Exoskeleton robot muscle simulation method and device based on bionic electromyographic signals

Abstract

The invention belongs to the technical field of exoskeleton robot muscle simulation, and particularly discloses an exoskeleton robot muscle simulation method and device based on bionic electromyographic signals. According to the method, firstly, a speed-torque curve is obtained, then bionic electromyographic signals are generated through a PD algorithm, the bionic electromyographic signals are preprocessed to obtain motion instruction data, then the motion instruction data are input into a joint motor to drive the joint motor to move, and the output force value of the joint motor is collected; the output force value of the joint motor is compared with the output force value of the established Hille muscle model, and the generation parameters of the bionic electromyographic signals are corrected until the output force value of the joint motor is close to the output force value of the Hille muscle model; the device comprises a curve generation module, a signal generation module, a signal preprocessing module, joint motors of all parts, a signal acquisition module, a model generation module and a correction module. According to the method, the muscle simulation efficiency and accuracy of the exoskeleton robot are effectively improved.

Description

Technical Field

[0001] The invention belongs to the technical field of exoskeleton robot muscle simulation, and in particular relates to an exoskeleton robot muscle simulation method and device based on bionic electromyographic signals. Background Art

[0002] In the process of modern society, robots are gradually changing the production and lifestyle of human beings, and improving the production efficiency and product quality of various industries. Exoskeleton robots refer to robots that are worn outside the human body, also known as "wearable robots". They are integrated technologies that integrate sensing, control, information, fusion, and mobile computing to provide a wearable mechanical mechanism for people as operators. Exoskeletons in the existing technology are mainly divided into two types. One is a human enhancement exoskeleton that assists specific joints of the human body. This exoskeleton is mainly used to increase human strength and expand the upper limit of ability; the other is a rehabilitation exoskeleton, which is mainly used in the field of medical rehabilitation, such as assisting paralyzed patients to walk.

[0003] The Chinese invention patent with the authorization announcement number CN 112587242 B discloses a surgical robot master hand simulation method, master hand and application, which collects the electromyographic signals generated by the target muscles of the human body, processes the electromyographic signals, obtains the movement angles of the metacarpophalangeal joint and the wrist joint, and drives the slave hand end, i.e., the exoskeleton robot, to move. This disclosed invention is used to solve the problems of large systems, complex structures, long training time, and easy fatigue of operators in mechanical master hands. However, in this technology, the electromyographic signals need to be collected from the human body before being processed, which makes the efficiency and accuracy of the exoskeleton robot muscle simulation low; at the same time, because the electromyographic signals need to be collected from the human body, the person being tested needs to cooperate continuously during the collection process, which requires a lot of manpower. Summary of the invention

[0004] In order to solve the above-mentioned deficiencies in the known technology, the present invention aims to provide an exoskeleton robot muscle simulation method based on electrical stimulation. The method is implemented by an exoskeleton robot muscle simulation device based on electrical stimulation. The method drives the movement of joint motors through bionic electromyographic signals, and improves the efficiency and accuracy of exoskeleton robot muscle simulation by collecting the motion data of the joint motors and training them through algorithms.

[0005] Another object of the present invention is to provide an exoskeleton robot muscle simulation device based on electrical stimulation to improve the efficiency and accuracy of exoskeleton robot muscle simulation.

[0006] To achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0007] A method for simulating muscles of an exoskeleton robot based on bionic myoelectric signals comprises the following steps performed in sequence:

[0008] S1. Establish the speed-torque curve required for the joint movement of each part of the exoskeleton robot;

[0009] S2, generating a bionic electromyographic signal according to the speed-torque curve by using the generation parameters of the bionic electromyographic signal and the PD algorithm;

[0010] S3, preprocessing the bionic electromyographic signal to obtain motion command data of the joint motors of each part;

[0011] S4, inputting the motion instruction data of the joint motors of each part into the joint motors of each part respectively;

[0012] The motors of the joints of each part receive the motion instruction data and move, and the output force values ​​of the motors of the joints of each part are collected;

