Adaptive sliding mode position control method for motor reducer system

Abstract

This invention discloses an adaptive sliding mode position control method for a motor reducer system. Addressing the position tracking error, overshoot, and steady-state oscillation problems caused by backlash switching, nonlinear friction, load disturbances, and measurement noise in permanent magnet synchronous motor reducer transmission systems, a cascaded position control structure is constructed. The speed loop combines a proportional-integral controller with a disturbance observer to estimate the total system disturbance and converts the filtered disturbance estimate into a current feedforward compensation quantity. The position loop employs a composite control law consisting of proportional-integral control, speed damping feedback, and finite-time sliding mode compensation, and utilizes an adaptive recurrent neural network to estimate the disturbance intensity online and adaptively adjust the sliding mode approach gain. This method improves the convergence speed and disturbance rejection capability during commutation and backlash switching, reduces chattering in the steady-state stage, and enhances positioning accuracy and robustness.

Description

Technical Field

[0001] This invention belongs to the field of motor servo control technology, specifically relating to an adaptive sliding mode position control method and system for a backlash-containing motor reducer system. It is applicable to robot joints, precision turntables, automated equipment, servo actuators, and other motor reducer systems that require high-precision position control. Background Technology

[0002] In robot joints and precision servo systems, motors are typically used in conjunction with reducers. Reducers lower output speed and increase output torque, but due to factors such as gear manufacturing, assembly, and wear, backlash usually exists within them. When the motor moves in both directions, this backlash causes idle travel and hysteresis between the motor and load sides, preventing the reducer's output position from promptly following the motor's movement. This phenomenon further leads to position tracking errors, commutation shocks, overshoot, low-speed oscillations, and steady-state micro-jitter.

[0003] Existing position control methods typically employ proportional-integral (PI) control or disturbance observation compensation methods. PI control has a simple structure, but it is prone to overshoot and oscillations in backlash-containing systems. Disturbance observers can compensate for some load and friction disturbances, but their ability to compensate for nonlinear disturbances caused by backlash switching is limited. Sliding mode control has strong robustness, but a fixed sliding mode gain makes it difficult to simultaneously achieve fast convergence and low chattering. Therefore, a position control method for motor reducers is needed that can maintain fast response under backlash switching and external disturbances while reducing oscillations in the steady-state phase. Summary of the Invention

[0004] Technical Problem: To address the shortcomings of existing technologies, this invention discloses an adaptive sliding mode position control method for a backlash-driven motor reducer system. This method improves the position tracking accuracy and disturbance rejection capability of the backlash-driven motor reducer system; reduces overshoot and oscillation through a velocity damping term; and enhances control strength under large disturbances and reduces control chattering in the steady-state phase with small errors by adaptively adjusting the sliding mode gain.

[0005] Technical Solution: The adaptive sliding mode position control method for a motor reducer system of the present invention uses a cascaded controller to realize the position control of the motor reducer system. The cascaded controller includes a position loop controller, a speed loop controller, and a current loop controller. The position loop controller, speed loop controller, and current loop controller are connected in cascade. The position loop controller outputs a speed command to the speed loop controller, the speed loop controller outputs a current command to the current loop controller, and the current loop controller outputs a motor drive control quantity to drive the motor to run.

[0006] The position loop controller includes a position PI controller, a velocity damping module, a position sliding mode controller module, and a recurrent neural network disturbance observation module. The recurrent neural network disturbance observation module predicts nonlinear disturbance values ​​based on system operating status signals and adjusts the sliding mode parameters of the position sliding mode controller in real time based on the nonlinear disturbance values. The basic velocity command output by the position PI controller, the velocity damping compensation output by the velocity damping module, and the sliding mode compensation output by the position sliding mode controller module are superimposed at the velocity command superposition terminal to obtain the output velocity command of the position loop controller.

[0007] The speed loop controller includes a speed PI controller and a disturbance observer module. The base current command output by the speed PI controller is superimposed with the current compensation amount output by the disturbance observer module to obtain the speed loop output current command.

[0008] The current loop controller is used to generate motor drive control quantities based on current commands and current feedback signals; the control method includes the following steps:

[0009] S1, Collect the operating status signal of the motor reducer system. The operating status signal includes position command, actual position of reducer output end, motor side speed, actual speed of reducer output end and motor current feedback signal.

