average of two consecutive time points, that is:

[0035] y1=(x1+x2) / 2; y2=(x3+x4) / 2

[0036] When scale is 3, the coarse-grained time series is composed of the average values ​​of three consecutive time points, that is, the definition:

[0037] y1=(x1+x2+x3) / 3; y2=(x4+x5+x6) / 3

[0038] And so on.

[0039] The coarse-graining process can be divided into two forms: one is the non-overlapping type, in which τ data are jumped each time, and the τ data are averaged to generate new data; the other is the overlapping type, in which 1-τ data are jumped each time, and the τ data are averaged.

[0040] The above coarse-graining process is non-overlapping, and its mathematical definition is:

[0041]

[0042] Among them, τ represents the time scale, that is, each jump is τ data, j represents the number of data collected, Represents the data collected for the jth time with a time scale of τ.

[0043] Then, the sample entropy corresponding to each scale or resolution is calculated and a sample entropy-scale curve is plotted. The area under the curve is the sum of the sample entropy values ​​within the selected scale range, which is a measure of multi-scale entropy. A time series with large fluctuations will produce a larger entropy value and can therefore be considered a signal with higher complexity. Similarly, a signal with a high degree of regularity will have a lower entropy value.

[0044] Wavelet packet decomposition can decompose both the low-frequency part of the signal and the high-frequency part, and this decomposition has neither redundancy nor omission, so it can perform better time-frequency localization analysis on signals containing a large amount of medium and high-frequency information.

[0045]

[0046] Where i = 0, 1, 2, ..., 2 j -1,f j,i (t i ) is the reconstructed signal of the j-th layer node (j,i) after wavelet packet decomposition.

[0047] Support vector machines (SVM) are a binary classification model. Its basic model is a linear classifier with the largest margin defined in the feature space. The largest margin makes it different from the perceptron. SVM also includes kernel techniques, which makes it a nonlinear classifier in essence. The learning strategy of SVM is to maximize the margin, which can be formalized as a problem of solving a convex quadratic programming, which is equivalent to the problem of minimizing the regularized hinge loss function. The learning algorithm of SVM is the optimization algorithm for solving convex quadratic programming.

[0048] 3. Hand movement function monitoring and intelligent assessment module

[0049] This module is used for motion function feature extraction, functional status evaluation and action completion evaluation. The motion function feature extraction combines consistency, non-negative matrix decomposition and other methods to decompose the synchronous coupling characteristics of electromyography and motion information; the functional status evaluation is based on the synchronous coupling characteristics of the extracted electromyography and motion information to obtain objective quantitative indicators that reflect the functional status of the hand movement, and realize the evaluation of the functional status of the hand movement; the action completion evaluation is based on the synchronous coupling characteristics of the extracted electromyography and motion information to obtain objective quantitative indicators that reflect the hand movement trajectory, and realize the evaluation of the hand movement completion;

[0050] Among them, in the extraction of motor function features, consistency analysis verifies the consistency of the distribution changes of each signal, that is, when the value of x increases, does the value of y increase or decrease accordingly. The mutual inhibition relationship should be one increase while the other decreases, while the reciprocal relationship often follows each other, with the same high and low. It is the distribution consistency level, or the distribution trend consistency.

[0051] The basic idea of ​​non-negative matrix factorization (NMF) can be simply described as follows: for any given non-negative matrix V, the NMF algorithm can find a non-negative matrix W and a non-negative matrix H, thereby decomposing a non-negative matrix into the product of the left and right non-negative matrices:

[0052] V≈WH

[0053] In R n A bunch of spatially distributed data has certain distribution patterns. So we need to find a set of bases that can more intuitively reflect these patterns, and then project the original data onto this set of bases to represent them. This will facilitate subsequent applications.

[0054] The V in the formula is an n×m dimensional matrix, each column of which is R n A vector in space, a total of m vectors; W is an n×k dimensional matrix, that is, k basis; H is a k×m matrix, each column is the vector obtained by projecting V onto W.

