An adaptive control method for upper limb rehabilitation training device

Abstract

The application belongs to the technical field of control and discloses a self-adaptive control method for upper limb rehabilitation training device. A permanent magnet synchronous motor is used as a driving device of a joint of a mechanical arm, and a double closed loop vector control strategy of a current loop and a speed loop is used to control the motor, wherein the inner loop is the current loop, the current loop uses a PI control strategy for control, the outer loop is the speed loop, and the speed loop uses a PID control strategy for control. The method is characterized in that, in a dynamic balance state, a formula is used to calculate an arm main driving torque, wherein F 手 =F 阻 ‑F 电 the arm main driving torque is calculated, F 电 a motor output torque is calculated, F 手 the arm main driving torque is calculated, F 阻 a corresponding load torque of the joint of the mechanical arm is calculated, and a value of an expected speed of arm movement corresponding to the arm main driving torque calculated according to the arm main driving torque is used as a given speed value of the speed loop. The application can achieve the same effect as self-rehabilitation training without mechanical assistance under the condition that the patient's activity ability is insufficient to complete the training action.

Description

Technical Field

[0001] This invention belongs to the field of control technology, and relates to the control technology of upper limb rehabilitation training equipment, specifically an adaptive control method for an upper limb rehabilitation training equipment. Background Technology

[0002] Stroke survivors and those with limb motor dysfunction due to trauma require active and passive limb movement training to stimulate nerves and muscles and restore basic limb function. During rehabilitation training, there is a relationship between arm strength and the speed of training movements. Greater arm strength allows for faster training, while slower training indicates weaker arm strength. Therefore, the actual achievable movement speed of the human arm is an important factor reflecting its health status.

[0003] Devices for upper limb rehabilitation training can be divided into two categories: non-powered and powered. The upper limb rehabilitation trainer described in this article is a portable, wearable, powered device driven by an electric motor, belonging to the field of mechatronics integrated automation equipment. Existing industrial or military mechanical exoskeletons are generally designed to save effort, and their structure typically requires force distribution, balance, and unidirectional support. Whether powered or non-powered, the application goal is the coordinated movement of normal limbs. When lifting objects, the control system of a powered upper limb exoskeleton can provide motor assistance based on the assessment of the object's weight. The relatively ideal assistance effect is that it can fully utilize the autonomous strength of the human arm while avoiding the upper limb rehabilitation trainer restricting the speed of training movements, allowing the training movements to reach the desired speed corresponding to the arm strength. This allows trainees to achieve essentially the same effect as those who can complete training movements without mechanical assistance, even when their own mobility is insufficient.

[0004] Addressing the needs of upper limb rehabilitation trainees with upper limb movement disorders, this rehabilitation training device's control system proactively analyzes the strength of the human arm—that is, when the arm has the ability to actively swing or bend—and uses this strength as input to the control system. This allows for adaptive adjustments to reduce motor assistance, maximizing the trainee's ability to perform voluntary strength training movements.

[0005] To address the aforementioned need for significantly stimulating trainees' voluntary strength training, Chinese Patent Application No. CN113633521A, published on November 12, 2021, discloses a control system and method for an upper limb exoskeleton rehabilitation robot. This control method uses a sensing and information acquisition module to acquire real-time angular displacement and angular velocity information of each joint of the upper limb rehabilitation trainer, as well as the interaction force information between the trainee and the upper limb rehabilitation trainer. This information is transmitted to a host computer for analysis to obtain the target impedance force. A fuzzy impedance controller is used to adjust the target impedance in real time to obtain the target impedance force. Then, a PID force control closed loop is used to eliminate force errors, thereby obtaining the final control torque, which ultimately drives the motor of the upper limb rehabilitation trainer.

[0006] While the control method mentioned in this patent can largely stimulate the trainee's need for voluntary strength training, it has two significant drawbacks. First, the patent fails to reflect the actual speed of the human arm's movement, thus failing to use arm movement speed to reflect the arm's health status. It uses the angular velocity and displacement of the upper limb rehabilitation trainer body, along with the interaction force between the affected limb and the trainer, to construct an impedance control model to determine the target impedance force. The deviation between the target impedance force and the interaction force between the affected limb and the trainer fed back from the force sensor is used as the input to a PID force controller. A closed-loop system eliminates force errors, thereby obtaining the final control torque for rehabilitation training. However, it does not involve acquiring the speed of the human arm's movement. Second, the control system obtains the interaction force between the trainee and the upper limb rehabilitation trainer, i.e., the human arm torque value, through a force sensor. This is detrimental to the portable and lightweight design of upper limb rehabilitation equipment. Summary of the Invention

[0007] This invention addresses the problem mentioned above that, when aiming to maximize the stimulation of trainees' voluntary strength training movements, it is impossible to simultaneously reflect the health status of the human arm through the speed of arm movement. It provides an adaptive control method for an upper limb rehabilitation training device.