[0010] S2, the position error is obtained based on the deviation between the position command and the actual position of the reducer output terminal, the position error change rate is obtained based on the position error, and the position error is input into the position loop controller;

[0011] S3, the position loop controller generates a compensated speed command based on the position error;

[0012] The position PI controller generates a base speed command based on the position error.

[0013] The speed damping module generates a speed damping compensation amount based on the motor side speed, which is used to suppress position overshoot and low-speed oscillation at the output end of the reducer.

[0014] The position sliding mode controller constructs sliding mode variables based on the position error and the rate of change of the position error, and generates sliding mode compensation amount based on the sliding mode variables;

[0015] The recurrent neural network disturbance observation module predicts the magnitude of nonlinear disturbances caused by backlash, friction, and load changes in real time based on the system operating status signal, and adaptively adjusts the sliding mode compensation parameters in the position sliding mode controller module online based on the prediction results.

[0016] The base speed command, the speed damping compensation amount, and the sliding mode compensation amount are superimposed to obtain the compensated speed command;

[0017] S4, input the compensated speed command into the speed loop controller;

[0018] The speed PI controller generates a base current command based on the speed error between the compensated speed command and the motor-side speed feedback.

[0019] The disturbance observer module estimates the system disturbance caused by backlash, friction and load changes based on speed error, motor-side speed feedback and / or motor current feedback signals, and converts the estimated system disturbance into a current compensation amount;

[0020] The current compensation amount is adjusted to the base current command to obtain the compensated current command;

[0021] S5, input the compensated current command into the current loop controller;

[0022] The current loop controller generates motor drive control quantity based on the current error between the compensated current command and the motor current feedback signal.

[0023] S6, drive the motor to run according to the motor drive control quantity, so that the actual position of the reducer output end tracks the position command.

[0024] The position loop controller satisfies:

[0025] ,

[0026] in, Output the speed command to the position loop, i.e., the compensated speed command, k p k i Here are the parameters for the PI controller, e(t) is the position error, and n is the position error. m k is the motor speed. d u is the velocity damping coefficient of the velocity damping module. s This is the sliding compensation amount for the position sliding mode controller.

[0027] The sliding mode variable and sliding mode compensation amount of the position sliding mode controller module satisfy:

[0028] ,

[0029] Where s is the sliding mode variable and us is the sliding mode compensation amount. For position tracking error, , is the boundary layer width of the position tracking error and the sliding mode variable, k1 is the terminal approaching gain, k2 is the linear stability gain, α and P are the weights of the nonlinear terms, where 0 < a < 1 and 0 < P < 1, and c is the sliding mode surface gain coefficient. is the derivative of the position tracking error.

[0030] The described recurrent neural network disturbance observation module takes the position error, the rate of change of the position error, the sliding mode variable, the speed feedback signal, and the current feedback signal as inputs and outputs the disturbance estimation value; it determines whether the current system is in a strong disturbance state based on the disturbance estimation value and adjusts k1 and k2 of the sliding mode compensation amount in real time.

[0031] ,

[0032] where k 1FT , k 2FT are the initial values of the terminal reaching gain and the linear stability gain, β1 and β2 are the gain coefficients of the nonlinear term and the linear term, and f ARNN is the magnitude of the disturbance predicted by the recurrent neural network in real time.

[0033] The described disturbance observer module converts the observed disturbance estimation value into a current compensation amount, and the current compensation amount satisfies:

[0034] ,

[0035] where i qD (k) is the compensation current value of the disturbance observer module, k t is the torque coefficient, i q is the q-axis current feedback value, and J is the equivalent moment of inertia referred to the motor side. is the rate of change of the angular velocity on the motor side, that is, the angular acceleration on the motor side.

[0036] The output q-axis current reference value of the described speed loop controller satisfies:

[0037] ,

[0038] where is the output current command of the speed loop, that is, the compensated q-axis current command, and i qs (k) is the basic q-axis current command generated by the speed loop PI controller according to the speed error.

[0039] Beneficial effects: The present invention combines speed loop disturbance compensation and position loop adaptive sliding mode compensation to improve the position tracking accuracy and anti-disturbance ability of the motor reducer system with backlash; reduces overshoot and oscillation through the speed damping term; and improves the control strength during large disturbances and reduces control chattering during the small error steady state stage by adaptively adjusting the sliding mode gain. Attached Figure Description

[0040] Figure 1 Block diagram of an adaptive finite-time sliding mode position control method for perturbation observation using a recurrent neural network.

[0041] Figure 2 This is a block diagram of the controller structure.

