A method for effectively inhibiting position response overshoot of a series elastic actuator sliding mode control

Abstract

This invention discloses an effective sliding mode control method for suppressing overshoot in the position response of a series elastic actuator. The method includes: designing a sliding mode controller and a lumped disturbance decoupling observer based on a dynamic model of the series elastic actuator system considering lumped disturbances. The sliding mode controller includes a Gaussian dynamic gain sliding surface and an anti-integral saturation superspiral reaching law; inputting the state feedback vector of the series elastic actuator system into the lumped disturbance decoupling observer, processing it, and outputting the lumped disturbance torque of the motor side and load side of the series elastic actuator system, then inputting it together with the preset position command of the load side into the sliding mode controller, processing it, and outputting the q-axis current reference value, thereby performing closed-loop control of the load side position of the series elastic actuator system. This invention, in point-to-point position control of the load side position of a series elastic actuator, can effectively reduce the overshoot of the load side position response while maintaining the system's high dynamic response capability.

Description

Technical Field

[0001] This invention relates to an actuator sliding mode control method, which relates to the field of robot joint motor control technology, and specifically to an effective sliding mode control method for suppressing position response overshoot of a series elastic actuator. Background Technology

[0002] The series elastic actuator consists of a motor, a high transmission ratio reducer, and an elastic torsion spring. It has advantages such as high-precision transmission and high safety, and is a key execution unit in robots. Its precise and effective control has always been a hot research topic in the robotics industry.

[0003] Due to the presence of elastic torsion springs, series elastic actuators are more sensitive to external disturbances than rigid actuators, increasing the difficulty of position control. Sliding mode variable structure control, as a highly robust algorithm with low model dependence and strong anti-interference ability, is increasingly being used by researchers for the position control of series elastic actuators. Some researchers have proposed sliding mode position controllers that combine integral reaching laws and quasi-continuous sliding mode control signals into a new reaching law, achieving higher position control bandwidth and less chattering compared to constant velocity reaching laws. Other researchers have adopted the concept of 0th-order sliding mode, designing a terminal sliding surface containing control variables to avoid the problem of terminal sliding mode singularity and optimizing the selection of state variables, enabling the controller to achieve position control without steady-state error under load.

[0004] Current research on sliding mode position control of series elastic actuators focuses on steady-state accuracy and chatter reduction, lacking research on reducing overshoot during transient processes. In certain scenarios of point-to-point control, position response overshoot is unacceptable; therefore, researching how to suppress position response overshoot in sliding mode position control of series elastic actuators is significant. Studies have found that the presence of the elastic torsion spring increases the system's dynamic order. To design a sliding mode position controller, higher-order differential terms, namely acceleration error terms and jerk error terms, need to be added to the sliding surface. The accelerating current provided by these terms during deceleration causes position response overshoot. The constant sliding surface used in current research cannot effectively weaken or suppress this accelerating current, thus failing to adequately solve the position response overshoot problem. Furthermore, the integral element in the reaching law (such as the superspiral reaching law) exacerbates this overshoot phenomenon. Summary of the Invention

[0005] To address the problems existing in the background art, this invention provides an effective sliding mode control method for suppressing overshoot in the position response of a series elastic actuator. This invention addresses the problem that sliding mode position controllers for series elastic actuators cannot effectively suppress overshoot by introducing a Gaussian function into the design of the sliding surface, creating a variable-parameter sliding surface, namely a Gaussian dynamic gain sliding surface. This cancels out the accelerating current provided by the acceleration error term and the acceleration error term during the deceleration phase in the load positioning process, reducing the overshoot in the position response. Furthermore, the integral element in the reaching law is optimized, and an anti-integral saturation superspiral reaching law is designed to further suppress the overshoot in the position response.

[0006] The technical solution adopted in this invention is:

[0007] The present invention provides an effective sliding mode control method for suppressing position response overshoot of a series elastic actuator, comprising:

[0008] Step 1) Based on the dynamic model of the series elastic actuator system considering lumped disturbances, the dynamic model takes into account the deceleration ratio. A Gaussian dynamic gain sliding mode position controller and a lumped disturbance decoupling observer are designed. The Gaussian dynamic gain sliding mode position controller includes a Gaussian dynamic gain sliding surface and an anti-integral saturation superspiral reaching law.

