CN116211656A

CN116211656A - Near-infrared brain function imaging-based lower limb rehabilitation training system and method

Info

Publication number: CN116211656A
Application number: CN202310177518.1A
Authority: CN
Inventors: 杨巍; 桂梁鸿; 曹巧欢; 颜泽皓; 冯陆颖; 余林繁; 杨灿军; 刘晓健
Original assignee: Zhejiang University ZJU
Current assignee: Zhejiang University ZJU
Priority date: 2023-02-28
Filing date: 2023-02-28
Publication date: 2023-06-06

Abstract

The invention discloses a lower limb rehabilitation training system based on near infrared brain function imaging, which is characterized by comprising a lower limb exoskeleton robot, data acquisition equipment and an upper computer; the lower limb exoskeleton robot comprises a waist binding part, wherein thigh rods, shank rods and foot pedals which are sequentially connected are respectively arranged on two sides of the waist binding part, and the thigh rods, the shank rods and the foot pedals are connected through motors so as to realize the respective driving of hip joints, knee joints and ankle joints of legs; the data acquisition equipment comprises a near infrared light transmitting module, a photoelectric conversion receiving module, a demodulation module and a control module, wherein the upper computer is used for obtaining rehabilitation training evaluation indexes and operating a loop optimization algorithm of a person; the invention also discloses a lower limb rehabilitation training method based on near infrared brain function imaging. According to the invention, the lower limb rehabilitation exoskeleton robot is used for participating in the fNIRS-based rehabilitation technology, so that the automation and the intellectualization of rehabilitation training are realized.

Description

Near-infrared brain function imaging-based lower limb rehabilitation training system and method

Technical Field

The invention relates to the technical field of rehabilitation equipment, in particular to a lower limb rehabilitation training system and method based on near infrared brain function imaging.

Background

Cerebral apoplexy is commonly called as "apoplexy", is an acute disease caused by vascular diseases for conveying blood to the brain, comprises ischemic and hemorrhagic strokes, has the characteristics of high disability rate, high recurrence rate and high economic burden besides high death rate, and is one of the primary reasons for adult disability.

In early rehabilitation, the muscle strength of a plurality of patients is low, a complete rehabilitation training action cannot be completed, the lower limb rehabilitation exoskeleton can assist the patients to complete corresponding actions, a passive training mode or an auxiliary training mode combining the passive and the active is provided, the patients with better muscle strength can perform active training by utilizing the muscle strength of the patients, and meanwhile, the rehabilitation exoskeleton can also provide proper resistance to increase training effects. Previously, a learner developed exoskeleton robots for rehabilitation, called hybrid assisted limbs (hybrid assistive limb, HAL), based on the "interactive biofeedback" theory.

The actions of the HAL robot are triggered by bioelectric signals (BES) detected from muscles, the HAL supports autonomous movements of the impaired limb and feeds back the sensation of successful movements to the brain, which is activated by the sensory feedback, so that the signals of the motor network are enhanced, the HAL treatment forms a closed loop, inducing an improvement of the function of the impaired limb and potentially promoting neural plasticity. Many studies on HAL have demonstrated the effectiveness of exoskeleton robots for neurological rehabilitation.

Although the various training modes of the rehabilitation exoskeleton enable the cerebral apoplexy patient to take part in the training process in a relatively early stage, the problems of how the active participation degree of the patient, how the muscle strength recovery condition is, how the activation degree of the related area of the cerebral cortex is, how the rehabilitation training effect is still to be quantified and solved, and most of the problems are judged by only depending on the experience of rehabilitation doctors at present.

The rise of near infrared brain function imaging (fNIRS) technology, which is an emerging non-invasive brain imaging technology that uses oxygen-carrying hemoglobin (HbO) in brain tissue, brings about the resolution of the above-mentioned problems ₂ ) And deoxyhemoglobin (HbR) has differential characteristics on near infrared light absorptivity of 600-900nm wavelength, hbO2 and HbR of the cerebral cortex and the absolute or relative change of the concentration of HbT of the total HbT of the HbO2 and HbR can be directly detected in real time by attaching the optical probe to the Brain to send and receive near infrared light, and a coupling relationship exists between nerves and blood vessels, so that the neuron activity of the cerebral cortex can be further deduced by using a modified Beer-Lambert law, and the novel Brain-computer interface (Brain-computer interface, BCI) can be further used. Since stroke patients are often accompanied by limb movement disorder, and the fNIRS technique does not impose too much restriction on the posture of the subject as other techniques, it is well suited for evaluation and study of the effects of rehabilitation training on stroke patients.