[0008] To achieve the above objectives, the technical solution adopted by the present invention is: an adaptive control method for an upper limb rehabilitation trainer, wherein a permanent magnet synchronous motor is used as the driving device for the joints of the upper limb rehabilitation trainer, and a dual closed-loop vector control strategy of current loop and speed loop is used to control the operation of the motor, wherein the inner loop is the current loop, which is controlled by a PI control strategy, and the outer loop is the speed loop, which is controlled by a fuzzy adaptive PID control strategy.

[0009] Its characteristic is that, under dynamic equilibrium (when the motor reaches a stable state), according to the formula... F 手 =F 阻 -F电 The active torque of the arm is calculated. F 电 For the motor output torque, F 手 For the active torque of the arm, F 阻 This refers to the load torque of the joints in the corresponding rehabilitation training device.

[0010] The desired velocity value of the arm movement corresponding to the arm's active torque, calculated based on the arm's active torque, is used as the given velocity value of the velocity loop.

[0011] The above control method uses PID speed control, which reduces the motor's auxiliary torque accordingly when the active torque of the human arm increases, so that the trainee can fully exert their autonomous activity ability. At the same time, the movement speed of the upper limb rehabilitation trainer is dynamically controlled based on the active torque of the arm to solve the constraint of the upper limb rehabilitation trainer's speed control on the training movement speed. This makes the movement state assisted by the upper limb rehabilitation trainer closer to the movement state of the trainee's autonomous rehabilitation training, thereby achieving an effect similar to autonomous rehabilitation training without mechanical assistance.

[0012] In addition, the above control method uses a torque balance formula to calculate the active torque of the arm, eliminating the need for a torque sensor.

[0013] Furthermore, before the motor reaches a stable state, a preset speed value is used as the given speed value for the speed loop.

[0014] Furthermore, the permanent magnet synchronous motor outputs torque F 电 Based on the equation of motion of the electric motor (relationship between current and torque): F 电 =3 / 2p n i q [i d （L d -L q ) + φ f ] ,in, p n This refers to the number of pole pairs in a permanent magnet synchronous motor. i d , i q The dq-axis component of the stator current. i d Also known as magnetizing current, i qAlso known as torque current, L d , L q For the dq axis inductance component, φ f It is a permanent magnet flux linkage.

[0015] Furthermore, load torque F 阻 Obtained through the following formula: F 阻 = G(q) + F 摩擦 ,in, q This refers to the joint angle value. G(q) The motor at the joint of the upper limb rehabilitation trainer needs to overcome the gravitational torque of the arm and the upper limb rehabilitation trainer's weight. F 摩擦 This refers to the frictional torque corresponding to the joint friction and the rotational friction of the motor in the upper limb rehabilitation training device. Frictional torque F 摩擦 Determined in advance through experiments.

[0016] Torque of gravity of arm and upper limb rehabilitation training equipment G(q) Obtained by querying a pre-defined gravitational moment table.

[0017] The gravitational moment table is established as follows:

[0018] The positional relationship between the end effector and the base of the upper limb rehabilitation trainer was established using the Denavit-Hartenberg method; the link offset was used to establish this relationship. d i Joint angle θ i Link length a i-1 Linkage torsion angle α i-1 The four parameters describe the transformation relationship between the various joint coordinate systems; the joint coordinate systems in two adjacent joint coordinate systems { i} relative to the joint coordinate system { i The transformation matrix of {-1} is:

[0019] ,

[0020] in, d i This is the link offset. θ i Joint angle, a i-1 The length of the link. α i-1 This is the torsion angle of the connecting rod.

[0021] Linkage offset d i Linkage torsion angle α i-1 Link length a i-1 The three parameters are known quantities. The permanent magnet synchronous motor contains a magnetic encoder, and the joint angle value of each joint is obtained by using the angle signal output by the magnetic encoder at each joint. After determining the relative relationship of the joint coordinate systems of adjacent joints, the joint coordinate system of the end joint relative to the head joint can be obtained by matrix multiplication.

[0022] The upper limb rehabilitation trainer uses the shoulder joint as the primary joint. The DH method allows for the determination of the elbow and wrist joint positions relative to the shoulder joint, thus providing all possible positions of the trainer. The range of motion of the shoulder, elbow, and wrist joints is divided equally, corresponding to all possible positions of the trainer. Using a dynamometer, the forces acting on the center of the palm, forearm, and upper arm are calculated. These forces are then multiplied by half the length of the palm, forearm, and upper arm, respectively, to obtain the gravitational torques acting on the wrist, elbow, and shoulder joints in each position. A gravitational torque table is then created based on the equal division of the range of motion of the shoulder, elbow, and wrist joints.