[0055] 4. Wearable manipulator adaptive control module

[0056] This module is used for rehabilitation mode selection and key parameter control. The rehabilitation mode selection module reasonably selects the rehabilitation training mode based on fine intention analysis, motion state function evaluation and action completion evaluation. Key parameter control is based on fine intention analysis, motion state function evaluation and action completion evaluation, and uses dynamic feedback algorithm to achieve adaptive adjustment of rehabilitation hand control parameters in different modes.

[0057] The rehabilitation training modes in the rehabilitation mode selection include: active mode, passive mode, active and passive combined mode; the key parameters controlled by the key parameter control include: speed, finger angle, finger strength, etc.

[0058] 5. Functional electrical stimulation dynamic control module

[0059] This module is used for dynamic control and parameter selection of functional electrical stimulation. Based on the deep learning algorithm, the online learning and closed-loop adjustment strategy of FES combination parameters is constructed to achieve dynamic control of key parameters such as stimulation position and intensity; parameter selection is based on the extracted features, and appropriate parameters are selected for control.

[0060] The parameters regulated by dynamic control of functional electrical stimulation include: frequency, pulse, duty cycle, wave rise / wave fall, current intensity, stimulation site, etc.

[0061] The specific training plan is as follows:

[0062] During the flaccid paralysis period of rehabilitation, the task-associated agonist muscles of the patient's intended movement are positively stimulated by optimizing features and optimizing stimulation parameters, thereby inducing active movement in the patient and accelerating the early rehabilitation process. During the spasticity period of rehabilitation, FES is used to stimulate the antagonist muscles to reduce the degree of spasticity, while inducing the completion of simple single-joint movements, inducing abnormal movement patterns to develop in the direction of separation and autonomous movement. During the rehabilitation recovery period, fine motor function recovery is completed by identifying behavioral intentions and optimizing stimulation parameters. Guided by the completion of daily tasks, the difficulty of tasks is increased and their interest is increased, thereby promoting a higher level of activation of movement-related brain areas.

[0063] VI. Information Transmission and Database Module

[0064] This module includes communication and database, and the communication realizes information transmission between systems through wireless mode. The database contains user's personal case information, test information, evaluation indicators, system parameters and other information, realizes multi-level data storage and management, and visualizes the user's status assessment and rehabilitation process.

[0065] 7. Wearable rehabilitation hardware

[0066] The hardware includes a wearable manipulator and a functional electrical stimulation unit. The wearable manipulator is responsible for receiving instructions sent by the wearable manipulator adaptive control module and generating corresponding actions; the functional electrical stimulation receives instructions sent by the functional electrical stimulation dynamic control module, adjusts electrical stimulation parameters, and outputs electrical stimulation. The two complement each other to realize the patient's fine motor intention.

[0067] The dual closed-loop control system for hand function rehabilitation that integrates a manipulator and functional electrical stimulation in the embodiment of the present invention utilizes the characteristics and advantages of electromyographic-inertial signals and embedded systems, and through the collection and analysis of electromyographic-inertial signals, realizes the recognition of the user's hand's fine intentions and the evaluation of the user's motor function status and movement completion, and further realizes the adaptive selection of rehabilitation modes of the rehabilitation manipulator, the dynamic modulation of key parameters, and the dynamic regulation of FES key parameters; by combining the rehabilitation manipulator with functional electrical stimulation, it helps to accelerate the recovery of neuroplasticity functions and the rehabilitation process.

[0068] like Figure 2 As shown, corresponding to a dual closed-loop control system for hand function rehabilitation integrating a manipulator and functional electrical stimulation in the above-mentioned embodiment, the present invention provides a dual closed-loop control method for hand function rehabilitation integrating a manipulator and functional electrical stimulation, the method comprising:

[0069] Synchronously acquiring the patient's electromyographic and inertial signals, and preprocessing the signals;

[0070] Feature extraction based on preprocessed electromyographic signals and inertial signals;

[0071] Perform movement intention analysis and movement function status assessment based on the extracted features;

[0072] Based on the analysis of the user's movement intention and the evaluation of the movement function status, a dynamic feedback algorithm is used to achieve adaptive adjustment of the control parameters of the rehabilitation hand in different modes, and the rehabilitation mode is selected according to the patient's functional status;

[0073] Based on the analysis of the user's movement intention and the evaluation of the movement function status, parameters are selected, and online learning and closed-loop adjustment of FES combination parameters are achieved through deep learning algorithms.