Academic research shows that the functional connection strength (functional connectivity, FC) between the bilateral primary cerebral motor cortex (Primary Motor Cortex, M1) of a person with unilateral motor function impaired after cerebral stroke is obviously positively correlated with the motor function score, and after rehabilitation training for a certain number of weeks, the brain function connection of the bilateral M1 area is gradually enhanced, and the motor function is also gradually recovered. Therefore, FC can be used as a quantitative index for assessing rehabilitation effect. In summary, the fNIRS technology can be used for quantitatively evaluating and feeding back information to guide rehabilitation training, but many existing researches only take the fNIRS technology as a means for evaluating rehabilitation effect, and although some researches take the fNIRS technology as a brain-computer interface, the feedback information is used for controlling a rehabilitation exoskeleton robot, only the exoskeleton can perform a few very simple control actions, and the active rehabilitation effect is not well exerted.

The application with publication number of CN114869232A discloses a rehabilitation training feedback method and system based on near infrared brain imaging, which can feed back and guide rehabilitation training auxiliary equipment of a trainee in real time, quantify rehabilitation training quality, take brain imaging map change characteristic data under standard rehabilitation training action guided by a rehabilitation engineer for the first time as a base line, remind the trainee through a bracelet, and ensure that the trainee completes rehabilitation training with high quality. According to the scheme, whether the rehabilitation actions of the trainee reach the standard or not can be reminded by the aid of the fNIRS feedback information and the bracelet, so that the workload of a rehabilitation doctor is reduced to a certain extent, but the target reference value cannot be dynamically adjusted along with the recovery of the trainee, and when the muscle strength of the trainee recovers a little, the rehabilitation doctor is required to reset target data, the operation is complicated, and the personal subjective experience of the doctor is relied on.

The application with publication number CN114391815a discloses a near-infrared brain function imaging device, which has a headgear for wearing on the head of a subject, wherein the headgear is provided with a plurality of probes for transmitting and/or receiving near-infrared signals to collect near-infrared signals of a plurality of corresponding channels, and data collection is performed for corresponding brain regions, so that the collection of near-infrared signals in a large-scale brain region is avoided, analysis and evaluation are completed in a short time, so that efficiency is improved, and meanwhile, signal drift and frequency band interference can be reduced, so that accuracy of analysis results is improved. According to the scheme, only near-infrared brain function imaging is used for evaluation and analysis, drift and disturbance are improved, a specific rehabilitation training adjustment is needed to be carried out, an adjustment strategy is provided after an evaluation result is read by a rehabilitation doctor, and at present, resources of the rehabilitation doctor are relatively tense, so that some patients cannot be timely trained.

In addition, the fNIRS-based rehabilitation technique is not combined with intelligent rehabilitation training equipment such as rehabilitation training exoskeleton and the like, so that the automation degree and the intelligent degree are low.

Disclosure of Invention

The invention aims to provide a lower limb rehabilitation training system based on near infrared brain function imaging, which uses a lower limb rehabilitation exoskeleton robot to participate in a rehabilitation technology based on fNIRS, uses relevant indexes fed back by fNIRS for evaluating the rehabilitation training effect of a patient to guide and modify a power-assisted control law or a position control law of the lower limb exoskeleton robot, and realizes automation and intellectualization of rehabilitation training.

A lower limb rehabilitation training system based on near infrared brain function imaging comprises a lower limb exoskeleton robot, data acquisition equipment and an upper computer;

the lower limb exoskeleton robot comprises a waist binding part, wherein thigh rods, shank rods and foot pedals which are sequentially connected are respectively arranged on two sides of the waist binding part, and the thigh rods, the shank rods and the foot pedals are connected through motors so as to realize the respective driving of hip joints, knee joints and ankle joints of legs; the motor is fixedly provided with an encoder for measuring the angular displacement and the angular speed of the motor; the thigh rod and the shank rod are fixedly provided with inertial measurement units for measuring the postures of different limb segments of the human body; the lower limb exoskeleton robot further comprises a signal receiving and transmitting module used for communicating with the upper computer;

the data acquisition equipment comprises a near infrared light emitting module, a photoelectric conversion receiving module, a demodulation module and a control module, wherein the near infrared light emitting module is used for receiving control signals of the control module and sending needed near infrared light to the brain scalp of a human body, the photoelectric conversion module is used for receiving near infrared light signals absorbed and attenuated by human tissues, converting the light signals into electric signals and transmitting the electric signals to the demodulation module, the demodulation module demodulates the converted electric signals, distinguishes the signals of all channels and converts the signals into digital signals, and then transmits the digital signals to the control module, and the control module transmits the data to the upper computer;

the upper computer obtains corresponding rehabilitation training evaluation index data by obtaining activation information of different brain areas, optimizes the data by an optimization algorithm, sends control parameters to the lower limb exoskeleton robot, and obtains final auxiliary parameters of the lower limb exoskeleton robot through multiple iterations, so that the optimal rehabilitation training effect is achieved under the assistance of the auxiliary parameters.

