A multi-motion mode control system and method for a rehabilitation training exoskeleton robot

Abstract

The present invention proposes a multi - motion - mode control system and method for a lower - limb rehabilitation training exoskeleton robot. The surface electromyogram signal is acquired through the electromyogram sensor of the robot; the surface electromyogram signal is input into a neuromuscular activation model to generate muscle activation degree; the human - machine interaction torque is acquired by using a pressure sensor; the surface electromyogram signal eigenvalue RMS, muscle activation degree and human - machine interaction torque are input into a fuzzy adaptive switching signal regulator to obtain a training mode switching signal; the training modes include a passive training mode, an assisted training mode, an active training mode and a resistance training mode; according to the training mode switching signal, the corresponding training mode is switched, so as to perform training control on the lower - limb exoskeleton rehabilitation robot under the corresponding training mode. The present invention evaluates the patient's motion state according to the patient's physiology and force signals, performs fuzzy rule reasoning to achieve adaptive hierarchical switching, enhances the rehabilitation effect, and makes the rehabilitation training more in line with the individual needs of the patient.

Description

Technical Field

[0001] The present invention relates to the technical field of rehabilitation training, and particularly to a multi-motion mode control system and method for a rehabilitation training exoskeleton robot. Background Art

[0002] Stroke is an acute cerebrovascular event closely related to brain tissue damage and is one of the main causes of global death and disability. Statistical data shows that in the past 30 years, the incidence of stroke has increased by 70%, and the overall disability rate exceeds 80%. Stroke often leads to severe motor dysfunction, especially limited lower limb movement, significantly affecting the patient's independent walking ability and quality of daily life.

[0003] Research shows that systematic rehabilitation training helps improve muscle strength, promote nerve remodeling, and accelerate blood circulation. However, traditional rehabilitation methods mainly rely on the patient's self-training or the assistance of rehabilitation therapists, with problems such as low patient active participation, high training repeatability, and limited human resources, resulting in difficult-to-guarantee rehabilitation effects. Therefore, rehabilitation robots that can accurately and timely respond to the patient's movement intention and provide precise and efficient training programs have gradually become a research hotspot.

[0004] However, the existing rehabilitation robot training methods have the following problems: 1) The training mode switch depends on manual intervention and lacks intelligent adjustment: Traditional rehabilitation robots usually rely on rehabilitation therapists to manually adjust the training mode according to the patient's rehabilitation progress. The mode switch lacks intelligent adjustment, affecting the coherence and adaptability of the training effect. 2) Lack of a real-time evaluation mechanism, resulting in insufficient basis for training mode switching, that is, the existing in-bed lower limb exoskeleton rehabilitation system has limited monitoring of the patient's motion state and lacks real-time evaluation of multi-source information such as surface electromyogram (sEMG) and human-machine interaction force, resulting in unscientific training mode switching. 3) Although the existing in-bed lower limb exoskeleton rehabilitation robots can provide assisted training, they generally have insufficiently rich training modes, lack multi-mode collaboration, and lack adaptive adjustment for different rehabilitation stages, making it difficult to meet the needs of different individuals and rehabilitation stages. Common rehabilitation robot training modes are few, such as single passive training or active training, which cannot cover the different needs of the entire rehabilitation cycle. At different stages of the patient's recovery, relying solely on a single mode may lead to insufficient or excessive training intensity, affecting the rehabilitation effect. 4) Traditional rehabilitation systems usually operate each training mode according to fixed rules, with modes being independent of each other, lacking mode fusion and dynamic adjustment capabilities, and the modes operate independently without collaborative optimization.

[0005] The control strategy of the present invention comprehensively considers multi-source information such as the patient's motor ability, electromyogram signal, and human-machine interaction force, ensures the reasonable switching of the training mode, and dynamically adjusts the training parameters to optimize the rehabilitation effect. At the same time, this control method is applied to the lower limb exoskeleton rehabilitation robot in bed, enabling the robot to provide accurate and personalized training programs and the most appropriate training intensity at different rehabilitation stages, improving the intelligent level and adaptability of the training, and accelerating the recovery of the patient's lower limb function. Summary of the Invention

[0006] The purpose of the present invention is to provide a multi-motion mode control system or method for a rehabilitation robot that can adapt to the changes in the patient's motion state in real time, automatically switch to the corresponding training mode, and enhance the rehabilitation effect, so as to improve the problems of the current lower limb exoskeleton rehabilitation robot in bed, such as single training mode, fixed control strategy, lack of adaptive adjustment for different rehabilitation stages, and difficulty in meeting the needs of different individuals and rehabilitation stages.