[0023] Furthermore, the desired velocity value of the arm movement was obtained as follows: Healthy individuals were used as experimental subjects, and training movements were performed, during which data on their upper limbs were collected. Four experimental values ​​were taken: minimum torque, minimum velocity (expressed as angular velocity) corresponding to the minimum torque, maximum torque, and maximum velocity corresponding to the maximum torque. The minimum and maximum velocities expressed as angular velocities were converted into rotational speeds. The ratio of the difference between the maximum and minimum velocities converted into rotational speeds to the difference between the maximum and minimum torques was used as the proportionality coefficient between the active torque of the arm and the desired velocity of the arm movement. K The obtained active torque value of the arm is multiplied by the proportional coefficient to obtain the desired speed of the arm movement.

[0024] Furthermore, the speed loop employs a fuzzy adaptive PID controller.

[0025] PID controller proportional coefficient K p Integral coefficient K i Differential coefficients K d The three parameters are modified online using fuzzy rules to ensure that the controlled object has good static and dynamic performance.

[0026] The input to the fuzzy controller is the deviation between the given velocity value and the feedback velocity value. e and rate of change of deviation ec Defined as: e(t) = v ref (t) - v(t) , ec = e(t) - e(t - 1) ,in, v ref Enter the speed value for the speed loop. v The speed value is fed back by the speed loop.

[0027] Manually set deviation amount e Deviation change rate ec The critical value is used as the input value for the fuzzy controller. When the absolute value of the input signal is greater than or equal to the critical value, the critical value is used; when the absolute value of the input signal is less than the critical value, the actual value of the input signal is used. For both positive and negative critical values, the signal is divided into five equal segments (NB, NS, ZE, PS, PB), representing negative large, negative small, zero, positive small, and positive large, respectively, to determine the deviation. e Deviation change rate ec The fuzzy subsets are all {NB, NS, ZE, PS, PB}. Meanwhile, the proportional gain of the output PID controller is... K p Integral coefficient K i Differential coefficients K d The three parameters are initially set to a moderate value. The output value is then divided into three equal segments (S, M, B), representing small, moderate, and large segments respectively. The proportional coefficient of the output value is then determined. K p Integral coefficient K i Differential coefficients K d The three parameters are all fuzzy subsets {S, M, B}. Design the proportional gain of the PID controller. K p Integral coefficient K i Differential coefficients K d Fuzzy rules for three parameters. Using fuzzy rules to adjust the proportional coefficient of a PID controller. K p Integral coefficient K i Differential coefficients K d The three parameters are modified. The proportional gain of the PID controller is then derived. K p Integral coefficient K i Differential coefficients K d The optimal values ​​of the three parameters are input to the PID controller.

[0028] The beneficial effects of this invention are as follows:

[0029] (1) Control objective of the present invention: In rehabilitation training without assisted transfer, there is a certain relationship between arm strength and training movement speed. When arm strength is high, training speed is fast, and when training speed is slow, arm strength is weak. Therefore, the actual movement speed that the human arm can achieve is one of the important factors reflecting the health status of the human arm. This patent reflects the human arm movement speed value through the active torque value of the human arm, and then drives the motor to perform training movements through this speed value, which is beneficial to the accuracy of the rehabilitation status assessment results.

[0030] (2) Based on a relatively simple speed control strategy, it can fully utilize the trainee's autonomous activity ability while avoiding the upper limb rehabilitation trainer's constraint on the speed of training movements. This allows the trainee to achieve the same effect as the trainee's autonomous rehabilitation training without mechanical assistance when their own activity ability is insufficient to complete the training movements.

[0031] (3) No sensors were used for data detection. From a control perspective, this provides a feasible solution for the development of practical, portable, wearable intelligent upper limb rehabilitation devices. Attached Figure Description

[0032] Figure 1 This is a block diagram illustrating the principle of the control method for the upper limb rehabilitation training device of the present invention. Detailed Implementation

[0033] The upper limb rehabilitation training device of this invention consists of a back support frame, a chest and abdomen strap, an exoskeleton-like arm, and fixing straps. Permanent magnet synchronous motors are installed in the shoulder, back, shoulder side, and elbow; a permanent magnet synchronous motor can also be installed in the wrist. This invention describes a technical method for designing an upper limb rehabilitation training device for the following application scenarios:

[0034] (1) A wearable lightweight upper limb rehabilitation training device with power motors at the corresponding positions of the shoulder, elbow and wrist joints, allowing the arm to extend and swing freely with or without power at multiple angles (front or side).

[0035] (2) When there is an active swinging or bending force input in the human arm, the permanent magnet synchronous motor control system can actively analyze and sense the input of external force, and adaptively adjust to reduce the assistance, thereby stimulating autonomous force training exercise.