[0013] S5, using the output force value of the Hill muscle model of the joint motor of each part as a training data set, training through the Boltzmann machine algorithm, comparing the actual output force value of the joint motor of each part with the output force value of the Hill muscle model of the joint motor of each part, and updating the generation parameters of the bionic electromyographic signal;

[0014] Establishing the Hill muscle model of the joint motor of each part, and obtaining the output force value of the Hill muscle model of the joint motor of each part is performed in any step before executing step S5;

[0015] S6, if the difference between the output force value of the joint motor of each part and the output force value of the Hill muscle model of the joint motor of each part is less than or equal to 3% of the maximum joint torque, continue to execute step S7;

[0016] Otherwise, jump to step S2, and repeat steps S2 to S5 until the difference in the output force values ​​of the Hill muscle model of the joint motors of each part is less than or equal to 3% of the maximum joint torque;

[0017] S7. Install the joint motors of each part to the corresponding joints of the exoskeleton robot to control the movement of the robot exoskeleton.

[0018] As a limitation, the process of establishing the speed-torque curve required for the movement of each joint of the exoskeleton robot in step S1 includes:

[0019] The human body's hip joint, knee joint, ankle joint, shoulder joint, elbow joint, and wrist joint are taken as the six major motion nodes, and the muscle groups of the six major motion nodes are used as driving force. Through simulation calculation and combining with human working movements, the motion spectrum of the exoskeleton robot's joints is analyzed, and the speed-torque curves required for the movement of each joint of the exoskeleton robot are established.

[0020] As a second limitation, the PD algorithm in step S2 is expressed as:

[0021] A(t)=P*(P cmd -P act )-D*V act ,

[0022] Among them, A(t) is the amplitude of the bionic electromyographic signal, P cmd is the joint motor command position, P act is the actual position of the joint motor, V act is the actual speed of the joint motor, P is the stiffness coefficient of the joint motor, and D is the damping coefficient of the joint motor. The selection of the stiffness coefficient P and the damping coefficient D is determined by the speed-torque curve;

[0023] The generation parameters of the electromyographic signal include stimulation duration, stimulation point direction and stimulation signal amplitude.

[0024] As a third limitation, the bionic electromyographic signal preprocessing in step S3 adopts the following steps performed in sequence:

[0025] a1) Establishing a regression model to perform regression processing on the bionic electromyographic signal;

[0026] a2) amplifying the bionic electromyographic signal after the regression process to obtain an amplified bionic electromyographic signal;

[0027] a3) rectifying the amplified bionic electromyographic signal to obtain a rectified bionic electromyographic signal;

[0028] a4) filtering the bionic myoelectric signal after rectification using a first-order low-pass filtering algorithm to obtain the bionic myoelectric signal after filtering;

[0029] a5) organizing the bionic electromyographic signal after filtering, constructing a numerical model of the bionic electromyographic signal, and determining the effective bionic electromyographic signal according to the numerical model.

[0030] As a fourth limitation, the method for establishing the Hill muscle model of the joint motor of each part in step S5 and obtaining the output force value of the Hill muscle model of the joint motor of each part is:

[0031] The motion data of people walking in the exoskeleton is collected, and a computer is used for simulation to generate electromyographic signals to drive the motors of the joints of various parts. The collected speed, position, and torque of the motors of the joints of various parts of the exoskeleton robot are regressed according to the Hill muscle model to obtain a Hill muscle model with the input of the electromyographic signal amplitude and the output of the muscle force.

[0032] Among them, the Hill muscle model includes a contractile element, a parallel elastic element, and a series elastic element;

[0033] The tension generated by shrinking the original is:

[0034] F cc =f(l)f(v)a

[0035] Among them, f(l) is the instantaneous muscle contraction length coefficient of the contractile element; f(v) is the instantaneous muscle contraction velocity coefficient of the contractile element, and a is the muscle activation state;

[0036] The instantaneous muscle contraction length coefficient f(l) of the contractile element and the instantaneous muscle contraction velocity coefficient f(v) of the contractile element satisfy:

[0037] f(l)=F max (1-(l ce -l ce0 ) 2 / ω 2 (l ce0 ) 2 )