[0009] Step 2) Input the state feedback vector of the series flexible actuator system into the lumped disturbance decoupling observer, process it, and output the lumped disturbance torque of the motor side and load side of the series flexible actuator system. Then, input it together with the preset position command of the load side into the Gaussian dynamic gain sliding mode position controller, process it, and output the q-axis current reference value.

[0010] Step 3) Perform closed-loop control on the load side position of the series elastic actuator system based on the q-axis current reference value to achieve overshoot sliding mode control that suppresses position response.

[0011] In step 1), the series elastic actuator system includes a motor, a harmonic reducer, an elastic torsion spring, and a load; the Gaussian dynamic gain sliding mode position controller is as follows:

[0012] ;

[0013] Among them, i q-ref Indicates the q-axis current reference value; N represents the transmission ratio of the harmonic reducer; J M and J L K represents the moment of inertia on the motor side and the load side, respectively. t K represents the torque coefficient on the motor side. s The nominal value representing the stiffness of an elastic torsion spring; f d f represents the relevant components of the dynamic model; s Indicates the design components of the Gaussian dynamic gain sliding surface; This represents the design component of the anti-integral saturation superspiral reaching law.

[0014] The relevant components f of the dynamic model d Specifically as follows:

[0015] ;

[0016] in, Indicates the preset position command on the load side. The fourth derivative; θ M and θ L These indicate the positions on the motor side and the load side, respectively; T M-lump and T L-lump These represent the lumped disturbance torque on the motor side and the load side, respectively; It represents the second derivative of the lumped disturbance torque on the load side.

[0017] The Gaussian dynamic gain sliding surface design component f s Specifically as follows:

[0018] ;

[0019] Where c1, c2, c3 and c4 represent the first, second, third and fourth sliding surface coefficients of the Gaussian dynamic gain sliding surface s, respectively; x1, x2, x3 and x4 represent the position difference, speed difference, acceleration difference and jerk difference on the load side, respectively; m and n represent the first and second anti-overshoot coefficients, respectively.

[0020] The design components of the anti-integral saturation superspiral reaching law Specifically as follows:

[0021] ;

[0022] ;

[0023] ;

[0024] Where, r ori and r sat These represent the opposite of the reaching law and its saturation value, respectively; and These represent the first and second reaching law coefficients, respectively; s represents the Gaussian dynamic gain sliding surface; Represents a symbolic function; I sat This represents the q-axis saturation current on the motor side.

[0025] In step 1), the Gaussian dynamic gain sliding surface s is specifically as follows:

[0026] ;

[0027] ;

[0028] ;

[0029] Where h() and g() represent the first and second anti-overshoot dynamic gains, respectively.

[0030] In step 1), the anti-integral saturation superhelical reaching law Specifically as follows:

[0031] .

[0032] In step 1), the lumped disturbance decoupling observer is specifically as follows:

[0033] ;

[0034] Among them, s lump ω represents the sliding mode surface vector of the lumped disturbance decoupling observer; ω represents the speed measurement vector. and Let e ​​and e represent the vector of observed rotational speeds and their first derivatives, respectively. Let J and K represent the error vectors between the measured speed vector and the observed speed vector, and their first derivatives, respectively; J represents the moment of inertia matrix; and K represents the input gain matrix. Represents the state feedback vector, u=[i q , θ M / N-θ L ] T i q This represents the measured q-axis current value on the motor side, θ. M and θ L These represent the positions on the motor side and the load side, respectively, and N represents the transmission ratio of the harmonic reducer. and Let them represent the lumped perturbation observation vector and its first derivative, respectively. =[ , ] T T M-lump and T L-lump denoted by and , respectively, the lumped disturbance torque on the motor side and the load side; k represents the lumped disturbance switching gain matrix; p represents the switching vector; and g is the lumped disturbance feedback gain matrix.