[0042] Figure 3 This is a block diagram of a recurrent neural network structure.

[0043] Figure 4 The waveform diagrams are for the step position experiments of the output terminal at angles of 100, 200, and 300 degrees.

[0044] Figure 5 The experimental waveform diagram is for the sinusoidal position command.

[0045] Figure 6 The waveforms show a comparison of three algorithms for PI, DOB perturbation compensation, and the DANN-FSTM method proposed in this invention for a 100-degree step jump experiment. Figure 6 (a) in the figure represents the output angle curve. Figure 6 (b) in the figure represents the position error angle curve. Detailed Implementation

[0046] The specific embodiments of the present invention are described clearly and completely below with reference to the accompanying drawings. These described embodiments are only a part of, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0047] The present invention will be further described below with reference to specific embodiments. For example... Figure 1 As shown, this embodiment takes a servo positioning system driven by a permanent magnet synchronous motor and a reducer as the object. A position sensor is installed at the output end of the motor and a position sensor is installed at the output end of the reducer. The controller uses the position at the motor end as the current control feedback quantity and the position at the output end of the reducer as the position closed-loop feedback quantity.

[0048] The adaptive sliding mode position control method for a motor reducer system of the present invention employs a cascaded controller to achieve position control of the motor reducer system. The cascaded controller includes a position loop controller, a speed loop controller, and a current loop controller. These controllers are cascaded sequentially. The position loop controller outputs a speed command to the speed loop controller, the speed loop controller outputs a current command to the current loop controller, and the current loop controller outputs a motor drive control quantity to drive the motor. The position loop controller includes a position PI controller, a speed damping module, a position sliding mode controller module, and a recurrent neural network disturbance observation module. The recurrent neural network disturbance observation module predicts nonlinear disturbance values ​​based on the system operating state signal and adjusts the sliding mode parameters of the position sliding mode controller in real time based on these nonlinear disturbance values. The basic speed command output by the position PI controller, the speed damping compensation output by the speed damping module, and the sliding mode compensation output by the position sliding mode controller module are superimposed at the speed command superposition terminal to obtain the output speed command of the position loop controller. The speed loop controller includes a speed PI controller and a disturbance observer module. The base current command output by the speed PI controller is superimposed with the current compensation output by the disturbance observer module to obtain the speed loop output current command. The current loop controller is used to generate motor drive control quantities based on the current command and the current feedback signal.

[0049] The control method includes the following steps:

[0050] S1, Collect the operating status signal of the motor reducer system. The operating status signal includes position command, actual position of reducer output end, motor side speed, actual speed of reducer output end and motor current feedback signal.

[0051] S2, the position error is obtained based on the deviation between the position command and the actual position of the reducer output terminal, the position error change rate is obtained based on the position error, and the position error is input into the position loop controller;

[0052] S3, the position loop controller generates a compensated speed command based on the position error;

[0053] The position PI controller generates a base speed command based on the position error.

[0054] The speed damping module generates a speed damping compensation amount based on the motor side speed, which is used to suppress position overshoot and low-speed oscillation at the output end of the reducer.

[0055] The position sliding mode controller constructs sliding mode variables based on the position error and the rate of change of the position error, and generates sliding mode compensation amount based on the sliding mode variables;

[0056] The recurrent neural network disturbance observation module predicts the magnitude of nonlinear disturbances caused by backlash, friction, and load changes in real time based on the system operating status signal, and adaptively adjusts the sliding mode compensation parameters in the position sliding mode controller module online based on the prediction results.

[0057] The base speed command, the speed damping compensation amount, and the sliding mode compensation amount are superimposed to obtain the compensated speed command;

[0058] S4, input the compensated speed command into the speed loop controller;

[0059] The speed PI controller generates a base current command based on the speed error between the compensated speed command and the motor-side speed feedback.

[0060] The disturbance observer module estimates the system disturbance caused by backlash, friction and load changes based on speed error, motor-side speed feedback and / or motor current feedback signals, and converts the estimated system disturbance into a current compensation amount;

[0061] The current compensation amount is adjusted to the base current command to obtain the compensated current command;

[0062] S5, input the compensated current command into the current loop controller;

[0063] The current loop controller generates motor drive control quantity based on the current error between the compensated current command and the motor current feedback signal.

[0064] S6, drive the motor to run according to the motor drive control quantity, so that the actual position of the reducer output end tracks the position command.