[0038] f(v)=(v ce0 -v ce ) / (v ce0 +(v ce / c))

[0039] Among them, F max is the maximum isometric contraction tension of the contraction element; l ce0 is the optimal length of the contraction element; l ce is the length of the contraction element after the tension changes; ω is the range of force generated by the contraction element; v ce0 is the optimal contraction speed of the contraction element; v ce is the contraction speed of the contraction element after the tension changes, c is the hyperbolic shape factor;

[0040] The tension of the parallel elastic element is:

[0041] F pee =K pee (l ce -l ce0 ) 2

[0042] Among them, K pee is the elastic modulus of the parallel elastic element;

[0043] The damping force generated by the contraction element is:

[0044] f c =Cv ce

[0045] Where C is the damping coefficient of the contraction element;

[0046] The muscle force generated by the i-th target muscle is:

[0047] F i = F ce + F pee + f c ;

[0048] In the Hill muscle model, the main control input variable is the muscle activation state a(t). Suppose the neural activation state at time t is:

[0049] u(t) = k × A(t)(t - d) - l1 × a(t - 1) - l2 × a(t - 2)

[0050] where A(t) is the amplitude of the bionic electromyogram signal, k, l1, and l2 are neural activation coefficients respectively, d is the time delay; among them, the neural activation coefficients satisfy:

[0051] l1 = α1 + α2 (|α1| < 1; |α2| < 1)

[0052] l2 = α1 · α2 (|α1| < 1; |α2| < 1)

[0053] k - l1 - l2 = 1

[0054] where α1 and α2 are electromyogram signal delay coefficients;

[0055] The muscle activation state a(t) is:

[0056]

[0057] where c, d, m, and b are muscle activation coefficients; u(t) is the neural activation state at time t;

[0058] When the muscle activation state a(t) = 0, the target muscle is not activated;

[0059] When the muscle activation state 0 < a(t) < 1, the target muscle is partially activated;

[0060] When the muscle activation state a(t) = 1, the target muscle is fully activated.

[0061] The present invention also discloses an exoskeleton robot muscle simulation device based on electrical stimulation, which is applied to the exoskeleton robot muscle simulation method based on electrical stimulation, and includes a curve generation module, a signal generation module, a signal preprocessing module, joint motors of each part, a signal acquisition module, a model generation module, and a correction module;

[0062] The curve generation module is used to establish the speed-torque curve required for the joint movement of each part of the exoskeleton robot and output it to the bionic electromyogram signal generation module;

[0063] A signal generation module receives the speed-torque curve, generates a bionic electromyographic signal through the generation parameters of the bionic electromyographic signal and the PD algorithm, and transmits the signal to the signal preprocessing module;

[0064] The signal preprocessing module receives the bionic electromyographic signal generated by the signal generating module, preprocesses the bionic electromyographic signal, obtains the motion command data of the joint motors of each part, and transmits the motion command data to the joint motors of each part accordingly;

[0065] The joint motors of each part respectively receive the motion command data of the joint motors of each part generated by the signal preprocessing module and generate motion;

[0066] The signal acquisition module receives the output force value generated by the motor movement of each joint and outputs it to the correction module;

[0067] The model generation module is used to collect the motion data of a person wearing an exoskeleton and perform regression processing on the collected motion data according to the Hill muscle model to obtain the Hill muscle model output force value of the joint motor of each part, and output it to the correction module;

[0068] The correction module receives the output force value of the joint motor of each part output by the signal acquisition module and the output force value of the Hill muscle model of the joint motor of each part output by the model generation module, takes the output force value of the Hill muscle model of the joint motor of each part obtained by the model generation module as a training data set, trains through the Boltzmann machine algorithm, compares the actual output force value of the joint motor of each part with the output force value of the Hill muscle model of the joint motor of each part, and updates the generation parameters of the bionic electromyographic signal.