[0035] The electronic device of the present invention includes: a memory and a processor coupled to each other, wherein the memory stores program data, and the processor invokes the program data to execute the method described above.

[0036] The present invention provides a computer-readable storage medium having program data stored thereon, characterized in that the program data, when executed by a processor, implements the method described above.

[0037] The beneficial effects of this invention are:

[0038] 1. The Gaussian dynamic gain sliding surface of the present invention increases the deceleration current provided by the approaching law component in the control law of the load-side deceleration stage, which offsets part of the acceleration current provided by the acceleration error term and the jerk error term during the deceleration stage, so that the load side can decelerate better, thereby suppressing the overshoot of the load-side position response.

[0039] 2. The Gaussian dynamic gain sliding surface of the present invention introduces two additional velocity error terms into the sliding surface component of the q-axis reference current control law, which increases the deceleration current during the deceleration phase and partially offsets the acceleration current provided by the acceleration error term and the jerk error term during the deceleration phase, so that the load side can decelerate better and further suppress the overshoot of the load side position response.

[0040] 3. The anti-integral saturation superspiral approach law of the present invention suppresses the overshoot caused by the integral component. Combined with the above two points, it suppresses the overshoot of the load side position response of the series elastic actuator during the point-to-point positioning process.

[0041] 4. The lumped disturbance decoupling observer of the present invention realizes the observation of lumped disturbances on the motor side and the load side, and decouples them from the speed observation, thus meeting the input variable requirements of the Gaussian dynamic gain sliding mode position controller. Attached Figure Description

[0042] Figure 1 This is a simplified mechanical structure diagram of the series elastic actuator SEA of the present invention;

[0043] Figure 2 This is a structural block diagram of the anti-integral saturation superspiral reaching law of the present invention;

[0044] Figure 3 This is a block diagram of the lumped disturbance decoupling observer of the present invention;

[0045] Figure 4 This is a block diagram of the Gaussian dynamic gain sliding mode position control structure of the series elastic actuator of the present invention;

[0046] Figure 5 The figures show the position step response waveforms under the present invention and the traditional sliding mode control method, wherein... Figure 5 (a) is a load step position response curve under the present invention and the traditional sliding mode control method. Figure 5 (b) is a load speed curve diagram under the present invention and the traditional sliding mode control method. Figure 5(c) is a current component diagram of the sliding surface provided by the present invention and the conventional sliding mode control method. Figure 5 (d) is the current component diagram provided by the reaching law under the present invention and the traditional sliding mode control method. Figure 5 (e) is a current component diagram provided by the integral term in the reaching law under the present invention and the conventional sliding mode control method. Figure 5 (f) is a schematic diagram of the s-values ​​corresponding to the system state points under the present invention and the traditional sliding mode control method. Detailed Implementation

[0047] To describe the present invention in more detail, the technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0048] like Figure 1 As shown, the Series Elastic Actuator (SEA) system includes a motor, a harmonic reducer, an elastic torsion spring, and a load. The harmonic reducer achieves torque amplification and high-precision transmission. The elastic torsion spring, as a core flexible component, not only enhances the system's safety and interaction compliance but also strengthens the shock resistance and transient force output performance of both the harmonic reducer and the load side. The ratio of the speeds on both sides of the harmonic reducer is strictly equal to the transmission ratio N and does not change with operating conditions; however, the ratio of the torques on both sides will vary with operating conditions. Furthermore, due to measurement errors, aging, and other factors, the actual stiffness of the elastic torsion spring will deviate from its nominal value.