[0074] The embodiment of the present invention provides a dual closed-loop control method for hand function rehabilitation that integrates a manipulator and functional electrical stimulation. It utilizes the characteristics and advantages of electromyographic-inertial signals and embedded systems, and collects electromyographic signals and inertial signals to realize the recognition of the user's hand's fine intentions and the evaluation of the user's motor function status and movement completion. Furthermore, it realizes the adaptive selection of rehabilitation modes of the rehabilitation manipulator, the dynamic control of key parameters, and the dynamic regulation of FES key parameters. By combining the rehabilitation manipulator with functional electrical stimulation, it helps to accelerate the recovery of neuroplasticity functions and the rehabilitation process.

[0075] In the embodiments provided by the present invention, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic, for example, the unit division can be a logical function division, and there may be other division methods in actual implementation, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, units or modules, which can be electrical or other forms.

[0076] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed over multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0077] In addition, in various embodiments of the present invention, each functional unit may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0078] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory, random access memory, mobile hard disk, magnetic disk or optical disk, etc., which can store program code.

[0079] 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 replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A dual closed-loop control system for hand function rehabilitation integrating a manipulator and functional electrical stimulation, characterized in that: It includes the module of synchronous acquisition of electromyography-inertial signals and data preprocessing, the module of fine hand movement feature extraction and intention analysis, the module of hand movement function monitoring and intelligent evaluation, the module of wearable manipulator adaptive control, the module of functional electrical stimulation dynamic regulation, the module of information transmission and database, and the hardware of wearable rehabilitation hand; The myoelectric-inertial signal synchronous acquisition and data preprocessing module is connected to the hand fine movement feature extraction and intention analysis module and the hand movement function monitoring and intelligent evaluation module through wireless communication. The myoelectric-inertial signal synchronous acquisition and data preprocessing module is used for myoelectric and inertial data acquisition and signal preprocessing; the signal preprocessing first uses an adaptive high-pass filter and an adaptive 50Hz power frequency notch filter to remove the baseline drift and power frequency interference in the signal respectively, and then uses a Butterworth third-order bandpass FIR filter to process and select the effective frequency band characteristics of the signal; the hand fine movement feature extraction and intention analysis module and the hand movement function monitoring and intelligent evaluation module are connected to the hand fine movement feature extraction and intention analysis module and the hand movement function monitoring and intelligent evaluation module. The data of the estimation module are stored in the database through wireless communication, and are connected to the wearable manipulator adaptive control module and the functional electrical stimulation dynamic regulation module. The hand fine movement feature extraction and intention analysis module is used for fine movement feature extraction and intention analysis; the fine movement feature extraction is based on multi-scale entropy and wavelet packet decomposition to obtain multi-domain dynamic features of electromyography and motion information, and extract stable feature indicators that reflect the fine intention of the hand; the intention analysis uses the support vector machine algorithm to obtain the user's fine movement intention through the extracted multi-domain dynamic features; wherein, the multi-scale entropy expands the sample entropy to multiple time scales, so as to provide additional observation perspectives when the time scale is uncertain; Multiscale entropy involves coarsening or downsampling the time series, analyzing the time series at increasingly coarse temporal resolutions; When scale is 1, the result of coarse-grained data is the original time series; When scale is 2, the coarse-grained series is formed by calculating the average of two consecutive time points, that is: y1=(x1+x2) / 2; y2=(x3+x4) / 2 When scale is 3, the coarse-grained time series is composed of the average values ​​of three consecutive time points, that is, the definition: y1=(x1+x2+x3) / 3; y2=(x4+x5+x6) / 3 And so on; There are two forms of coarse-graining process: one is non-overlapping, each time