[0069] Due to the adoption of the above technical solution, the present invention has the following beneficial effects compared with the prior art:

[0070] (1) The device of the present invention inputs bionic electromyographic signals into joint motors, collects motion data of joint motors, and continuously corrects the generation parameters of bionic electromyographic signals to obtain the efficiency and accuracy of exoskeleton robot muscle simulation;

[0071] (2) The method of the present invention generates myoelectric bionic signals by adopting the PD algorithm combined with bionic myoelectric signal generation parameters and speed-torque curves, thereby avoiding the process of directly collecting myoelectric signals from the human body, saving manpower, and at the same time, by directly and continuously correcting the generation parameters of the bionic signal at several points, the output force value of the joint motor of each part is made close to the output force value of the Hill muscle model, thereby improving the efficiency and accuracy of the exoskeleton robot muscle simulation;

[0072] (3) The present invention enables the joint motor to exhibit the motion effect of bionic muscle characteristics, increases the flexibility of the exoskeleton robot during operation, increases the ease of interaction between man and machine and the environment, and avoids impact and collision.

[0073] In summary, the present invention can avoid the continuous cooperation of personnel during the acquisition process and improve the efficiency and accuracy of exoskeleton robot muscle simulation. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0075] Figure 1 This is a flow chart of the method of Example 1 of the present invention;

[0076] Figure 2 This is a speed-torque curve diagram of Example 1 of the present invention;

[0077] Figure 3 It is an output curve diagram of the joint motor and the Hill muscle model of Example 1 of the present invention;

[0078] Figure 4 This is a principle block diagram of embodiment 2 of the present invention. DETAILED DESCRIPTION

[0079] In order to better explain the present invention and facilitate understanding, the preferred embodiments of the present invention are described in detail below through specific implementation modes in conjunction with the accompanying drawings.

[0080] Example 1 A method for simulating muscles of an exoskeleton robot based on electrical stimulation

[0081] This embodiment provides an exoskeleton robot muscle simulation method based on electrical stimulation, such as Figure 1 The embodiment shown includes the following steps performed in sequence:

[0082] S1. Establish the speed-torque curve required for the joint movement of each part of the exoskeleton robot;

[0083] S2, generating a bionic electromyographic signal according to the speed-torque curve by using the generation parameters of the bionic electromyographic signal and the PD algorithm;

[0084] S3, preprocessing the bionic electromyographic signal to obtain motion command data of the joint motors of each part;

[0085] S4, inputting the motion instruction data of the joint motors of each part into the joint motors of each part respectively;

[0086] The motors of the joints of each part receive the motion instruction data and move, and the output force values ​​of the motors of the joints of each part are collected;

[0087] S5, using the output force value of the Hill muscle model of the joint motor of each part as a training data set, training through the Boltzmann machine algorithm, comparing the actual output force value of the joint motor of each part with the output force value of the Hill muscle model of the joint motor of each part, and updating the generation parameters of the bionic electromyographic signal;

[0088] Establishing the Hill muscle model of the joint motor of each part, and obtaining the output force value of the Hill muscle model of the joint motor of each part is performed in any step before executing step S5;

[0089] S6, if the difference between the output force value of the joint motor of each part and the output force value of the Hill muscle model of the joint motor of each part is less than or equal to 3% of the maximum joint torque, continue to execute step S7;

[0090] Otherwise, jump to step S2, and repeat steps S2 to S5 until the difference in the output force values ​​of the Hill muscle model of the joint motors of each part is less than or equal to 3% of the maximum joint torque;

[0091] S7. Install the joint motors of each part to the corresponding joints of the exoskeleton robot to control the movement of the robot exoskeleton.

[0092] In this embodiment, the process of establishing the speed-torque curve required for the movement of each joint of the exoskeleton robot in step S1 includes: taking the human hip joint, knee joint, ankle joint, shoulder joint, elbow joint, and wrist joint as the six major motion nodes, and the muscle groups of the six major motion nodes as the driving force, through simulation calculation, combining the human working movements to analyze the motion map of the exoskeleton robot joints, and establishing the speed-torque curve required for the movement of each joint of the exoskeleton robot.