[0065] The position loop controller satisfies:

[0066] ,

[0067] in, Output the speed command to the position loop, i.e., the compensated speed command, k p k i Here are the parameters for the PI controller, e(t) is the position error, and n is the position error. m k is the motor speed. d u is the velocity damping coefficient of the velocity damping module. s This is the sliding compensation amount for the position sliding mode controller module.

[0068] Figure 1 The expression for the mid-position sliding mode controller module is: ,

[0069] ,

[0070] where s is the sliding mode variable, us is the sliding mode compensation quantity, is the position tracking error, , are the boundary layer widths of the position tracking error and the sliding mode variable, k1 is the terminal approaching gain, k2 is the linear stability gain, α and P are the weights of the non-linear terms, and 0 < a < 1, 0 < P < 1, c is the sliding mode surface gain coefficient, is the derivative of the position tracking error.

[0071] The adaptive sliding mode compensation quantity gains k1 and k2 can be expressed as: ,

[0072] where k 1FT , k 2FT are the initial values of the terminal reaching gain and the linear stability gain, β1 and β2 are the gain coefficients of the non-linear and linear terms, f ARNN is the disturbance magnitude predicted by the recurrent neural network in real time.

[0073] The sliding mode compensation quantity gain is adaptively adjusted in real time according to the disturbance magnitude predicted by the recurrent neural network. When the error changes rapidly or the commutation disturbance of the system increases, the sliding mode variable increases and the sliding mode compensation term increases; when the system enters the vicinity of the target position and the disturbance weakens, the sliding mode variable decreases and the sliding mode compensation term weakens accordingly, so as to avoid generating an overly strong control effect within a small error range.

[0074] Figure 1 The expression of the disturbance observer module in

[0075] ,

[0076] where f d is the initial value of the observed disturbance, α is the first-order low-pass filtering coefficient, is the filtered disturbance value, i qD is the disturbance observer compensation current quantity.

[0077] The speed loop disturbance observer estimates the equivalent disturbance received by the system based on the current command, the actual rotational speed and the motor parameters. The equivalent disturbance includes the disturbances caused by load changes, friction changes and backlash contact state changes. The disturbance estimation value is not directly used for the control output, but is first filtered and then converted into a current compensation quantity to avoid high-frequency noise entering the current loop.

[0078] Figure 3The diagram shows the structure of a recurrent neural network (RNN) for real-time prediction of system disturbances. During control, the RNN continuously receives position error, error rate of change, and sliding mode variable signals, and outputs a disturbance estimate. This disturbance estimate is used to determine whether the system is under strong disturbance. A large disturbance estimate indicates that the system may be in a backlash switching, commutation impact, or load disturbance phase; a small disturbance estimate indicates that the system is close to steady state or the disturbance is weak.

[0079] In this embodiment, the adaptive neural network can adopt a lightweight network structure. Its input signal is normalized and limited before entering the network, and the network output is also limited to ensure that the controller can run in real time in the embedded chip.

[0080] Figure 4 It can be seen that the algorithm of this invention can achieve good tracking performance for position step commands in different ranges, with fast response speed and no overshoot.

[0081] Figure 5 The algorithm of this invention can track the response well by using a sinusoidal command indicating the position.

[0082] Figure 6 The waveform of the 100-degree position step command experiment shows that the proposed algorithm has the fastest response speed, the shortest rise time, and the smallest overshoot. From the perspective of position error, the algorithm proposed in this invention has the smallest final angle error.

[0083] Through the above implementation methods, the system can quickly approach the target position during the large error stage, improve the anti-disturbance capability during the commutation and backlash switching stages, reduce control chattering during the small error steady-state stage, and quickly converge the error, thereby achieving high-precision position control of the backlash motor reducer system.

Claims

1. An adaptive sliding mode position control method for a motor reducer system, characterized in that: The method employs a cascaded controller to achieve position control of the motor reducer system. The cascaded controller includes a position loop controller, a speed loop controller, and a current loop controller. The position loop controller, speed loop controller, and current loop controller are cascaded sequentially. The position loop controller outputs a speed command to the speed loop controller, the speed loop controller outputs a current command to the current loop controller, and the current loop controller outputs a motor drive control quantity to drive the motor to run.