[0036] (3) Under specific standard postures, the active force input of the arm can be analyzed and statistically analyzed, thereby providing support for the evaluation of rehabilitation training effects and the adjustment of training plans.

[0037] The upper limb rehabilitation trainer is an integrated electromechanical control system, and the key to its design lies in the control method of each permanent magnet synchronous motor control system. The three design goals mentioned above are fundamentally different from the functional goals of traditional industrial upper limb rehabilitation trainers and upper limb exoskeleton upper limb rehabilitation trainers, and therefore the focus of the design is also significantly different.

[0038] All motor control systems in the upper limb rehabilitation trainer are designed for assisted control. When the trainee's arm joints are completely unable to exert force independently, each joint motor should drive the arm in a fully driven manner according to the training plan for rehabilitation exercises. When the trainee's arm has a certain degree of independent movement, the permanent magnet synchronous motors at the relevant joints should drive the arm in an assisted manner for training. While all permanent magnet synchronous motor systems share the same requirements for adaptive assisted control, they differ in the input / output conditions and magnitude ranges for achieving control.

[0039] The control system's control process is divided into three stages. The first stage is the system startup stage. The trainee presses an external start button, the controller receives the initial speed setpoint from the software, and the system starts running. After a period of operation, the feedback speed value reaches the initial setpoint and enters a constant speed state, marking the start of the second stage. In the second stage, under constant speed conditions, the system feeds back the arm's active torque value. Based on the proportional relationship between the arm's active torque and the desired arm movement speed, the desired arm movement speed is obtained and used as the speed loop input signal to begin the second stage of control. The third stage is the system shutdown stage. The trainee presses an external stop button, the permanent magnet synchronous motor controller receives a zero-magnitude current setpoint from the software, and the motor stops rotating. The permanent magnet synchronous motor systems at each joint are uniformly designed as follows: Figure 1 The closed-loop control system shown is shown.

[0040] like Figure 1 As shown, the control method for the permanent magnet synchronous motor used in the upper limb rehabilitation training device is based on dual closed-loop vector control. The inner loop is the current loop, controlled by a PI controller to control the motor current (torque), enabling better dynamic current adjustment under varying motor load conditions. The outer loop is the speed loop, controlled by a fuzzy adaptive PID controller. The PID controller's proportional gain... K p Integral coefficient K i Differential coefficients K d The three parameters are dynamically determined by fuzzy inference to improve the system's robustness to the nonlinearity of the controlled object and the time-varying nature of the load. A key aspect of the closed-loop design is ensuring the speed and stability of the motor drive response.

[0041] To enable the detection and control of the human arm torque (hereinafter referred to as the arm active torque), an outer loop is added to the dual closed-loop control. When the motor output torque in the system... F 电 Arm active torque F 手 Load torque F 阻 When all three are in a dynamic equilibrium state, the motor output torque and load torque can be calculated, and then the active torque value of the arm can be obtained. Then, based on the proportional relationship between the active torque value of the arm and the desired speed value of the arm movement, the active torque value of the arm is converted into the desired speed value of the arm movement as the input value of the speed loop.

[0042] As shown in the figure, the active torque of the arm F 手 As part of the input signal, it acts directly on the permanent magnet synchronous motor, in conjunction with the load torque. F 负 Both together constitute the system input signal. Load torque F 负 With drive motor load torque model F 阻 These are quantities of different numerical values. F 负 This represents the actual workload during training; the specific value is unknown. F 阻 This is the predicted load value during training, based on the actual value calculated later.

[0043] The forces at each joint of the arm are satisfied: F 手 +F 电 -F 阻 =F 合 ,in, F 手 For the active torque of the arm, F 电 For the motor output torque, F 阻 This is the load torque.

[0044] When the permanent magnet synchronous motor at the joint is in a state of dynamic equilibrium F 合 = ma = 0 ,Right now F 手 +F 电 -F 阻 =0 . That is, F手 =F 阻 - F 电 。 From the above formula, we can see that the active torque of the arm F 手 It can be determined by the load torque F 阻 With motor output torque F 电 The result is obtained. Among them, the torque current of the permanent magnet synchronous motor is directly proportional to the motor output torque (i.e., electromagnetic torque).

[0045] To determine the relationship between the active torque of the arm and the desired speed of the arm movement: Before rehabilitation training, healthy individuals were used as experimental subjects to perform training movements, and data on their upper limbs were collected during the process. Four experimental values ​​were collected: the minimum torque value, the minimum speed value corresponding to the minimum torque value, the maximum torque value, and the maximum speed value corresponding to the maximum torque value. The minimum and maximum speeds, expressed as angular velocities, were converted into rotational speeds. The ratio of the difference between the maximum and minimum speeds converted to rotational speeds to the difference between the maximum and minimum torques was used as the proportionality coefficient between the active torque of the arm and the desired speed of the arm movement. K The obtained active torque value of the arm is multiplied by the proportional coefficient to obtain the desired speed of the arm movement.