[0093] The PD algorithm used in step S2 is expressed as:

[0094] A(t)=P*(P cmd -P act )-D*V act

[0095] Among them, A(t) is the amplitude of the bionic electromyographic signal, P cmd is the joint motor command position, P act is the actual position of the joint motor, V act is the actual speed of the joint motor, P is the stiffness coefficient of the joint motor, and D is the damping coefficient of the joint motor. The selection of the stiffness coefficient P and the damping coefficient D is determined by the speed-torque curve, with the rated speed as the boundary. When the actual speed of the joint motor is less than the rated speed, the speed-torque curve adopts the strategy of taking small P and large D. When the actual speed of the joint motor is greater than the rated speed, the speed-torque curve adopts the strategy of taking large P and small D. The generation parameters of the electromyographic signal include the stimulation duration, the direction of the stimulation point and the amplitude of the stimulation signal.

[0096] The bionic electromyographic signal preprocessing in step S3 is performed in the following steps in sequence:

[0097] a1) Establishing a regression model to perform regression processing on the bionic electromyographic signal;

[0098] a2) amplifying the bionic electromyographic signal after the regression process to obtain an amplified bionic electromyographic signal;

[0099] a3) rectifying the amplified bionic electromyographic signal to obtain a rectified bionic electromyographic signal;

[0100] a4) filtering the bionic myoelectric signal after rectification using a first-order low-pass filtering algorithm to obtain the bionic myoelectric signal after filtering;

[0101] a5) organizing the bionic electromyographic signal after filtering, constructing a numerical model of the bionic electromyographic signal, and determining the effective bionic electromyographic signal according to the numerical model.

[0102] The Hill muscle model of each joint motor in step S5 can be obtained in any step before step S5. The method for establishing the Hill muscle model of each joint motor and obtaining the output force value of the Hill muscle model of each joint motor is as follows:

[0103] The motion data of people walking in the exoskeleton is collected, and a computer is used for simulation to generate electromyographic signals to drive the motors of the joints of various parts. The collected speed, position, and torque of the motors of the joints of various parts of the exoskeleton robot are regressed according to the Hill muscle model to obtain a Hill muscle model with the input of the electromyographic signal amplitude and the output of the muscle force.

[0104] Among them, the Hill muscle model includes a contractile element, a parallel elastic element, and a series elastic element;

[0105] The tension generated by shrinking the original is:

[0106] F cc =f(l)f(v)a

[0107] Among them, f(l) is the instantaneous muscle contraction length coefficient of the contractile element; f(v) is the instantaneous muscle contraction velocity coefficient of the contractile element, and a is the muscle activation state;

[0108] The instantaneous muscle contraction length coefficient f(l) of the contractile element and the instantaneous muscle contraction velocity coefficient f(v) of the contractile element satisfy:

[0109] f(l)=F max (1-(l ce -l ce0 ) 2 / ω 2 (l ce0 ) 2 )

[0110] f(v)=(v ce0 -v ce) / (v ce0 +(v ce / c))

[0111] Among them, F max is the maximum isometric contraction tension of the contraction element; l ce0 is the optimal length of the contraction element; l ce is the length of the contraction element after the tension changes; ω is the range of force generated by the contraction element; v ce0 is the optimal contraction speed of the contraction element; v ce is the contraction speed of the contraction element after the tension changes, c is the hyperbolic shape factor;

[0112] The tension of the parallel elastic element is:

[0113] F pee =K pee (l ce -l ce0 ) 2

[0114] Among them, K pee is the elastic modulus of the parallel elastic element;

[0115] The damping force generated by the contraction element is:

[0116] f c =Cv ce

[0117] Where C is the damping coefficient of the contraction element;

[0118] The muscle force generated by the i-th target muscle is:

[0119] F i =F ce +F pee +f c ;

[0120] In the Hill muscle model, the input variable that plays a major control role is the muscle activation state a(t). The neural activation state at time t is:

[0121] u(t)=k×A(t)(td)-l1×a(t-1)-l2×a(t-2)

[0122] Among them, A(t) is the amplitude of the bionic electromyographic signal, k, l1, l2 are the neural activation coefficients, and d is the time delay; among them, the neural activation coefficient satisfies:

[0123] l1=α1+α2(|α1|<1;|α2|<1)

[0124] l2=α1·α2(|α1|<1;|α2|<1)

[0125] k - l1 - l2 = 1

[0126] Where α1 and α2 are the myoelectric signal delay coefficients;

[0127] The muscle activation state a(t) is:

[0128]

[0129] Where c, d, m, and b are muscle activation coefficients; u(t) is the neural activation state at time t;

[0130] When the muscle activation state a(t) = 0, the target muscle is not activated;

[0131] When the muscle activation state 0 < a(t) < 1, the target muscle is partially activated;

[0132] When the muscle activation state a(t) = 1, the target muscle is fully activated.

[0133] Figure 2 It is a speed - torque curve graph. In the graph, the solid line is the speed - torque envelope curve graph of the exoskeleton robot joint under the conventional algorithm, and the dashed line is the speed - torque envelope curve graph actually required for the exoskeleton robot joint to achieve the ideal effect. The range enveloped by the speed - torque curve together with the coordinate axes is the motor working range, and the working range under the conventional algorithm is smaller than that under the ideal effect.

[0134] Figure 3 It is the output curve graph of the joint motor and the Hill muscle model. In the graph, the solid line represents the output force value of the Hill muscle model, and the dashed line represents the actual output force value of the joint motor. After calibration, the output force value of the joint motor almost coincides with the output force value of the Hill muscle model.

[0135] In this embodiment, after completing step S5, the joint motors of each part can be installed on the corresponding joints of the exoskeleton robot to control the movement of the robot exoskeleton.

[0136] Embodiment 2 A muscle simulation device for an exoskeleton robot based on electrical stimulation

[0137] This embodiment is applied to Embodiment 1. As Figure 4 shown, this embodiment includes: a curve generation module, a signal generation module, a signal pre - processing module, joint motors of each part, a signal acquisition module, a model generation module, and a calibration module.

[0138] The curve generation module is used to establish the speed - torque curves required for the movement of each part joint of the exoskeleton robot and output them to the bionic myoelectric signal generation module;

[0139] The signal generation module generates a bionic electromyographic signal according to the speed-torque curve and through the generation parameters of the bionic electromyographic signal and the PD algorithm, and transmits the signal to the signal preprocessing module;

[0140] The signal preprocessing module receives the bionic electromyographic signal generated by the signal generating module, preprocesses the bionic electromyographic signal, obtains the motion command data of the joint motors of each part, and transmits the motion command data to the joint motors of each part accordingly;

[0141] The joint motors of each part respectively receive the motion command data of the joint motors of each part generated by the signal preprocessing module and generate motion;

[0142] The signal acquisition module receives the output force value generated by the motor movement of each joint and outputs it to the correction module;

[0143] The model generation module is used to collect the motion data of a person wearing an exoskeleton and perform regression processing on the collected motion data according to the Hill muscle model to obtain the Hill muscle model output force value of the joint motor of each part, and output it to the correction module;

[0144] The correction module receives the output force value of the joint motor of each part output by the signal acquisition module and the output force value of the Hill muscle model of the joint motor of each part output by the model generation module, takes the output force value of the Hill muscle model of the joint motor of each part obtained by the model generation module as a training data set, trains through the Boltzmann machine algorithm, compares the actual output force value of the joint motor of each part with the output force value of the Hill muscle model of the joint motor of each part, and updates the generation parameters of the bionic electromyographic signal.

[0145] In this embodiment, the joint motors of various parts include: joint motors of six nodes of the human body's hip joint, knee joint, ankle joint, shoulder joint, elbow joint, and wrist joint, and the joint motors are equipped with loads that are consistent with the human body's muscle inertia and gravity.