[0007] To achieve the above object, the present invention proposes a multi-motion mode control system for a lower limb rehabilitation training exoskeleton robot, including a lower limb exoskeleton rehabilitation robot body in bed, a neuromuscular activation model, a motion state monitoring module, and a training module:

[0008] The robot body obtains the human-machine interaction torque of the patient's trained lower limb through a pressure sensor, and obtains the surface electromyogram signal of the trained lower limb through an electromyogram sensor;

[0009] The neuromuscular activation model is used to convert the surface electromyogram signal into muscle activation;

[0010] The motion state monitoring module introduces an adaptive training mode switching method based on fuzzy rule inference, and uses a fuzzy adaptive switching signal regulator to generate a training mode switching signal from the surface electromyogram signal eigenvalue, muscle activation, and human-machine interaction torque;

[0011] The training module is used to receive the training mode switching signal and generate the corresponding training mode;

[0012] The training modes include a passive training mode, an assisted training mode, an active training mode, and a resistance training mode; the hierarchical switching mechanism of the training modes is increment or decrement step by step.

[0013] Furthermore, the robot training control method for the passive training mode is:

[0014] 1.1: Subtract the actual trajectory fed back by the robot encoder from the expected trajectory of the training task to obtain the trajectory error, and the calculation formula is:

[0015] ;

[0016] Among them, is the trajectory error, is the expected trajectory angle of the training task, is the actual angle of the joint motor of the lower limb exoskeleton robot in bed;

[0017] 1.2: Input the trajectory error into the sliding mode controller to obtain the expected joint auxiliary torque of the robot's motion. The calculation formula is:

[0018] ;

[0019] Among them, respectively represent the trajectory error and the change rate of the trajectory error, represents the sliding mode surface function, represents the modeling error of the system, represents the positive definite inertia matrix of the system, represents the centrifugal force and Coriolis force matrix of the system, represents the gravity matrix of the system, respectively represent the joint motor angle, angular velocity, and angular acceleration, is the controller parameter, represents the expected joint auxiliary torque of the robot's motion.

[0020] 1.3: Perform passive training control on the lower limb exoskeleton rehabilitation robot in bed according to the expected joint auxiliary torque. The control method is specifically as follows: 1) If the expected joint auxiliary torque is positive, the motion trend of the lower limb exoskeleton in bed is the same as the gravity direction; 2) If the expected joint auxiliary torque is negative, the motion trend of the lower limb exoskeleton in bed is opposite to the gravity direction.

[0021] Furthermore, the robot training control method in the assisted training mode is:

[0022] 2.1: Input the expected trajectory of the training task into the robot inverse dynamics model to obtain the expected joint torque of the lower limb exoskeleton rehabilitation robot in bed. The calculation formula is:

[0023] [[ID=4,7]]+ ;

[0024] Among them, is the expected joint torque, <( respectively represent the kinetic energy and potential energy of the robot joint under the Lagrange method, are the angle and angular velocity in the expected trajectory of the training task.

[0025] 2.2: Obtain the human-machine interaction torque through the pressure sensor of the robot;

[0026] 2.3: Input the desired joint torque and the human-machine interaction torque into the motion ability evaluation module to calculate and generate the assistance factor;

[0027] 2.4: Subtract the desired trajectory of the training task from the actual trajectory fed back by the robot encoder to obtain the trajectory error. The trajectory error is input into the sliding mode controller to generate the desired joint assistance torque for the robot's motion;

[0028] 2.5: Input the obtained desired joint torque, desired joint assistance torque, and assistance factor into the assistance training controller to obtain the joint assistance torque for the robot's motion;

[0029] 2.6: Perform assistance training control on the lower limb exoskeleton rehabilitation robot in bed according to the joint assistance torque.

[0030] Furthermore, the method for calculating the assistance factor by the motion ability evaluation module through the desired joint torque and the human-machine interaction torque includes the following steps:

[0031] 2.31: Calculate the patient state volatility through the human-machine interaction torque. The calculation formula is as follows:

[0032] ;

[0033] 2.32: Calculate the patient torque deviation rate through the human-machine interaction torque. The calculation formula is as follows:

[0034] ;

[0035] 2.33: Calculate the assistance factor through the patient state volatility and the torque deviation rate using the following formula;

[0036] ;

[0037] Wherein, represents the patient state volatility, represents the patient torque deviation rate, is the window period, is the current moment, is the human-machine interaction torque, is the desired joint torque; represents the assistance factor, is the weighting factor;

[0038] In step 2.5, input the desired joint torque, the desired joint assistance torque, and the assistance factor into the assistance training controller to obtain the joint assistance torque for the robot's motion. The calculation formula is:

[0039] ;

[0040] Wherein, Denotes the joint assistance torque, is the desired joint assistance torque, is the desired joint torque, is the assistance factor.

[0041] Furthermore, the robot training control method in the active training mode is as follows:

[0042] 3.1: Input the actual trajectory feedback by the robot encoder into the active training controller to obtain the joint assistance torque. The formula is as follows:

[0043] ;

[0044] Wherein, is the joint assistance torque, are respectively the inertia, damping and stiffness parameter matrices of the robot, respectively represent that the actual trajectory feedback by the robot encoder includes the actual angle, actual angular velocity and actual angular acceleration.