[0049] In a specific implementation of this invention, a Landau LSEA3 series elastic actuator is used as the motor under test (its parameters correspond to variables with subscript M in subsequent formulas), and a Huichuan MS1H3-44C15CD-A331R is used as the load motor (its parameters correspond to variables with subscript L in subsequent formulas). These are connected via a rigid coupling to form an algorithm testing platform to test the actual performance of this invention. The specific parameters of the motor under test and the load motor are shown in Table 1 below:

[0050] Table 1

[0051]

[0052] like Figure 4 As shown, the load-side position control system of the series elastic actuator is a dual-loop control structure consisting of an inner current loop and an outer position loop. The position loop is a Gaussian dynamic gain sliding mode position controller, whose inputs include the load-side position reference command. , Load side position θ L and velocity ω L θ position on the motor side M and velocity ω M The feedback value and the torque information matrix T obtained by the observer lump The output is the q-axis current reference value i.q-ref (i d-ref =0 control). Current loop control is a proportional-integral PI control based on the field-oriented control method. The motor stator current i is sampled and fed back to the proportional-integral PI controller, causing it to output a reference voltage. This reference voltage is then adjusted using Space Vector Pulse Width Modulation (SVPWM) technology to drive the motor. U dc This refers to the DC voltage of the inverter. When the current loop bandwidth is sufficiently high, the output q-axis current can completely track the q-axis current reference value in a very short time. Therefore, this invention considers the motor's current loop as an ideal execution unit, and designs the position loop based on this.

[0053] To achieve closed-loop control of the load-side position of a series elastic actuator, this embodiment provides an effective sliding mode control method to suppress position response overshoot of the series elastic actuator. The specific process is as follows:

[0054] Firstly, based on a series elastic actuator system considering lumped disturbances, a dynamic model of the series elastic drive system is established, taking into account the changes in the transmission efficiency of the harmonic reducer and the stiffness mismatch of the elastic torsion spring in the control system. The dynamic model incorporates the efficiency of the harmonic reducer into the scope of lumped disturbances, thus better explaining the source of external disturbances, as detailed below:

[0055]

[0056]

[0057]

[0058] Among them, J M ω M T M B M and θ M These represent the moment of inertia, rotational speed, electromagnetic torque, coefficient of friction, and position on the permanent magnet synchronous motor side, respectively. The derivative of the rotational speed on the permanent magnet synchronous motor side; K t This represents the motor torque coefficient on the permanent magnet synchronous motor side; T represents the q-axis current on the permanent magnet synchronous motor side. M-lump and T L-lump T represents the lumped disturbance torque on the motor side and the load side, respectively. s K represents the transmitted torque of a torsion spring, and its magnitude is related to the positional difference between the two sides of the torsion spring. s and ∆K s θ represents the nominal value of the stiffness of the elastic torsion spring and its error compared to the actual value, respectively; N and η represent the transmission ratio and transmission efficiency of the harmonic reducer, respectively;L J L ω L T L and B L These represent the position, moment of inertia, rotational speed, load torque, and coefficient of friction on the load side, respectively. The derivative of the rotational speed on the load side.

[0059] Series flexible actuators (SEAs) are typically equipped with motor-side and load-side encoders, which can be used to measure θ in the model separately. M ω M θ L ω L Take measurements. J M J L It can be identified offline through simple acceleration and deceleration experiments. M-lump T L-lump This necessitates the design of a torque observer to monitor the lumped disturbance torque in real time.

[0060] A dynamic model considering lumped disturbances is used for a series elastic actuator system. The dynamic model takes into account the deceleration ratio and a Gaussian dynamic gain sliding mode position controller and a lumped disturbance decoupling observer are designed.

[0061] The Gaussian dynamic gain sliding mode position controller of this invention addresses the position response overshoot caused by acceleration error terms and jerk error terms in a constant sliding mode surface. It designs a Gaussian dynamic gain sliding mode surface 's', which uses a Gaussian function as the sliding mode gain to construct a dynamic gain sliding mode surface. The Gaussian dynamic gain sliding mode surface 's' automatically adjusts its shape under different operating conditions, effectively suppressing the overshoot of the load-side position response. The state variables of the Gaussian dynamic gain sliding mode surface 's' are as follows:

[0062]

[0063] Where x1, x2, x3 and x4 represent the position difference, speed difference, acceleration difference and jerk difference on the load side, respectively; , and These represent the first derivatives of the position difference, speed difference, and acceleration difference on the load side, respectively. , and These represent the preset position commands on the load side. The first, second, and third derivatives; It represents the first derivative of the lumped disturbance torque on the load side.