jumping τ data, taking τ data as average to generate new data; the other is overlapping, each time jumping 1-τ data, taking τ data as average; The above coarse-graining process is non-overlapping, and its mathematical definition is: Among them, τ represents the time scale, that is, each jump is τ data, j represents the number of data collected, represents the data collected for the jth time with a time scale of τ; Then, the sample entropy corresponding to each scale or resolution is calculated, and the sample entropy-scale curve is plotted; the area under the curve is the sum of the sample entropy values ​​within the selected scale range, which is a measure of multi-scale entropy. A time series with large fluctuations will produce a large entropy value, so it can be considered as a signal with high complexity; wavelet packet decomposition can decompose both the low-frequency part of the signal and the high-frequency part, and this decomposition has neither redundancy nor omission, so it can perform better time-frequency localization analysis on signals containing a large amount of medium and high-frequency information; Where i = 0, 1, 2, ..., 2 j -1,f j,i (t i ) is the reconstructed signal of the j-th layer node (j,i) after wavelet packet decomposition; The functional electrical stimulation dynamic control module is used for dynamic control and parameter selection of functional electrical stimulation. Based on the deep learning algorithm, an online learning and closed-loop adjustment strategy of FES combination parameters is constructed to realize dynamic control of key parameters of stimulation position and intensity. Parameter selection is based on the extracted features, and appropriate parameters are selected for control. The parameters adjusted by dynamic control of functional electrical stimulation include: frequency, pulse, duty cycle, wave rise / wave fall, current intensity, and stimulation site. The specific training program is as follows: During the rehabilitation flaccid paralysis period, by optimizing features and optimizing stimulation parameters, the task-associated agonist muscles of the patient's intended movement are positively stimulated to induce active movement in the patient and accelerate the early rehabilitation process; during the rehabilitation spasticity period, FES is used to stimulate antagonist muscles to reduce the degree of spasticity, while inducing simple single-joint movements, inducing abnormal movement patterns to develop in the direction of separation and autonomous movement; during the rehabilitation recovery period, by identifying behavioral intentions and optimizing stimulation parameters, fine motor function recovery is completed, and the completion of daily tasks is taken as the guide to increase the difficulty of tasks and increase their interest, thereby promoting a higher level of activation of movement-related brain areas; the wearable robotic hand adaptive control module and the functional electrical stimulation dynamic regulation module both control the wearable rehabilitation hand hardware through wireless communication and thus control hand movements.

2. According to claim 1, a dual closed-loop control system for hand function rehabilitation integrating a manipulator and functional electrical stimulation, characterized in that: The hand motor function monitoring and intelligent evaluation module is used for motor function feature extraction, functional state evaluation and movement completion evaluation; the motor function feature extraction combines consistency and non-negative matrix decomposition to decompose the synchronous coupling characteristics of electromyography and motion information; the functional state evaluation is based on the synchronous coupling characteristics of the extracted electromyography and motion information, and obtains objective quantitative indicators that reflect the hand motor function state, thereby realizing hand motor function state evaluation; the movement completion evaluation is based on the synchronous coupling characteristics of the extracted electromyography and motion information, and obtains objective quantitative indicators that reflect the hand movement trajectory, thereby realizing hand movement completion evaluation.

3. According to claim 1, a dual closed-loop control system for hand function rehabilitation integrating a manipulator and functional electrical stimulation, characterized in that: The wearable manipulator adaptive control module is used for rehabilitation mode selection and key parameter control.

4. According to claim 1, a dual closed-loop control system for hand function rehabilitation integrating a manipulator and functional electrical stimulation, characterized in that: The information transmission and database module includes wireless communication and database.

5. According to claim 1, a dual closed-loop control system for hand function rehabilitation integrating a manipulator and functional electrical stimulation, characterized in that: The wearable rehabilitation hand hardware includes a wearable robotic hand and a functional electrical stimulation unit.

Images