[0045] 3.2: Conduct active training control on the lower limb exoskeleton rehabilitation robot in bed according to the joint assistance torque.

[0046] Furthermore, the robot training control method in the resistance training mode is as follows:

[0047] 4.1: Obtain the human-machine interaction torque through the robot pressure sensor;

[0048] 4.2: Input the human-machine interaction torque into the virtual impedance model to obtain the trajectory correction amount for adjusting the desired trajectory of the robot;

[0049] 4.3: Subtract the trajectory correction amount from the desired trajectory of the training task to obtain the adjusted desired trajectory;

[0050] 4.4: Subtract the actual trajectory feedback by the robot encoder from the adjusted desired trajectory to obtain the trajectory error;

[0051] 4.5: Input the obtained trajectory error into the sliding mode controller to obtain the desired joint assistance torque for the robot movement;

[0052] 4.6: Conduct resistance training control on the lower limb exoskeleton rehabilitation robot in bed according to the desired joint assistance torque.

[0053] Furthermore, the virtual impedance model is:

[0054] ;

[0055] Wherein, respectively represent the joint mass, virtual damping coefficient, virtual stiffness coefficient, represents the human - machine interaction force, respectively represent the trajectory correction amount, the angular velocity of the joint motor, and the angular acceleration of the joint motor.

[0056] Further, the neuromuscular activation model includes a neural activation function and a muscle activation function;

[0057] The neural activation function is:

[0058] ;

[0059] The muscle activation function is:

[0060] ;

[0061] Wherein, represents the current moment of the neural activation function, is the electromyogram signal, are respectively the neural activation coefficients, is the time delay, is the sampling time window period; represents the current moment of the muscle activation function, is the muscle activation coefficient.

[0062] Further, the fuzzy adaptive switching signal regulator includes input variables, output variables, and fuzzy rules. By establishing corresponding fuzzy rules, the motion state of the patient is evaluated based on the physiological and mechanical signals of the patient, and fuzzy rule reasoning is performed to output the most suitable training mode for switching to the corresponding motion state.

[0063] The input variables include three key physiological and mechanical parameters: the root mean square (RMS) of the surface electromyogram signal characteristics, the muscle activation degree, and the human - machine interaction torque;

[0064] The output variable includes a training mode switching signal;

[0065] In the fuzzy rules, the fuzzy subsets of the input variables include low (L), medium (M), and high (H); the fuzzy subsets of the output variables include passive, assistive, active, and impedance;

[0066] The passive training mode, assisted training mode, active training mode, and resistance training mode are integrated into a control architecture to achieve collaborative optimization among the modes. Among them, the passive training, assisted training, and resistance training modes all adopt a trajectory tracking control method based on sliding mode control. By combining multi-channel data such as lower limb surface electromyogram signals (sEMG), human-machine interaction force, and motion trajectory, the training mode can be optimized in real time according to the patient's state, enabling the overall control model to have dynamic adjustment capabilities and effectively reducing the computational complexity and power consumption of the robot.

[0067] The passive training mode, assisted training mode, active training mode, and resistance training mode respectively correspond to the acute phase, initial recovery phase, functional recovery phase, and strengthening phase of the patient's lower limb rehabilitation; corresponding rehabilitation training modes and control strategies are designed according to the four stages of lower limb rehabilitation (acute phase, initial recovery phase, functional recovery phase, and strengthening phase), namely passive training, assisted training, active training, and resistance training. By integrating passive training, assisted training, active training, and resistance training, a training system covering the four stages of lower limb rehabilitation (acute phase, initial recovery phase, functional recovery phase, and strengthening phase) is formed. By constructing a multi-mode fusion framework, full-cycle rehabilitation coverage is achieved, enabling the patient to smoothly transition between different modes and ensuring that the training effect is more scientific and reasonable.

[0068] According to the characteristics of physiological signals and force signals under different training task trajectories, the corresponding physiological parameter trapezoidal membership functions are set respectively to dynamically adjust the training mode switching strategy. The formula is as follows:

[0069] ;

[0070] Among them, The input variables include three key physiological and mechanical parameters: the root mean square (RMS) of the surface electromyogram signal feature value, muscle activation degree, and human-machine interaction torque. The output variable includes the training mode switching signal.

[0071] The training mode switching adopts a hierarchical switching mechanism and can only be switched by gradually increasing or decreasing, that is, the training mode can only be switched gradually in an increasing order of "passive training mode → assisted training mode → active training mode → resistance training mode" or a decreasing order of "resistance training mode → active training mode → assisted training mode → passive training mode" to prevent mode "jumping" and avoid injuries caused by the muscles not being adapted.