[0046] Phase current is acquired and analyzed on the motor driver, and the torque current is obtained after conversion from the phase current. i q Thus, the output torque of the motor can be determined. F 电 。 Motor output torque F 电 Based on the equation of motion of the electric motor (relationship between current and torque):

[0047] F 电 =3 / 2p n i q [i d （L d -L q ) + φ f ] ,in, p n This refers to the number of pole pairs in a permanent magnet synchronous motor. i d , i qThe dq-axis component of the stator current. i d Also known as magnetizing current, i q Also known as torque current, L d , L q For the dq axis inductance component, φ f It is a permanent magnet flux linkage.

[0048] Number of pole pairs of permanent magnet synchronous motor p n dq axis inductance component L d , L q Permanent magnet magnetic flux φ f All of these are related to the internal structural parameters of the permanent magnet synchronous motor and are constant values. For surface-mounted permanent magnet synchronous motors, L d =L q Motor output torque and torque current i q Proportional. For built-in permanent magnet synchronous motors, the excitation current is selected... i d =0 The control method also achieves the same motor output torque and torque current. i q The effect is directly proportional.

[0049] Establish load torque models for the permanent magnet synchronous motors in the shoulder, elbow, and wrist sections of the upper limb rehabilitation training device. F 阻 : F 阻 = G(q) + F 摩擦 ,in, q This refers to the joint angle value. G(q) The permanent magnet synchronous motor at the joint of the upper limb rehabilitation trainer needs to overcome the gravitational torque of the arm and the trainer's weight. F 摩擦 The frictional torque is the frictional force corresponding to the joint friction of the upper limb rehabilitation training device and the rotational friction of the permanent magnet synchronous motor.

[0050] Torque of gravity of arm and upper limb rehabilitation training equipment G(q) The magnitude of the gravitational torque of the arm and upper limb rehabilitation trainer is related to the position and posture of the trainer. The Denavit-Hartenberg method (DH method) can be used to establish the position and posture relationship between the end effector of the upper limb rehabilitation trainer and the base. This relationship can be established through the linkage offset. di Joint angle θ i Link length a i-1 Linkage torsion angle α i-1 The four parameters describe the transformation relationship between the various joint coordinate systems; the joint coordinate systems in two adjacent joint coordinate systems { i} relative to the joint coordinate system { i The transformation matrix of {-1} is:

[0051] ,

[0052] in, d i This is the link offset. θ i Joint angle, a i-1 The length of the link. α i-1 This is the torsion angle of the connecting rod.

[0053] Linkage offset d i Linkage torsion angle α i-1 Link length a i-1 All three parameters are related to the mechanical structure of the trainer and are known quantities. The permanent magnet synchronous motor contains a magnetic encoder. The angle value of each joint is obtained by using the angle signal output from the magnetic encoder at each joint (i.e., the joint angle in the DH method). θ i After determining the relative relationship between the joint coordinate systems of adjacent joints, the joint coordinate system of the distal joint relative to the joint coordinate system of the proximal joint can be obtained by matrix multiplication.

[0054] The upper limb rehabilitation trainer uses the shoulder joint as the primary joint. The DH method allows for the determination of the elbow and wrist joint positions relative to the shoulder joint, thus providing all possible positions of the trainer. The range of motion of the shoulder, elbow, and wrist joints is divided equally, corresponding to all possible positions of the trainer. Using a dynamometer, the forces acting on the center of the palm, forearm, and upper arm are calculated. These forces are then multiplied by half the length of the palm, forearm, and upper arm, respectively, to obtain the gravitational torques acting on the wrist, elbow, and shoulder joints in each position. A gravitational torque table is then created based on the equal division of the range of motion of the shoulder, elbow, and wrist joints.

[0055] The frictional torque corresponding to the joint friction of the trainer and the rotational friction of the motor. F 摩擦When the arm is in a horizontal position, take a point at the forearm position and measure the force at that point using a dynamometer. This is the magnitude of the force under the influence of joint friction. Divide this force by the distance from that point to the center of joint rotation to obtain the friction torque value.

[0056] Establish the torque balance equations at the shoulder, elbow, and wrist joints of the upper limb rehabilitation training device: F 手 -G(q) - F 摩擦 +F 电 =0 Note: 1) F 手 This refers to the active torque of the arm. Unlike the output torque of the arm on the upper limb rehabilitation trainer, the output torque of the arm on the upper limb rehabilitation trainer can be considered as the force resulting from the active torque of the arm overcoming the arm's gravity. The gravitational torque in the above equilibrium equation... G(q) The equilibrium equations are for when the motor rotates forward (i.e., overcoming gravity during training). If the motor rotates in reverse, the equilibrium equations are: F 手 +G(q) - F 摩擦 + F 电 =0 .