[0064] First derivative of the jerk difference on the load side Specifically as follows:

[0065]

[0066] The Gaussian dynamic gain sliding surface s is as follows:

[0067]

[0068]

[0069]

[0070] Where c1, c2, c3, and c4 represent the first, second, third, and fourth sliding surface coefficients of the Gaussian dynamic gain sliding surface s, respectively, all greater than 0. In specific implementation, c1=50, c2=1, c3=0.05, and c4=0.0005 are selected. m and n represent the first and second anti-overshoot coefficients, respectively, m and n>0. m and n are the standard deviations of the Gaussian function, and their values ​​determine the central width of the Gaussian function. In specific implementation, m=0.9×θ is selected. L-ref n=0.1×θ L-ref h() and g() represent the first and second anti-overshoot dynamic gains, respectively.

[0071] Differentiating the Gaussian dynamic gain sliding surface s, we get:

[0072]

[0073] Among them, h() and Let g( ) and g( ) represent the first anti-overshoot dynamic gain and its first derivative, respectively. These represent the second anti-overshoot dynamic gain and its first derivative, respectively. It represents the first derivative of the jerk difference on the load side.

[0074] To address the overshoot problem in the load-side position response caused by the saturation of the integral element in traditional superspiral reaching laws, this invention designs an anti-integral-saturation superspiral reaching law. ,like Figure 2 As shown, the details are as follows:

[0075] .

[0076]

[0077]

[0078] Where r ori and r sat These represent the opposite of the reaching law and its saturation value, respectively; and y1 and y2 represent the first and second reaching law coefficients, respectively, where y1 and y2 > 0. In practice, y1 = 5 and y2 = 600 are selected; s represents the Gaussian dynamic gain sliding surface. Represents a symbolic function; I sat This represents the q-axis saturation current on the motor side. In specific implementation, I is selected. sat =1.5×I n . Figure 2 In this context, u represents s, and z is a symbol used in discrete mathematics.

[0079] Based on sliding mode control theory, by substituting the anti-integral saturation superspinning approach law and the first derivative of the jerk difference on the load side into the differential of the Gaussian dynamic gain sliding surface s, the Gaussian dynamic gain sliding mode position controller is finally obtained as follows:

[0080]

[0081]

[0082]

[0083]

[0084] Among them, i q-ref Indicates the q-axis current reference value; N represents the transmission ratio of the harmonic reducer; J M and J L K represents the moment of inertia on the motor side and the load side, respectively. t K represents the torque coefficient on the motor side. s The nominal value representing the stiffness of an elastic torsion spring; f d f represents the relevant components of the dynamic model; s Indicates the design components of the Gaussian dynamic gain sliding surface; Represents the design components of the anti-integral saturation superspiral reaching law; Indicates the preset position command on the load side. The fourth derivative; θ M and θ L These indicate the positions on the motor side and the load side, respectively; T M-lump and T L-lump These represent the lumped disturbance torque on the motor side and the load side, respectively; It represents the second derivative of the lumped disturbance torque on the load side.

[0085] like Figure 3 As shown, the designed lumped disturbance decoupling observer is as follows:

[0086] ;

[0087] J= K= k= g=

[0088] Among them, s lump s represents the sliding mode surface vector of the lumped disturbance decoupling observer. lump =[s M , s L ] T s M and s L Let T and ω represent the sliding surfaces on the motor side and load side of the lumped disturbance decoupling observer, respectively; T denotes matrix transpose; and ω denotes the speed measurement vector. and Let e ​​and e represent the vector of observed rotational speeds and their first derivatives, respectively. Let ω = [ω_0, ... M , ω L ] T , , and These represent the observed speed values ​​on the motor side and the load side, respectively, e=[e M , e L ] T e M and e L These represent the errors between the observed and measured speed values ​​on the motor side and the load side, respectively; J represents the moment of inertia matrix; K represents the input gain matrix; Represents the state feedback vector, u=[i q , θ M / N-θ L ] T i q This represents the measured q-axis current value on the motor side, θ. M and θ L These represent the positions on the motor side and the load side, respectively, and N represents the transmission ratio of the harmonic reducer. and Let them represent the lumped perturbation observation vector and its first derivative, respectively. =[ , ] T T M-lump and T L-lump Let represent the lumped disturbance torque on the motor side and the load side, respectively; k represents the lumped disturbance switching gain matrix. and These represent the lumped disturbance switching gain on the motor side and the load side, respectively. In specific implementation, k is selected. M =-30,k L=-30; p represents the switching vector, p=[ , ] T g is the lumped disturbance feedback gain matrix. and These represent the lumped disturbance feedback gain on the motor side and the load side, respectively. In specific implementation, g is selected. M =1, g L =1.