[0072] The present invention also proposes a multi-motion mode control method for a lower limb rehabilitation training exoskeleton robot. Based on the above-mentioned multi-motion mode control system for a lower limb rehabilitation training exoskeleton robot, the method includes the following steps:

[0073] S1: Surface electromyographic signals are obtained through the electromyographic sensors of the lower limb exoskeleton robot on the bed;

[0074] S2: Extract the RMS of surface electromyography signal eigenvalues;

[0075] S3: inputting the surface electromyographic signal into a neuromuscular activation model to generate a muscle activation degree;

[0076] S4: Use the robot's pressure sensor to obtain the human-robot interaction torque;

[0077] S5: Inputting the surface electromyography signal characteristic value RMS, muscle activation degree and human-computer interaction torque into the motion state monitoring module, and obtaining the training mode switching signal by using the fuzzy adaptive switching signal regulator;

[0078] S6: The training modes include passive training mode, assisted training mode, active training mode and resistance training mode; according to the training mode switching signal, switch to the corresponding training mode, so as to perform training control on the lower limb exoskeleton rehabilitation robot in the corresponding training mode.

[0079] Compared with the prior art, the advantages of the present invention are:

[0080] 1. The control strategy of the present invention introduces an adaptive training mode switching method based on fuzzy rule reasoning. It combines the three key physiological and mechanical parameters of the lower limb surface electromyographic signal characteristic value (RMS), muscle activation degree and human-computer interaction torque to establish corresponding fuzzy rules. The patient's motion state is evaluated according to the patient's physiological and force signals, and the fuzzy rule reasoning output is switched to the most appropriate training mode under the corresponding motion state. It can adapt to the changes in the patient's motion state in real time, realize adaptive switching to the corresponding training mode, enhance the rehabilitation effect, make rehabilitation training more in line with the patient's individual needs, enable the robot to provide accurate and personalized training plans and the most appropriate training intensity at different rehabilitation stages, improve the intelligence level and adaptability of training, and accelerate the recovery of the patient's lower limb function.

[0081] 2. The method of the present invention adopts a rule-based hierarchical switching mechanism. The training mode can only increase or decrease step by step, preventing mode "jumping" and avoiding the occurrence of skipping levels, thereby avoiding injuries caused by muscle failure to adapt and improving the safety of the control method.

[0082] 3. Based on the four stages of lower limb rehabilitation for bedridden patients, namely the acute stage, the initial recovery stage, the functional recovery stage and the strengthening stage, the present invention designs four rehabilitation exercise modes, namely passive training, assisted training, active training and resistance training, and integrates them into one, constructing a multi-modal fusion framework to achieve full-cycle rehabilitation coverage, enabling patients to smoothly transition between different modes and ensuring a more scientific and reasonable training effect.

[0083] 4. The control strategy of the present invention comprehensively considers multi-source information such as the patient's motor ability, EMG signal, and human-machine interaction force, ensures the reasonable switching of the training mode, and dynamically adjusts the training parameters to optimize the rehabilitation effect. At the same time, this control method is applied to the lower limb exoskeleton rehabilitation robot in bed, enabling the robot to provide accurate and personalized training programs and the most suitable training intensity at different rehabilitation stages, improving the intelligent level and adaptability of the training, and accelerating the recovery of the patient's lower limb function.

[0084] 5. According to the characteristics of physiological signals and force signals under different training task trajectories, the present invention respectively sets corresponding membership functions, dynamically adjusts the training mode switching strategy, and ensures the accuracy of the judgment of the patient's motion state and the rationality of the training mode switching under different training tasks.

[0085] 6. In the present invention, four training modes are integrated into a control architecture, adopting a unified basic control loop and a unified input-output channel to achieve collaborative optimization between modes, enabling the overall control model to have the ability of dynamic adjustment, and effectively reducing the computational complexity and power consumption of the robot. BRIEF DESCRIPTION OF THE DRAWINGS

[0086] Figure 1 is the overall architecture diagram of the multi-motion mode control system of the rehabilitation training exoskeleton robot of the present invention;

[0087] Figure 2 is the architecture diagram of the adaptive hierarchical mode switching method based on multi-motion modes in the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0088] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be further described below.

[0089] In response to the rehabilitation training needs of bedridden elderly people, this embodiment proposes a multi-motion mode control system for a rehabilitation training exoskeleton robot, as shown in Figure 1 and Figure 2 shown, the system includes a lower limb exoskeleton rehabilitation robot body in bed, a neuromuscular activation model, a motion state monitoring module, and a training module.

[0090] Among them, the lower limb exoskeleton rehabilitation robot body in bed obtains the human-machine interaction torque of the patient's trained lower limb through a pressure sensor, and obtains the surface EMG signal of the trained lower limb through an EMG sensor; the neuromuscular activation model is used to convert the surface EMG signal into muscle activation degree; in this embodiment, the neuromuscular activation model includes a neural activation function and a muscle activation function, specifically:

[0091] The neural activation function is:

[0092] ;

[0093] The muscle activation function is:

[0094] ;

[0095] Among them, represents the neural activation function at the current moment of, is the electromyogram signal, are the neural activation coefficients respectively, is the time delay, is the sampling time window period, represents the current moment of the muscle activation function, is the muscle activation coefficient.