[0057] Speed ​​loop control design method:

[0058] The outer loop of the control system is the speed loop, which uses a fuzzy adaptive PID controller. The proportional gain of the PID controller is... K p Integral coefficient K i Differential coefficients K d The three parameters are modified online using fuzzy rules to ensure that the controlled object has good static and dynamic performance.

[0059] The input to the fuzzy controller is the deviation between the given velocity value and the feedback velocity value. e and rate of change of deviation ec Defined as: e(t) = v ref (t) - v(t) , ec = e(t) - e(t - 1) ,in, v ref Enter the speed value for the speed loop. v The speed value is fed back by the speed loop.

[0060] Fuzzy adaptive PID control working process: The deviation is manually set. e and rate of change of deviation ecThe critical value is used as the input value for the fuzzy controller. When the absolute value of the input signal is greater than or equal to the critical value, the critical value is used; when the absolute value of the input signal is less than the critical value, the actual value of the input signal is used. For both positive and negative critical values, the signal is divided into five equal segments (NB, NS, ZE, PS, PB), representing negative large, negative small, zero, positive small, and positive large, respectively, to determine the deviation. e and rate of change of deviation ec The fuzzy subsets are all {NB, NS, ZE, PS, PB}. Meanwhile, the proportional gain of the output PID controller is... K p Integral coefficient K i Differential coefficients K d The three parameters are initially tuned to a moderate value. The output value is then divided into three equal parts (S, M, B), representing small, moderate, and large segments, respectively, to determine the output value scaling factor. K p Integral coefficient K i Differential coefficients K d The three parameters are all fuzzy subsets {S, M, B}. Design the proportional gain of the PID controller. K p Integral coefficient K i Differential coefficients K d Fuzzy rules for three parameters. Using fuzzy rules to adjust the proportional coefficient of a PID controller. K p Integral coefficient K i Differential coefficients K d The three parameters are modified. The proportional gain of the PID controller is then derived. K p Integral coefficient K i Differential coefficients K d The optimal values ​​of the three parameters are input to the PID controller, enabling the system to have good speed regulation performance under different operating conditions and achieving the goal of optimal system control.

[0061] Fuzzy rule design:

[0062] proportionality coefficient K p Its function is to accelerate system response, eliminate errors, and improve system adjustment accuracy. Proportional coefficient. K p The larger the value, the faster the system response speed and the higher the system adjustment accuracy, but it is prone to overshoot.

[0063] Integral coefficient K i Its function is to eliminate the steady-state error of the system. Integral coefficient K i The larger the integral coefficient, the faster the system's static error is eliminated, but the integral coefficient... K i If the value is too large, integral saturation may occur in the early stages of the adjustment process, resulting in a large overshoot during the adjustment process.

[0064] Differential coefficients K d Its function is to reflect the trend of signal changes and to introduce an effective early correction signal into the system before the deviation signal changes too much, thereby speeding up the response, reducing the adjustment time, eliminating oscillations, and ultimately changing the dynamic performance of the system.

[0065] Therefore, the proportional coefficient of the PID controller K p Integral coefficient K i Differential coefficients K d The tuning of the three parameters must take into account their effects at different times and their interconnections.

[0066] Generally speaking:

[0067] (1) In e When the absolute value is large, the proportionality coefficient K p Take the larger value, differential coefficient K d Take the smaller value, integral coefficient K i Setting the value to zero speeds up system response and reduces overshoot.

[0068] (2) In e When the absolute value is moderate, the proportionality coefficient K p Take the smaller value, differential coefficient K d and integral coefficient K i Choose an appropriate value to reduce overshoot and steady-state error.

[0069] (3) In e When the absolute value is small, the proportionality coefficient K p and integral coefficient K i Take the larger value, differential coefficient K dChoosing a moderate value avoids oscillations around the equilibrium point, resulting in better steady-state performance of the system.

[0070] In this control, the deviation between the given speed value and the feedback speed value is... e and rate of change of deviation ec Divide the data into 5 fuzzy subsets on average and output the scaling factor. K p Integral coefficient K i Differential coefficients K d The three parameters are then divided into three fuzzy subsets. Based on the fuzzy rules, the proportional coefficients can be obtained. K p Integral coefficient K i Differential coefficients K d The fuzzy rules for the three parameters are shown in the table below.