[0089] Then, the state feedback vector of the series flexible actuator system is input into the lumped disturbance decoupling observer. After processing, the lumped disturbance torque of the motor side and load side of the series flexible actuator system is output. This torque, along with the preset position command on the load side, is then input into the Gaussian dynamic gain sliding mode position controller. After processing, the q-axis current reference value is output. Finally, closed-loop control of the load side position of the series flexible actuator system is performed based on the q-axis current reference value, which conforms to the following... Figure 4 The position loop section shown implements overshoot sliding mode control to suppress position response.

[0090] On the aforementioned algorithm testing platform, a step command from position 0 to 2π was used, carrying a load T L Taking a 28Nm start-up as an example, performance tests and analyses were conducted on the method proposed in this invention and the traditional sliding mode control method (taking a sliding mode controller using a linear sliding surface and a superspiral reaching law as an example). The results are as follows: Figure 5 As shown. Under the Gaussian dynamic gain sliding mode position control method of this invention, when the load side approaches the target position, the load side position error x1 is positive, and the load side speed error x2 is negative. Therefore, the second and fifth terms in the q-axis current reference value control law are negative. The current component provided by the sliding surface in the proposed method is smaller than that in the traditional control method, such as... Figure 5 As shown in (c). Furthermore, from... Figure 5 From (f), we can see that the phase point is initially in the region s>0, with the same sign as x1. As the load side approaches the target position, the coefficient of x1 in the proposed Gaussian dynamic gain sliding surface decreases, while the coefficient of x2 increases. Compared to a fixed-parameter sliding surface, the proposed sliding surface allows the phase point to move to the region s=0 or even s<0 more quickly. Therefore, the current component provided by the approaching law of the proposed method is less than that of the traditional control method, such as... Figure 5 As shown in (d). Finally by Figure 5 As shown in (e), the integral term in the proposed anti-integral saturation superspiral reaching law provides a smaller positive current during the deceleration phase, which is more conducive to deceleration on the load side. The above analysis shows that the proposed control method effectively counteracts the accelerating current provided by the x3 and x4 components, allowing for better deceleration on the load side. Figure 5 As shown in (b), this reduces overshoot in the position response, as Figure 5 As shown in (a).

[0091] The embodiments described above are merely some preferred embodiments of the present invention, and are not intended to limit the invention. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the invention. Therefore, all technical solutions obtained by equivalent substitution or equivalent transformation fall within the protection scope of the present invention.

Claims

1. A sliding mode control method for effectively suppressing position response overshoot of a series elastic actuator, characterized in that, include: Step 1) Based on the dynamic model of the series elastic actuator system considering lumped disturbances, design a Gaussian dynamic gain sliding mode position controller and a lumped disturbance decoupling observer. The Gaussian dynamic gain sliding mode position controller includes a Gaussian dynamic gain sliding surface and an anti-integral saturation superspiral reaching law. Step 2) Input the state feedback vector of the series flexible actuator system into the lumped disturbance decoupling observer, process it, and output the lumped disturbance torque of the motor side and load side of the series flexible actuator system. Then, input it together with the preset position command of the load side into the Gaussian dynamic gain sliding mode position controller, process it, and output the q-axis current reference value. Step 3) Perform closed-loop control on the load side position of the series elastic actuator system based on the q-axis current reference value to achieve sliding mode control that suppresses position response overshoot; In step 1), the Gaussian dynamic gain sliding surface s is specifically as follows: ； ； ； Where c1, c2, c3 and c4 represent the first, second, third and fourth sliding surface coefficients of the Gaussian dynamic gain sliding surface s, respectively; h() and g() represent the first and second anti-overshoot dynamic gains, respectively; x1, x2, x3 and x4 represent the position difference, speed difference, acceleration difference and jerk difference on the load side, respectively; m and n represent the first and second anti-overshoot coefficients, respectively.