[0096] The human-machine interaction torque, the root mean square (RMS) of the surface electromyogram signal features, and the muscle activation degree are input into the motion state monitoring module. The motion state monitoring module introduces an adaptive training mode switching method based on fuzzy rule inference. A fuzzy adaptive switching signal regulator is used to generate a training mode switching signal from the surface electromyogram signal features, the muscle activation degree, and the human-machine interaction torque, and complete the adaptive hierarchical switching of the training mode. The training module is used to receive the training mode switching signal and generate the corresponding training mode. According to the four stages of lower limb rehabilitation for bedridden patients, namely the acute stage, the initial recovery stage, the functional recovery stage, and the strengthening stage, the control strategy of the training mode is designed as a passive training mode, an assisted training mode, an active training mode, and a resistance training mode. At the same time, the hierarchical switching mechanism of this training mode can only be incremented or decremented step by step, that is, the training mode can only be switched incrementally in the order of "passive training mode → assisted training mode → active training mode → resistance training mode" or decremented step by step in the order of "resistance training mode → active training mode → assisted training mode → passive training mode" to prevent mode "jumping" and avoid injuries caused by the muscles not being adapted. By integrating passive training, assisted training, active training, and resistance training, a training system covering the four stages of lower limb rehabilitation (acute stage, initial recovery stage, functional recovery stage, and strengthening stage) is formed. By constructing a multi-mode fusion framework, full-cycle rehabilitation coverage is achieved, enabling patients to smoothly transition between different modes and ensuring that the training effect is more scientific and reasonable.

[0097] Meanwhile, in this embodiment, the passive training mode, assisted training mode, active training mode, and resistance training mode all adopt a unified basic control loop and a unified input-output channel to achieve collaborative optimization among modes. Among them, the passive training, assisted training, and resistance training modes all adopt a trajectory tracking control method based on sliding mode control. By combining multi-channel data such as surface electromyography (sEMG) signals of the lower limbs, human-computer interaction force, and motion trajectory, the training mode can be optimized in real time according to the patient's state, enabling the overall control model to have dynamic adjustment capabilities and effectively reducing the computational complexity and power consumption of the robot.

[0098] In this embodiment, as Figure 1 shown, the robot training control methods for the four training modes of passive training mode, assisted training mode, active training mode, and resistance training mode are as follows:

[0099] (1) The robot training control method for the passive training mode is:

[0100] 1.1: Subtract the actual trajectory fed back by the robot encoder from the desired trajectory of the training task. Among them, the desired trajectory includes the desired trajectory angle , desired trajectory angular velocity and desired trajectory angular acceleration ;

[0101] The trajectory error calculation formula is:

[0102] ;

[0103] Among them, is the trajectory error, is the desired trajectory angle of the training task, is the actual angle of the joint motor of the lower limb exoskeleton robot in bed;

[0104] 1.2: Input the trajectory error into the sliding mode controller to obtain the desired joint auxiliary torque for the robot movement. The specific calculation formula is as follows:

[0105] ;

[0106] Among them, respectively represent the trajectory error and the change rate of the trajectory error, represents the sliding mode surface function, represents the modeling error of the system, represents the positive definite inertia matrix of the system, represents the centrifugal force and Coriolis force matrix of the system, represents the gravity matrix of the system, respectively represent the joint motor angle, angular velocity, and angular acceleration, is the controller parameter, indicating the desired joint assistance torque for the robot motion.

[0107] 1.3: Perform passive training control on the lower limb exoskeleton rehabilitation robot in bed according to the desired joint assistance torque. The specific control method is as follows: 1) If the obtained desired joint assistance torque is positive, the motion trend of the lower limb exoskeleton in bed is the same as the direction of gravity; 2) If the obtained desired joint assistance torque is negative, the motion trend of the lower limb exoskeleton in bed is opposite to the direction of gravity.

[0108] (2) The robot training control method for the assisted training mode is:

[0109] 2.1: Input the desired trajectory of the training task into the robot inverse dynamics model to obtain the desired joint torque for the motion of the lower limb exoskeleton rehabilitation robot in bed , and the calculation formula is:

[0110] + ;

[0111] where, is the desired joint torque, respectively represent the kinetic energy and potential energy of the robot joints under the Lagrangian method, are the angles and angular velocities in the desired trajectory of the training task.

[0112] 2.2: Obtain the human-machine interaction torque through the pressure sensor of the robot ;

[0113] 2.3: Input the desired joint torque and the human-machine interaction torque into the motion ability evaluation module to calculate and generate the assistance factor , and its specific calculation method is as follows:

[0114] 2.31: Calculate the patient state volatility through the human-machine interaction torque, and the calculation formula is as follows:

[0115] ;

[0116] where, represents the patient state volatility, is the window period, is the current moment, is the human-machine interaction torque;

[0117] 2.32: Calculate the patient torque deviation rate through the human-machine interaction torque, and the calculation formula is as follows:

[0118] ;

[0119] Wherein, represents the patient torque offset rate, is the window period, is the current moment, is the human-machine interaction torque, is the desired joint torque;

[0120] 2.33: Calculate the assistance factor through the patient state volatility and the torque offset rate , and the formula is as follows;

[0121] ;

[0122] Wherein, represents the assistance factor, is the weighting factor, is the patient state volatility, is the patient torque offset rate.