[0071] Table 1 K p Fuzzy rules:

[0072] ,

[0073] Table 2 K i Fuzzy rules:

[0074] ,

[0075] Table 3 K d Fuzzy rules:

[0076] ,

[0077] System Startup: When the trainee intentionally initiates training movements, the system needs to be started for control. There are two startup methods: First, when the trainee's arm torque is sufficient to move the upper limb rehabilitation trainer, the system detects the speed signal, estimates the arm torque, and then controls and starts the system. Second, when the trainee's arm torque is insufficient to overcome the resistance of the upper limb rehabilitation trainer for effective movement, a manual on / off method can be used. The trainee presses the external start button, and the controller receives the initial speed setpoint from the software, causing the motor to operate. The set speed is then adjusted according to the magnitude of the arm torque.

[0078] System Stop Detection: When the trainee intentionally stops the training movement, the system can stop the movement in two ways: First, a dead zone can be set. When the speed of the permanent magnet synchronous motor or the calculated hand torque falls below a certain range and remains below that range for a certain period, it can be considered that the trainee wants to stop exercising, and the system can enter standby mode without control. Second, a manual switch can be used. When the trainee presses the stop switch, the system receives a zero-magnitude current command from the software, and the permanent magnet synchronous motor stops rotating.

[0079] Adaptive control is manifested in addressing situations where the controlled system experiences parameter uncertainties or unknown parameter changes. From a transient real-time adjustment perspective, when the arm's active torque is used as the system input, transient adjustment is achieved to match the transient assistance of the motor.

[0080] Based on the motor output torque F 电 This feedback signal and the load torque obtained from the motor load model analysis F 阻 The active torque of the arm is obtained from the two. F 手 When the automatic arm torque is used as the system input, the active arm torque is constructed. F 手 The relationship between the load torque and speed is directly proportional. F 阻 With motor output torque F 电 The upper and lower limits of both are obtained to assess the active torque of the arm. F 手 Upper and lower limits. Determined by the arm's active torque. F 手 Matching the upper and lower limits with the required rotational speed to construct the active torque of the arm. F 手 The relationship between torque and speed is directly proportional, enabling an adaptive control system that uses the automatic torque of the arm as the input signal for the system's speed loop.

[0081] In terms of adjusting training programs, the existing training data is used for evaluation to adjust the established training plan. The mechanism for adjusting training programs is based on the active torque of the arm. F 手 Changes and their speed of change adjust the active torque of the arm. F 手 The value is directly proportional to speed. Evaluated based on current training data, the active torque of the arm. F 手 The comparison and analysis with existing data shows a changing trend (increase), proving that the rehabilitation effect is good. When the change value reaches a certain value, the active torque of the arm should be appropriately reduced. F 手The value directly proportional to speed, i.e., the active torque of the arm. F 手 The larger the value, the smaller the corresponding system speed input value. (This applies to the arm's active torque.) F 手 The matching relationship between the change value and the proportional relationship value can be established through experiments using experimental data, and the data relationship table between the two can be constructed and stored in the database.

Claims

1. An adaptive control method for an upper limb rehabilitation trainer, wherein a permanent magnet synchronous motor is used as the driving device for the joints of the upper limb rehabilitation trainer, and a dual closed-loop vector control strategy of current loop and speed loop is used to control the operation of the motor, wherein the inner loop is the current loop, which is controlled by a PI control strategy, and the outer loop is the speed loop, which is controlled by a fuzzy adaptive PID control strategy. Its features are, In a dynamic equilibrium state, according to the formula F 手 =F 阻 -F 电 The active torque of the arm is calculated; where F 电 For the motor output torque, F 手 For the active torque of the arm, F 阻 The load torque for the joints of the corresponding upper limb rehabilitation training device; The desired velocity value of the arm movement corresponding to the arm's active torque, calculated based on the arm's active torque, is used as the given velocity value of the velocity loop; Load torque F 阻 Obtained through the following formula: F 阻 =G(q)+F 摩擦 Where q is the joint angle value. G(q) The permanent magnet synchronous motor at the joint of the upper limb rehabilitation trainer needs to overcome the gravitational torque of the arm and the upper limb rehabilitation trainer. F 摩擦 The frictional torque corresponding to the joint friction force and the rotational friction force of the permanent magnet synchronous motor in the upper limb rehabilitation training device; Frictional torque F 摩擦 The gravitational torque of the arm and upper limb rehabilitation training device was determined in advance through experiments. G(q) Obtained by querying a pre-defined gravitational moment table.

2. The adaptive control method for the upper limb rehabilitation training device according to claim 1, characterized in that, Before the permanent magnet synchronous motor reaches a stable state, a preset speed value is used as the given speed value for the speed loop.

3. The adaptive control method for the upper limb rehabilitation training device according to claim 1, characterized in that, The permanent magnet synchronous motor outputs torque F 电 Based on the equation of motion of the electric motor: F 电 =3 / 2p n i q [i d （L d -L q )+φ f ] ,in, p n This refers to the number of pole pairs in a permanent magnet synchronous motor. i d , i q The dq-axis component of the stator current. i d Also known as magnetizing current, i q Also known as torque current, L d , L q For the dq axis inductance component, φ f It is a permanent magnet flux linkage.