2. The sliding mode control method for effectively suppressing position response overshoot of a series elastic actuator according to claim 1, characterized in that: In step 1), the series elastic actuator system includes a motor, a harmonic reducer, an elastic torsion spring, and a load; the Gaussian dynamic gain sliding mode position controller is as follows: ； Among them, i q-ref Indicates the q-axis current reference value; N represents the transmission ratio of the harmonic reducer; J M and J L K represents the moment of inertia on the motor side and the load side, respectively. t K represents the torque coefficient on the motor side. s The nominal value representing the stiffness of an elastic torsion spring; f d f represents the relevant components of the dynamic model; s Indicates the design components of the Gaussian dynamic gain sliding surface; This represents the design component of the anti-integral saturation superspiral reaching law.

3. The sliding mode control method for effectively suppressing position response overshoot of a series elastic actuator according to claim 2, characterized in that: The relevant components f of the dynamic model d Specifically as follows: ； in, Indicates the preset position command on the load side. The fourth derivative; θ M and θ L These indicate the positions on the motor side and the load side, respectively; T M-lump and T L-lump These represent the lumped disturbance torque on the motor side and the load side, respectively; It represents the second derivative of the lumped disturbance torque on the load side.

4. The sliding mode control method for effectively suppressing position response overshoot of a series elastic actuator according to claim 2, characterized in that: The Gaussian dynamic gain sliding surface design component f s Specifically as follows: 。 5. The sliding mode control method for effectively suppressing position response overshoot of a series elastic actuator according to claim 4, characterized in that: The design components of the anti-integral saturation superspiral reaching law Specifically as follows: ； ； ； Where, r ori and r sat These represent the opposite of the reaching law and its saturation value, respectively; and These represent the first and second reaching law coefficients, respectively; s represents the Gaussian dynamic gain sliding surface; Represents a symbolic function; I sat This represents the q-axis saturation current on the motor side.

6. The sliding mode control method for effectively suppressing position response overshoot of a series elastic actuator according to claim 5, characterized in that: In step 1), the anti-integral saturation superhelical reaching law Specifically as follows: 。 7. The sliding mode control method for effectively suppressing position response overshoot of a series elastic actuator according to claim 1, characterized in that: In step 1), the lumped disturbance decoupling observer is specifically as follows: ； Among them, s lump ω represents the sliding mode surface vector of the lumped disturbance decoupling observer; ω represents the speed measurement vector. and Let e ​​and e represent the vector of observed rotational speeds and their first derivatives, respectively. Let J and K represent the error vectors between the measured speed vector and the observed speed vector, and their first derivatives, respectively; J represents the moment of inertia matrix; and K represents the input gain matrix. Represents the state feedback vector, u=[i q , θ M / N-θ L ] T i q This represents the measured q-axis current value on the motor side, θ. M and θ L These represent the positions on the motor side and the load side, respectively, and N represents the transmission ratio of the harmonic reducer. and Let them represent the lumped perturbation observation vector and its first derivative, respectively. =[ , ] T T M-lump and T L-lump denoted by and , respectively, the lumped disturbance torque on the motor side and the load side; k represents the lumped disturbance switching gain matrix; p represents the switching vector; and g is the lumped disturbance feedback gain matrix.

8. An electronic device, characterized in that, include: A memory and a processor are coupled to each other, wherein the memory stores program data, and the processor invokes the program data to perform the method as described in any one of claims 1-7.

9. A computer-readable storage medium storing program data thereon, characterized in that, When the program data is executed by the processor, the method as described in any one of claims 1-7 is implemented.

Images