[0123] 2.4: Subtract the desired trajectory of the training task from the actual trajectory fed back by the robot encoder to obtain the trajectory error , and input the trajectory error into the sliding mode controller to generate the desired joint assistance torque for the robot motion;

[0124] 2.5: Input the obtained desired joint torque , the desired joint assistance torque and the assistance factor into the assistance training controller to obtain the joint assistance torque[[ID=SS]] for the robot motion; The specific calculation formula is as follows:

[0125] ;

[0126] Wherein, represents the joint assistance torque, is the desired joint assistance torque, is the desired joint torque, is the assistance factor;

[0127] 2.6: Perform assistance training control on the lower limb exoskeleton rehabilitation robot in bed according to the joint assistance torque .

[0128] (3) The robot training control method in the active training mode is as follows:

[0129] 3.1: The actual trajectory fed back by the robot encoder Input it into the active training controller to obtain the joint assistance torque , and the specific calculation formula is as follows:

[0130] ;

[0131] Among them, is the joint assistance torque, are respectively the inertia, damping, and stiffness parameter matrices of the robot, respectively represent the actual angle, actual angular velocity, and actual angular acceleration feedback by the robot encoder.

[0132] 3.2: Perform active training control on the lower limb exoskeleton rehabilitation robot in bed according to the joint assistance torque;

[0133] The robot training control method in the resistance training mode is:

[0134] 4.1: Obtain the human-machine interaction torque through the robot pressure sensor ;

[0135] 4.2: Input the human-machine interaction torque into the virtual impedance model to obtain the trajectory correction amount for adjusting the desired trajectory of the robot; among them, the virtual impedance model is:

[0136] ;

[0137] Among them, respectively represent the joint mass, virtual damping coefficient, and virtual stiffness coefficient, represents the human-machine interaction force, respectively represent the trajectory correction amount, joint motor angular velocity, and joint motor angular acceleration;

[0138] 4.3: Subtract the trajectory correction amount from the desired trajectory of the training task to obtain the adjusted desired trajectory, including the angle of the desired trajectory after being adjusted by the virtual impedance model and the angular velocity of the adjusted desired trajectory in the resistance training mode;

[0139] 4.4: Subtract the adjusted desired trajectory from the actual trajectory feedback by the robot encoder to obtain the trajectory error , and the formula is:

[0140] ;

[0141] Among them, is the trajectory error, is the angle of the adjusted desired trajectory, is the actual angle of the joint motor of the lower limb exoskeleton robot in bed.

[0142] 4.5: Input the obtained trajectory error into the sliding mode controller to obtain the expected joint auxiliary torque for the robot's movement , thereby achieving the effect of resistance;

[0143] 4.6: Conduct resistance training control on the lower limb exoskeleton rehabilitation robot in bed according to the expected joint auxiliary torque.

[0144] In this embodiment, the fuzzy adaptive switching signal regulator includes input variables, output variables, and fuzzy rules. By establishing corresponding fuzzy rules, the patient's motion state is evaluated based on the patient's physiological and mechanical signals, and fuzzy rule reasoning is performed to output the most suitable training mode for switching to the corresponding motion state.

[0145] Among them, the input variables include three key physiological and mechanical parameters: the root mean square (RMS) of the surface electromyogram signal feature value, muscle activation degree, and human-machine interaction torque; the output variable includes the training mode switching signal;

[0146] In the fuzzy rules, the fuzzy subsets of the input variables include low (L), medium (M), and high (H); the fuzzy subsets of the output variables include passive, assistive, active, and impedance;

[0147] The input variable-output variable mapping relationship is shown in Table 1:

[0148] Table 1

[0149] RMS value Muscle activation Human-machine interaction torque Training mode L L L Passive L L M Assistive L L H Assistive L M L Passive L M M Assistive L M H Active L H L Passive L H M Assistive L H H Active M L L Passive M L M Assistive M L H Active M M L Passive M M M Assistive M M H Active M H L Passive M H M Assistive M H H Active H L L Passive H L M Assistive H L H Active H M L Passive H M M Assistive H M H Active H H L Passive H H M Assistive H H H Resistance

[0150] The hierarchical switching rules of the training mode are shown in Table 2:

[0151] Table 2

[0152] Current training mode Output training mode Switching rule Passive mode Passive training Maintain passive training Passive mode Assistive training Passive mode → Assistive training Assistive mode Passive training Assistive training → Passive training Assistive mode Assistive training Maintain assistive training Assistive mode Active training Assistive training → Active training Active mode Assistive training Active training → Assistive training Active mode Active training Maintain active training Active mode Resistance training Active training → Resistance training Resistance mode Active training Resistance training → Active training Resistance mode Resistance training Maintain resistance training

[0153] According to the characteristics of the physiological signals and force signals under different training task trajectories, the corresponding physiological parameter trapezoidal membership functions are set respectively, and the training mode switching strategy is dynamically adjusted. The formula is as follows:

[0154] ;

[0155] Among them, is the input variable including three key physiological and mechanical parameters: the root mean square (RMS) of the surface electromyogram signal feature value, muscle activation degree, and human-machine interaction torque, is the output variable including the training mode switching signal.