Claims

1. A method for simulating muscles of an exoskeleton robot based on bionic electromyographic signals, characterized in that: The method comprises the following steps performed in sequence: S1. Establish the speed-torque curve required for the joint movement of each part of the exoskeleton robot; S2, generating a bionic electromyographic signal according to the speed-torque curve by using the generation parameters of the bionic electromyographic signal and the PD algorithm; S3, preprocessing the bionic electromyographic signal to obtain motion command data of the joint motors of each part; S4, inputting the motion instruction data of the joint motors of each part into the joint motors of each part respectively; The motors of the joints of each part receive the motion instruction data and move, and the output force values ​​of the motors of the joints of each part are collected; S5, using the output force value of the Hill muscle model of the joint motor of each part as a training data set, training through the Boltzmann machine algorithm, comparing the actual output force value of the joint motor of each part with the output force value of the Hill muscle model of the joint motor of each part, and updating the generation parameters of the bionic electromyographic signal; Wherein, establishing the Hill muscle model of the joint motor of each part and obtaining the output force value of the Hill muscle model of the joint motor of each part is performed in any step before executing step S5; S6, if the difference between the output force value of the joint motor of each part and the output force value of the Hill muscle model of the joint motor of each part is less than or equal to 3% of the maximum joint torque, continue to execute step S7; Otherwise, jump to step S2, and repeat steps S2 to S5 until the difference in the output force values ​​of the Hill muscle model of the joint motors of each part is less than or equal to 3% of the maximum joint torque; S7. Install the joint motors of each part to the corresponding joints of the exoskeleton robot to control the movement of the robot exoskeleton.

2. The exoskeleton robot muscle simulation method based on bionic myoelectric signals according to claim 1 is characterized in that: The process of establishing the speed-torque curve required for the movement of each joint of the exoskeleton robot in step S1 includes: The human body's hip joint, knee joint, ankle joint, shoulder joint, elbow joint, and wrist joint are taken as the six major motion nodes, and the muscle groups of the six major motion nodes are used as driving force. Through simulation calculation and combining with human working movements, the motion spectrum of the exoskeleton robot's joints is analyzed, and the speed-torque curves required for the movement of each joint of the exoskeleton robot are established.

3. The exoskeleton robot muscle simulation method based on bionic myoelectric signals according to claim 1 or 2, characterized in that: The PD algorithm in step S2 is expressed as: A(t)=P*(P cmd -P act )-D*V act , Among them, A(t) is the amplitude of the bionic electromyographic signal, P cmd is the joint motor command position, P act is the actual position of the joint motor, V act is the actual speed of the joint motor, P is the stiffness coefficient of the joint motor, and D is the damping coefficient of the joint motor. The selection of the stiffness coefficient P and the damping coefficient D is determined by the speed-torque curve; The generation parameters of the electromyographic signal include stimulation duration, stimulation point direction and stimulation signal amplitude.

4. The exoskeleton robot muscle simulation method based on bionic myoelectric signals according to claim 1 or 2, characterized in that: In step S3, the bionic electromyographic signal preprocessing adopts the following steps which are performed in sequence: a1) Establishing a regression model to perform regression processing on the bionic electromyographic signal; a2) amplifying the bionic electromyographic signal after the regression process to obtain an amplified bionic electromyographic signal; a3) rectifying the amplified bionic electromyographic signal to obtain a rectified bionic electromyographic signal; a4) filtering the bionic myoelectric signal after rectification using a first-order low-pass filtering algorithm to obtain the bionic myoelectric signal after filtering; a5) organizing the bionic electromyographic signal after filtering, constructing a numerical model of the bionic electromyographic signal, and determining the effective bionic electromyographic signal according to the numerical model.