Preferably, the device further comprises a textile headgear, wherein a plurality of pairs of the near infrared light emitting modules and the photoelectric conversion receiving modules are distributed on the textile headgear.

Preferably, the thigh rod and the shank rod are telescopic, and thigh binding and shank binding are respectively arranged on the thigh rod and the shank rod.

Preferably, the optimization algorithm adopts a human-in-loop black box optimization algorithm.

The invention further aims to provide a lower limb rehabilitation training method based on near infrared brain function imaging, which comprises an auxiliary mode parameterization step and a human in-loop optimization algorithm step according to evaluation data;

the auxiliary mode parameterization step specifically comprises the following steps:

position control strategy is adopted in early rehabilitation stage, and the hip joint angle scaling coefficient K as a control parameter is taken _h Knee joint angle scaling coefficient K _k Ankle angle scaling factor K _n And gait cycle completion time scaling factor T _k On the basis of the change of the angle of the joints of the normal gait of the human body, a new joint position track curve can be generated by multiplying the corresponding scaling coefficient, so that the stride is changed, the cooperative action relation of three joints is not changed, and the gait cycle completion time scaling coefficient is multiplied to change the gait cycle completion time, so that the pace speed is adjusted;

a power-assisted control strategy is adopted in the later period of rehabilitation, a power-assisted curve is parameterized, and each parameter on the power-assisted curve is positioned in the range of the parameter obtained after preliminary experiments are carried out on normal people;

the step of the human-in-loop optimization algorithm according to the evaluation data is specifically as follows:

the parameter value of the auxiliary mode is used as an input parameter, the evaluation index related to human brain cortex activation fed back by the near infrared brain function imaging technology is used as an output value, and the output is maximized through continuous iterative optimization.

Preferably, the position control strategy specifically includes: the lower limb exoskeleton robot directly drives the lower limb of a person to move to a designated position, a walking gait with a planned track in advance is realized, three motors on the hip joint, the knee joint and the ankle joint are cooperatively matched, and in a gait cycle, the rotation angles of the three motors follow the joint angle data of the normal gait of the human body after the corresponding parameters are scaled.

Preferably, the assistance control strategy is that the lower limb exoskeleton robot gives corresponding assistance in different stages of walking of a person, the current gait phase is obtained by learning periodic signals through the adaptive oscillator, and the required assistance value is directly obtained according to a predefined parameterized assistance curve.

Preferably, the evaluation index is a functional connection strength.

The invention has the beneficial effects that:

(1) According to the invention, the lower limb rehabilitation exoskeleton robot is used for participating in the rehabilitation technology based on the fNIRS, and the related index fed back by the fNIRS for evaluating the rehabilitation training effect of the patient is used for guiding and modifying the assistance control law or the position control law of the lower limb exoskeleton robot, so that the automation and the intellectualization of the rehabilitation training are realized.

(2) The lower limb exoskeleton robot adopted by the invention has a higher-level control system, and can provide rehabilitation assistance for patients by using different parameterized control laws, wherein the control laws can be specifically the position of the lower limb exoskeleton robot, namely the swinging angle of a joint motor, and can also be assistance provided by the exoskeleton for people. Therefore, the brain cortex activation information fed back by the fNIRS is used as an evaluation index to guide the exoskeleton to provide proper assistance, and the exoskeleton can provide variable actions and is beneficial to rehabilitation of patients.

(3) The invention automatically optimizes and adjusts the assistance mode or the position assistance mode provided by the exoskeleton by using a loop black box optimization algorithm, and aims to find an optimal assistance mode under the current state of a patient, so that the functional connection strength (FC) of a corresponding brain region fed back by fNIRS is maximized under the assistance of the mode, and a targeted rehabilitation training effect is realized.