[0156] Based on the above multi-motion mode control system of the lower limb rehabilitation training exoskeleton robot, a multi-motion mode control method for the lower limb rehabilitation training exoskeleton robot is provided. The method includes the following steps:

[0157] S1: Obtain surface electromyogram signals through the electromyogram sensors of the lower limb exoskeleton robot in bed;

[0158] S2: Extract the eigenvalue RMS of the surface electromyogram signals;

[0159] S3: Input the surface electromyogram signals into the neuromuscular activation model to generate muscle activation degrees;

[0160] S4: Obtain the human-machine interaction torque using the pressure sensors of the robot;

[0161] S5: Input the eigenvalue RMS of the surface electromyogram signals, the muscle activation degrees, and the human-machine interaction torque into the motion state monitoring module, and use the fuzzy adaptive switching signal regulator to obtain the training mode switching signal;

[0162] S6: The training modes include passive training mode, assisted training mode, active training mode, and resistance training mode; according to the training mode switching signal, switch to the corresponding training mode, so as to perform training control under the corresponding training mode on the lower limb exoskeleton rehabilitation robot in bed.

[0163] The above are only the preferred embodiments of the present invention and do not impose any limitation on the present invention. Any person skilled in the art within the technical field, without departing from the technical solution of the present invention, makes any form of equivalent substitution or modification and other changes to the technical solution and technical content disclosed by the present invention, all of which belong to the content of not departing from the technical solution of the present invention and still fall within the protection scope of the present invention.

Claims

1. A multi-motion mode control system for a lower limb rehabilitation training exoskeleton robot, characterized in that, It includes a lower limb exoskeleton rehabilitation robot body in bed, a neuromuscular activation model, a motion state monitoring module, and a training module: The robot body obtains the human-machine interaction torque of the patient's trained lower limb through a pressure sensor and obtains the surface electromyogram signal of the trained lower limb through an electromyogram sensor; The neuromuscular activation model is used to convert the surface electromyogram signal into muscle activation; The motion state monitoring module introduces an adaptive training mode switching method based on fuzzy rule inference, and uses a fuzzy adaptive switching signal regulator to generate a training mode switching signal from the surface electromyogram signal eigenvalue, muscle activation, and human-machine interaction torque; The training module is used to receive the training mode switching signal and generate a corresponding training mode; The training modes include a passive training mode, an assisted training mode, an active training mode, and a resistance training mode; The hierarchical switching mechanism of the training mode is incremental or decremental step by step; The fuzzy adaptive switching signal regulator includes input variables, output variables, and fuzzy rules; The input variables include the surface electromyogram signal eigenvalue RMS, muscle activation, and human-machine interaction torque; The output variable includes a training mode switching signal; In the fuzzy rules, the fuzzy subsets of the input variables include low (L), medium (M), and high (H); The fuzzy subsets of the output variables include passive, assisted, active, and impedance; The passive training mode, assisted training mode, active training mode, and resistance training mode are integrated into a control architecture; The passive training mode, assisted training mode, active training mode, and resistance training mode respectively correspond to the acute stage, initial recovery stage, functional recovery stage, and strengthening stage of the patient's lower limb rehabilitation; The training mode switching adopts a hierarchical switching mechanism and can only be switched by incremental or decremental step by step.

2. The multi-motion mode control system of the lower limb rehabilitation training exoskeleton robot according to claim 1, wherein The robot training control method for the passive training mode is: 1.1: Subtract the expected trajectory of the training task from the actual trajectory fed back by the robot encoder to obtain a trajectory error; 1.2: Input the trajectory error into a sliding mode controller to obtain the expected joint auxiliary torque of the robot motion; 1.3: Perform passive training control on the lower limb exoskeleton rehabilitation robot in bed according to the expected joint auxiliary torque. The specific control method is: 1) If the expected joint auxiliary torque is positive, the motion trend of the lower limb exoskeleton in bed is the same as the gravity direction; 2) If the expected joint auxiliary torque is negative, the motion trend of the lower limb exoskeleton in bed is opposite to the gravity direction.