5. The exoskeleton robot muscle simulation method based on bionic myoelectric signals according to claim 1 or 2, characterized in that: In step S5, the Hill muscle model of the joint motor of each part is established, and the method for obtaining the output force value of the Hill muscle model of the joint motor of each part is as follows: Collect the motion data of the exoskeleton worn by the collector, use a computer for simulation, generate EMG signals to drive the joint motors of each part, and perform regression processing on the speed, position, and torque of the joint motors of each part of the exoskeleton robot according to the Hill muscle model to obtain a Hill muscle model with the amplitude of the EMG signal as the input and the muscle force as the output; Among them, the Hill muscle model includes a contractile element, a parallel elastic element, and a series elastic element; The tension generated by the contractile element is: F cc =f(l)f(v)a Among them, f(l) is the instantaneous muscle contraction length coefficient of the contractile element; f(v) is the instantaneous muscle contraction speed coefficient of the contractile element, and a is the muscle activation state; The instantaneous muscle contraction length coefficient f(l) of the contractile element and the instantaneous muscle contraction speed coefficient f(v) of the contractile element satisfy: f(l)=F max (1-(l ce -l ce0 ) 2 / ω 2 (l ce0 ) 2 ) f(v)=(v ce0 -v ce ) / (v ce0 +(v ce / c)) Among them, F max is the maximum isometric contraction tension of the contraction element; l ce0 is the optimal length of the contraction element; l ce is the length of the contraction element after the tension changes; ω is the range of force generated by the contraction element; v ce0 is the optimal contraction speed of the contraction element; v ce is the contraction speed of the contraction element after the tension changes, c is the hyperbolic shape factor; The tension of the parallel elastic element is: F pee =K pee (l ce -l ce0 ) 2 Among them, K pee is the elastic modulus of the parallel elastic element; The damping force generated by the contractile element is: f c =Cv ce Among them, C is the damping coefficient of the contractile element; The muscle force generated by the i-th target muscle is: F i =F ce +F pee +f c ; In the Hill muscle model, the main input variable for control is the muscle activation state a(t). Let the neural activation state at time t be: u(t) = k×A(t)(t - d) - l1×a(t - 1) - l2×a(t - 2) Among them, A(t) is the amplitude of the bionic EMG signal, k, l1, and l2 are neural activation coefficients respectively, and d is the time delay; among them, the neural activation coefficients satisfy: l1＝α1+α2(|α1|<1；|α2|<1) l2＝α1·α2(|α1|<1；|α2|<1) k-l1-l2＝1 Among them, α1 and α2 are EMG signal delay coefficients; The muscle activation state a(t) is: Among them, c, d, m, and b are muscle activation coefficients; u(t) is the neural activation state at time t; When the muscle activation state a(t) = 0, the target muscle is not activated; When the muscle activation state 0 < a(t) < 1, the target muscle is partially activated; When the muscle activation state a(t) = 1, the target muscle is fully activated.

6. An exoskeleton robot muscle simulation device based on bionic electromyographic signals, applied to the exoskeleton robot muscle simulation method based on electrical stimulation according to any one of claims 1 to 5, characterized in that: It includes a curve generation module, a signal generation module, a signal preprocessing module, joint motors of each part, a signal acquisition module, a model generation module, and a correction module; The curve generation module is used to establish the speed-torque curve required for the movement of the joints of each part of the exoskeleton robot and output it to the bionic EMG signal generation module; The signal generation module receives the speed-torque curve and generates a bionic EMG signal through the generation parameters of the bionic EMG signal and the PD algorithm, and transmits it to the signal preprocessing module; The signal preprocessing module receives the bionic EMG signal generated by the signal generation module, preprocesses the bionic EMG signal, obtains the motion instruction data of the joint motors of each part, and transmits it to the joint motors of each part correspondingly; The joint motors of each part respectively receive the motion instruction data of the joint motors of each part generated by the signal preprocessing module and generate motion; The signal acquisition module receives the output force value generated by the motion of the joint motors of each part and outputs it to the correction module; The model generation module is used to collect the motion data of the exoskeleton worn by the collector and perform regression processing on the collected motion data according to the Hill muscle model to obtain the output force value of the Hill muscle model of the joint motors of each part and output it to the correction module; The correction module receives the output force value of the joint motor of each part output by the signal acquisition module and the output force value of the Hill muscle model of the joint motor of each part output by the model generation module, takes the output force value of the Hill muscle model of the joint motor of each part obtained by the model generation module as a training data set, trains through the Boltzmann machine algorithm, compares the actual output force value of the joint motor of each part with the output force value of the Hill muscle model of the joint motor of each part, and updates the generation parameters of the bionic electromyographic signal.

Images