Drawings

FIG. 1 is a schematic diagram of the structure of the present invention;

FIG. 2 is a schematic diagram of a lower extremity exoskeleton robot;

fig. 3 is a schematic structural view of a near infrared light transmitting and receiving device of fnigs;

FIG. 4 is a schematic view of a textile headset with multiple pairs of light sources and detectors;

FIG. 5 is a schematic diagram of a 16 channel fNIRS apparatus installation distribution;

FIG. 6 is a block diagram of the overall flow of lower limb rehabilitation training;

FIG. 7 is a block diagram of data acquisition and processing of brain-related assessment indicators;

FIG. 8 is a graph of hip, knee, ankle angle data for a human being walking normally during a gait cycle;

FIG. 9 is a schematic diagram of an adaptive oscillator;

FIG. 10 is a schematic diagram of a hip joint assist curve;

FIG. 11 is a schematic diagram of a parameterized boost curve.

Detailed Description

The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

As shown in fig. 1, a lower limb rehabilitation training system based on near infrared brain function imaging includes a lower limb exoskeleton robot 100, a data acquisition device 200 and an upper computer 300.

As shown in fig. 2, the lower limb exoskeleton robot 100 comprises a waist binding 1, wherein two sides of the waist binding 1 are respectively provided with a thigh rod 2, a shank rod 3 and a foot pedal 4 which are sequentially connected, the thigh rod 2, the shank rod 3 and the foot pedal 4 are connected with each other through motors 5, and the number of the motors 5 is six specifically, so as to realize the respective driving of hip joints, knee joints and ankle joints of legs; the rear end of the waist binding is provided with a battery for powering the motor 5 and the control circuit.

In this embodiment, the thigh bar 2 and the shank bar 3 are telescopic to accommodate different body sizes; thigh binding and shank binding are also respectively arranged on the thigh rod 2 and the shank rod 3, and the binding can be freely adjusted to adapt to different human bodies and is used for fixing the exoskeleton of the lower limbs and the human bodies.

An encoder for measuring the angular displacement and the angular velocity of the motor 5 is fixed on the motor 5; the thigh rod 2 and the shank rod 3 are fixedly provided with inertial measurement units IMU for measuring the postures of different limb segments of the human body; the lower extremity exoskeleton robot 100 further comprises a signal transceiver module for communicating with the upper computer 300.

The data acquisition device 200 includes a near infrared light emitting module 6, a photoelectric conversion receiving module 7, a demodulation module and a control module, wherein the near infrared light emitting module 6 is used for receiving a control signal of the control module, controlling the near infrared light emitting diode to emit the required near infrared light, and emitting the required near infrared light to the scalp of the human brain by adopting an optical fiber, the photoelectric conversion module 7 is used for receiving the near infrared light signal absorbed and attenuated by human tissue, converting the optical signal into an electric signal, transmitting the electric signal to the demodulation module, demodulating the converted electric signal by the demodulation module, distinguishing the signals of each channel, converting the signals into digital signals, transmitting the digital signals to the control module, and transmitting the data to the upper computer 300 by the control module.

As shown in fig. 3, the near infrared light emitting and receiving device of fnigs, wherein white is a light source, that is, a near infrared light emitting module 6, and black is a detector, that is, a photoelectric conversion receiving module 7, light emitted by the light source is scattered and absorbed multiple times in human tissue, the propagation direction is changed, the intensity is attenuated to a certain extent, and finally, the light is received by the detector, and because of the significant difference of the absorption rate of oxyhemoglobin and deoxyhemoglobin in the tissue to near infrared light, when the relative content of the oxyhemoglobin and deoxyhemoglobin changes, the intensity of emergent light received at the detector changes significantly, and the change of the received near infrared light can be converted into the change of hemoglobin concentration in human brain tissue by using the modified beer-lambert law, and the hemoglobin concentration has a coupling relation with the activation of nerves, so as to realize the measurement of cerebral cortex nerve activation.

In this embodiment, a plurality of pairs of light sources and detectors are fixed on the textile headgear 400, so as to measure the brain of the patient, and the schematic diagram is shown in fig. 4, and signals are transmitted to the demodulation module through the signal transmission line to realize modulation, demodulation and signal processing.

In this embodiment, a 16-channel fnigs device is used, and an installation and distribution schematic diagram is shown in fig. 5, where the installation and distribution schematic diagram includes 16 light sources (white points) and 16 detectors (black points), and the device used can cover most of brain areas related to lower limb movements of the cerebral cortex, so as to provide convenience for obtaining corresponding rehabilitation training evaluation indexes.