3. The multi-motion mode control system of the lower limb rehabilitation training exoskeleton robot according to claim 1, characterized in that, The robot training control method for the assisted training mode is: 2.1: Input the expected trajectory of the training task into the robot inverse dynamics model to obtain the expected joint torque of the lower limb exoskeleton rehabilitation robot in bed; 2.2: Obtain the human-machine interaction torque through the pressure sensor of the robot; 2.3: Input the expected joint torque and the human-machine interaction torque into the motion ability evaluation module to calculate and generate an assistance factor; 2.4: Subtract the expected trajectory of the training task from the actual trajectory fed back by the robot encoder to obtain a trajectory error, and input the trajectory error into a sliding mode controller to generate the expected joint auxiliary torque of the robot motion; 2.5: Input the obtained joint desired torque, joint desired assist torque, and assist factor into the assist training controller to obtain the joint assist torque for the robot's movement. 2.6: Perform assist training control on the lower limb exoskeleton rehabilitation robot in bed according to the joint assist torque.

4. The multi-motion mode control system of the lower limb rehabilitation training exoskeleton robot according to claim 3, characterized in that, The method for calculating the assist factor by the motion ability evaluation module through the joint desired torque and the human-machine interaction torque includes the following steps: 2.31: Calculate the patient state volatility through the human-machine interaction torque, and the calculation formula is as follows: 2.32: Calculate the patient torque deviation rate through the human-machine interaction torque, and the calculation formula is as follows: 2.33: Calculate the assist factor through the patient state volatility and torque deviation rate using the following formula; β = 1 - [A·SFR + (1 - A)·TOR]; Among them, SFR Indicates the patient state volatility, TOR Indicates the patient torque offset rate, where T is the window period, f Is the current moment, τ a Is the human-machine interaction torque, τ d Is the desired joint torque; β represents the assistance factor, and A is the weighting factor; In step 2.5, input the joint desired torque, joint desired assist torque, and assist factor into the assist training controller to obtain the joint assist torque for the robot's movement, and the calculation formula is: τ assisted = τ e - β·τ d ; Among them, τ assisted represents the joint assistance torque, τ e is the desired joint assistance torque, τ d is the desired joint torque, and β is the assistance factor.

5. The multi-motion mode control system for the lower limb rehabilitation training exoskeleton robot according to claim 1, wherein the robot training control method for the active training mode is: 3.1: Input the actual trajectory feedback by the robot encoder into the active training controller to obtain the joint assist torque; 3.2: Perform active training control on the lower limb exoskeleton rehabilitation robot in bed according to the joint assist torque.

6. The multi-motion mode control system for the lower limb rehabilitation training exoskeleton robot according to claim 1, wherein the robot training control method for the resistance training mode is: 4.1: Obtain the human-machine interaction torque through the robot pressure sensor; 4.2: Input the human-machine interaction torque into the virtual impedance model to obtain the trajectory correction amount for adjusting the robot's desired trajectory; 4.3: Subtract the trajectory correction amount from the desired trajectory of the training task to obtain the adjusted desired trajectory; 4.4: Subtract the actual trajectory feedback by the robot encoder from the adjusted desired trajectory to obtain the trajectory error; 4.5: Input the obtained trajectory error into the sliding mode controller to obtain the joint desired assist torque for the robot's movement; 4.6: Perform resistance training control on the lower limb exoskeleton rehabilitation robot in bed according to the joint desired assist torque.

7. The multi-motion mode control system of the lower limb rehabilitation training exoskeleton robot according to claim 6, characterized in that, The virtual impedance model is: where m, k, and p represent the joint mass, virtual damping coefficient, and virtual stiffness coefficient respectively, F represents the human-machine interaction force, and Δq, represent the trajectory correction amount, joint motor angular velocity, and joint motor angular acceleration respectively.

8. The multi-motion mode control system of the lower limb rehabilitation training exoskeleton robot according to claim 1, wherein The neuromuscular activation model includes a neural activation function and a muscle activation function; The neural activation function is: The muscle activation function is: Wherein, μ(t) represents the neural activation function at the current moment t, sEMG is the electromyogram signal, α, β1, and β2 are respectively the neural activation coefficients, d is the time delay, and T is the sampling time window period; u(t) represents the muscle activation function at the current moment t, and A is the muscle activation coefficient.

9. A multi-motion mode control method for a lower limb rehabilitation training exoskeleton robot, based on the multi-motion mode control system of the lower limb rehabilitation training exoskeleton robot according to any one of claims 1-8, characterized in that, It includes the following steps: S1: Obtain the surface electromyogram signal through the electromyogram sensor of the lower limb exoskeleton robot in bed; S2: Extract the surface electromyogram signal eigenvalue RMS; S3: Input the surface electromyogram signal into the neuromuscular activation model to generate the muscle activation degree; S4: Use the pressure sensor of the robot to obtain the human-machine interaction torque; S5: Input the surface electromyogram signal eigenvalue RMS, muscle activation degree, and human-machine interaction torque into the fuzzy adaptive switching signal regulator to obtain a training mode switching signal; S6: The training modes include a passive training mode, an assisted training mode, an active training mode, and a resistance training mode; The training mode switching adopts a hierarchical switching mechanism and can only be switched by gradually increasing or decreasing; according to the training mode switching signal, switch to the corresponding training mode, so as to perform training control in the corresponding training mode for the lower limb exoskeleton rehabilitation robot in bed.

Images