4. The adaptive control method for the upper limb rehabilitation training device according to claim 3, characterized in that, The gravitational moment table is established as follows: The positional relationship between the end effector and the base of the upper limb rehabilitation trainer was established using the Denavit-Hartenberg method; the link offset was used to establish this relationship. d i Joint angle θ i Link length a i-1 Linkage torsion angle α i-1 The four parameters describe the transformation relationship between the various joint coordinate systems; the joint coordinate systems in two adjacent joint coordinate systems { i } relative to the joint coordinate system { i The transformation matrix of {-1} is: ， in, d i This is the link offset. θ i Joint angle, a i-1 The length of the link. α i-1 The link twist angle; Linkage offset d i Linkage torsion angle α i-1 Link length a i-1 The three parameters are known quantities; the permanent magnet synchronous motor contains a magnetic encoder, and the angle signal value output by the magnetic encoder in the permanent magnet synchronous motor at each joint is used as the joint angle value of each joint; after the relative relationship of the joint coordinate systems of adjacent joints is determined, the joint coordinate system of the end joint relative to the joint coordinate system of the head joint can be obtained by matrix multiplication. The upper limb rehabilitation trainer uses the shoulder joint as the primary joint. The Denavit-Hartenberg method is used to obtain the positions of the elbow and wrist joints relative to the shoulder joint, thus providing all possible positions of the upper limb rehabilitation trainer. The range of motion of the shoulder, elbow, and wrist joints is divided equally, corresponding to all positions of the upper limb rehabilitation trainer. A force gauge is used to calculate the magnitude of the force at the center of the palm, forearm, and upper arm, respectively. These forces are then multiplied by half the length of the palm, forearm, and upper arm, respectively, to obtain the gravitational torque on the wrist, elbow, and shoulder joints in each position. A gravitational torque table is then created based on the equal division of the range of motion of the shoulder, elbow, and wrist joints.

5. The adaptive control method for the upper limb rehabilitation training device according to claim 1, characterized in that, The desired velocity of the arm movement was obtained as follows: Healthy individuals were used as experimental subjects, and training movements were performed, during which data on their upper limbs were collected. Four experimental values ​​were taken: the minimum torque value, the minimum velocity value corresponding to the minimum torque, the maximum torque value, and the maximum velocity value corresponding to the maximum torque. The minimum and maximum velocity values, expressed as angular velocities, were converted into rotational speed values. The ratio of the difference between the maximum and minimum velocities converted to rotational speeds to the difference between the maximum and minimum torques was used as the proportionality coefficient between the active torque of the arm and the desired velocity of the arm movement. K ; The obtained active torque value of the arm is multiplied by the proportional coefficient to obtain the desired speed of the arm movement.

6. The adaptive control method for the upper limb rehabilitation training device according to claim 1, characterized in that, The speed loop employs a fuzzy adaptive PID controller; PID controller proportional coefficient K p Integral coefficient K i Differential coefficients K d The three parameters can be modified online using fuzzy rules; The input to the fuzzy controller is the deviation between the given velocity value and the feedback velocity value. e and rate of change of deviation ec Defined as: e (t)=v ref (t)-v(t) , ec = e(t) - e(t-1) ,in, v ref Give the speed loop a speed value, v The speed loop provides feedback on the speed value; Manually set deviation amount e Deviation change rate ec The critical value is used as the input value for the fuzzy controller when the absolute value of the input signal is greater than or equal to the critical value; when the absolute value of the input signal is less than the critical value, the actual value of the input signal is used as the input value for the fuzzy controller. For both positive and negative critical values, the signal is divided into five equal segments: NB, NS, ZE, PS, and PB, representing negative large, negative small, zero, positive small, and positive large, respectively, to determine the deviation. e and rate of change of deviation ec The fuzzy subsets are all {NB, NS, ZE, PS, PB}; while the output PID controller proportional coefficients are... K p Integral coefficient K i Differential coefficients K d The three parameters are initially set to moderate values. The output value is then divided into three equal segments, S, M, and B, representing small, moderate, and large segments respectively, to determine the output value scaling factor. K p Integral coefficient K i Differential coefficients K d The three parameter fuzzy subsets are all {S, M, B}; design PID controller proportional coefficient K p Integral coefficient K i Differential coefficients K d Fuzzy rules for three parameters; using fuzzy rules to adjust the proportional coefficient of the PID controller. K p Integral coefficient K i Differential coefficients K d Modify the three parameters; deduce the proportional coefficient of the PID controller. K p Integral coefficient K i Differential coefficients K d The optimal values ​​of the three parameters are input to the PID controller.

Images