The upper computer 300 obtains the activation information of different brain areas through the resolving signals, so as to obtain corresponding rehabilitation training evaluation index data, optimizes the data through an optimization algorithm, and sends control parameters to the lower limb exoskeleton robot to change the auxiliary action of the lower limb rehabilitation exoskeleton, and the optimization algorithm can find the best auxiliary mode through multiple iterations to achieve the best rehabilitation training effect, namely, obtain the highest rehabilitation training evaluation index. The whole flow chart of the lower limb rehabilitation training is shown in figure 6.

The functional connection strength of the brain area required to be stimulated and recovered can be selected as a recovery training evaluation index according to the recovery requirement, and the subsequent optimization algorithm can realize targeted recovery training so as to enhance the activation of the appointed brain area.

Specifically, the upper computer 300 calculates the change in blood oxygen concentration of the corresponding channel according to the modified beer-lambert law, thereby reflecting the change in functional connection strength of the corresponding brain region. The data acquisition and processing of brain-related assessment indicators is shown in fig. 7.

A lower limb rehabilitation training method based on near infrared brain function imaging comprises an auxiliary mode parameterization step and a human in-loop optimization algorithm step according to evaluation data.

Control related description of the lower limb exoskeleton robot is carried out before parameterization of the auxiliary mode. The control strategy of the exoskeleton of the lower limb mainly comprises a position control strategy and a power-assisted control strategy, and the position control strategy and the power-assisted control strategy have important roles in different stages of rehabilitation respectively, and are described below.

Position control strategy: the exoskeleton robot directly drives the lower limbs of the person to move to the designated position to realize walking gait of a pre-planned track, wherein three motors on the hip joint, the knee joint and the ankle joint are matched in a cooperative manner, in a gait cycle, the rotation angles of the three motors strictly follow the joint angle data acquired from the normal gait of the human body, as shown in fig. 8, the motor position tracking of the bottom layer adopts a PID control algorithm, and the phase difference of two legs is 180 degrees.

Boost control strategy: the exoskeleton gives corresponding assistance to different stages of human walking, the basic control steps of the exoskeleton comprise gait phase recognition, a required assistance value is generated according to a predefined assistance curve, a gait phase recognition algorithm adopts an adaptive oscillator, the adaptive oscillator is described as shown in fig. 9, the current gait phase is obtained by learning periodic signals such as angle data of an IMU, the gait phase of 0-2 pi is converted into 0-100%, the required assistance value can be directly obtained according to the predefined assistance curve, a hip joint is taken as an example, the assistance curve is shown in fig. 10, and then the assistance application is realized through closed loop force control of a bottom layer.

In early rehabilitation, the patient's own mobility is weaker, so it is preferable to adopt a position control strategy, for normal walking of the human body, the main walking parameters are stride and pace speed, the stride response corresponds to the upper and lower limit values of the hip joint, knee joint and ankle joint angle change in fig. 8, when the upper and lower limit values are reduced, the stride is correspondingly reduced, and when the upper and lower limit values are increased, the stride is correspondingly increased. The pace corresponds to the time taken to complete the entire 100% gait cycle, the shorter the completion time, the faster the pace, thus the control parameter K is desirable _h 、K _k 、K _n 、T _k Respectively a hip joint angle scaling coefficient, a knee joint angle scaling coefficient, an ankle joint angle scaling coefficient and a gait cycle completion timeOn the basis of the joint angle change of fig. 8, a new joint position track curve can be generated by multiplying the corresponding scaling coefficient, so that the stride can be changed without changing the cooperative action relation of three joints. The pace can be adjusted by changing the gait cycle completion time, and in conclusion, the four parameters can well change the action mode of position control, and the actions are changeable and accord with expectations. The preliminary experiment is carried out on normal people, so that the value ranges of four parameters for ensuring comfortable and safe walking of people can be determined, the safety of the patients is ensured, and the method is simultaneously used for iteration of a follow-up optimization algorithm.

Later in rehabilitation, the patient has certain walking capacity and has the requirement of active walking, so that the assisting control strategy is better adopted, the interaction between the exoskeleton and the human body is more comfortable, the rehabilitation training effect is better, and the assisting curve corresponding to fig. 10 is parameterized, as shown in fig. 11.

According to 8 parameters shown in FIG. 11, positive peak phases are respectively

Positive rising phase->

Positive falling phase

Positive peak torque τ _PP Negative peak phase->

Negative rising phase->

Negative falling phase->

And negative peak torque τ _NP The corresponding point coordinates can be obtained through the parameters, a needed power-assisted curve is generated by using a segmented cubic spline interpolation mode, and the corresponding range of the parameters which are comfortable to a normal person can be obtained by performing a preliminary experiment on the normal person, wherein the parameter range is defined as follows:

positive rise phase

Positive peak phase

Positive peak torque τ _PP ∈[0Nm,10Nm]

Positive falling phase

Negative rise phase

Negative peak phase

Negative peak torque τ _NP ∈[0Nm,10Nm]

Negative falling phase

The human loop optimization algorithm mainly comprises Bayesian optimization and covariance matrix self-adaptive evolution strategies, which are typical black box optimization algorithms, wherein the object aimed by the black box optimization algorithm is a black box function, namely, the black box optimization algorithm has a plurality of input parameters, but only has one output value, and the optimization algorithm is iterated continuously to find a group of input parameters which enable the output value to be maximum. For the current lower limb rehabilitation training problem, the system formed by the human and the exoskeleton robot is a black box function, the input parameters are parameter values of an auxiliary mode, the evaluation indexes related to human cerebral cortex activation fed back by fNIRS are output, and the output can be maximized through continuous iterative optimization, so that the optimal rehabilitation training effect is achieved.

For the optimization algorithm, the invention takes Bayesian optimization as an example for carrying out algorithm description, and the following table is an algorithm pseudo code.

Bayesian optimization algorithm pseudocode description

The basic working principle is that n is randomly explored firstly ₀ The sampling points, i.e. randomly generating n in parameter space ₀ The method comprises the steps of combining auxiliary mode parameter values, calculating a corresponding brain activation evaluation index, namely a sampling value, continuously iterating, estimating the distribution of a black box function by using Gaussian process regression according to current sampled data in each iteration, selecting the next group of explored parameter values according to the maximum value of an acquisition function, selecting the maximum expected lifting function (EI) by the acquisition function, hopefully exploring the region where the maximum value of the black box function is located, and automatically optimizing in different rehabilitation stages of a patient by using the circulation after a certain number of iterations, wherein the auxiliary mode parameter value corresponding to the found maximum rehabilitation training evaluation index is the optimal auxiliary training action description.

Although the present invention has been described with reference to the foregoing embodiments, it will be apparent to those skilled in the art that modifications may be made to the embodiments described, or equivalents may be substituted for elements thereof, and any modifications, equivalents, improvements and changes may be made without departing from the spirit and principles of the present invention.

Claims

1. The lower limb rehabilitation training system based on near infrared brain function imaging is characterized by comprising a lower limb exoskeleton robot, data acquisition equipment and an upper computer;

the upper computer obtains corresponding rehabilitation training evaluation index data by obtaining activation information of different brain areas, optimizes the data by an optimization algorithm, sends control parameters to the lower limb exoskeleton robot, and obtains final auxiliary parameters of the lower limb exoskeleton robot through multiple iterations.

2. The lower limb rehabilitation training system according to claim 1, further comprising a textile headgear, wherein a plurality of pairs of the near infrared light emitting module and the photoelectric conversion receiving module are distributed on the textile headgear.

3. The lower limb rehabilitation training system according to claim 1, wherein the thigh bar and the shank bar are telescopic, and thigh binding and shank binding are further provided on the thigh bar and the shank bar, respectively.

4. The lower limb rehabilitation training system according to claim 1, wherein the optimization algorithm employs an on-loop black box optimization algorithm.

5. The lower limb rehabilitation training method based on near infrared brain function imaging is characterized by comprising an auxiliary mode parameterization step and a human in-loop optimization algorithm step according to evaluation data;

6. The lower limb rehabilitation training method according to claim 5, wherein the position control strategy specifically comprises: the lower limb exoskeleton robot directly drives the lower limb of a person to move to a designated position, a walking gait with a planned track in advance is realized, three motors on the hip joint, the knee joint and the ankle joint are cooperatively matched, and in a gait cycle, the rotation angles of the three motors follow the joint angle data of the normal gait of the human body after the corresponding parameters are scaled.

7. The lower limb rehabilitation training method according to claim 5, wherein the assistance control strategy is specifically that the lower limb exoskeleton robot gives corresponding assistance in different stages of walking of a person, learns periodic signals by using an adaptive oscillator to obtain a current gait phase, and directly obtains a required assistance value according to a predefined parameterized assistance curve.

8. The method of claim 5, wherein the evaluation index is a functional connection strength.