2. The adaptive sliding mode position control method for the motor reducer system according to claim 1, characterized in that: The position loop controller includes a position PI controller, a velocity damping module, a position sliding mode controller module, and a recurrent neural network disturbance observation module; The recurrent neural network disturbance observation module predicts nonlinear disturbance values ​​based on the system operating status signal, and adjusts the sliding mode parameters of the position sliding mode controller in real time based on the nonlinear disturbance values. The basic speed command output by the position PI controller, the speed damping compensation output by the speed damping module, and the sliding mode compensation output by the position sliding mode controller module are superimposed at the speed command superposition terminal to obtain the output speed command of the position loop controller.

3. The adaptive sliding mode position control method for the motor reducer system according to claim 2, characterized in that: The speed loop controller includes a speed PI controller and a disturbance observer module. The base current command output by the speed PI controller is superimposed with the current compensation amount output by the disturbance observer module to obtain the speed loop output current command.

4. The adaptive sliding mode position control method for the motor reducer system according to claim 3, characterized in that: The current loop controller is used to generate motor drive control quantities based on current commands and current feedback signals; the control method includes the following steps: S1, Collect the operating status signal of the motor reducer system. The operating status signal includes position command, actual position of reducer output end, motor side speed, actual speed of reducer output end and motor current feedback signal. S2, the position error is obtained based on the deviation between the position command and the actual position of the reducer output terminal, the position error change rate is obtained based on the position error, and the position error is input into the position loop controller; S3, the position loop controller generates a compensated speed command based on the position error; The position PI controller generates a base speed command based on the position error. The speed damping module generates a speed damping compensation amount based on the motor side speed, which is used to suppress position overshoot and low-speed oscillation at the output end of the reducer. The position sliding mode controller constructs sliding mode variables based on the position error and the rate of change of the position error, and generates sliding mode compensation amount based on the sliding mode variables; The recurrent neural network disturbance observation module predicts the magnitude of nonlinear disturbances caused by backlash, friction, and load changes in real time based on the system operating status signal, and adaptively adjusts the sliding mode compensation parameters in the position sliding mode controller module online based on the prediction results. The base speed command, the speed damping compensation amount, and the sliding mode compensation amount are superimposed to obtain the compensated speed command; S4, input the compensated speed command into the speed loop controller; The speed PI controller generates a base current command based on the speed error between the compensated speed command and the motor-side speed feedback. The disturbance observer module estimates the system disturbance caused by backlash, friction and load changes based on speed error, motor-side speed feedback and / or motor current feedback signals, and converts the estimated system disturbance into a current compensation amount; The current compensation amount is adjusted to the base current command to obtain the compensated current command; S5, input the compensated current command into the current loop controller; The current loop controller generates motor drive control quantity based on the current error between the compensated current command and the motor current feedback signal. S6, drive the motor to run according to the motor drive control quantity, so that the actual position of the reducer output end tracks the position command.

5. The adaptive sliding mode position control method for the motor reducer system according to claim 4, characterized in that, The position loop controller satisfies: ， in, Output the speed command to the position loop, i.e., the compensated speed command, k p k i Here are the parameters for the PI controller, e(t) is the position error, and n is the position error. m k is the motor speed. d u is the velocity damping coefficient of the velocity damping module. s This is the sliding compensation amount for the position sliding mode controller.

6. The adaptive sliding mode position control method for a motor reducer system according to claim 5, characterized in that, The sliding mode variable and sliding mode compensation amount of the position sliding mode controller module satisfy: ， where s is the sliding mode variable, us is the sliding mode compensation amount, is the position tracking error, and are the boundary layer widths of the position tracking error and the sliding mode variable, k1 is the terminal approaching gain, k2 is the linear stability gain, α and P are the weights of the nonlinear terms, and 0 < a < 1, 0 < P < 1, c is the sliding mode surface gain coefficient, is the derivative of the position tracking error.

7. The adaptive sliding mode position control method for a motor reducer system according to claim 6, characterized in that, The recurrent neural network disturbance observation module takes position error, position error change rate, sliding mode variable, velocity feedback signal, and current feedback signal as inputs and outputs disturbance estimate. It uses the disturbance estimate to determine whether the system is under strong disturbance and adjusts the sliding mode compensation values ​​k1 and k2 in real time. ， Where, k 1FT k 2FT The initial values ​​for the terminal to achieve the gain and linear stable gain are given by β1 and β2, which are the gain coefficients for the nonlinear and linear terms, respectively. ARNN This refers to the magnitude of the perturbation predicted in real time by the recurrent neural network.

8. The adaptive sliding mode position control method for a motor reducer system according to claim 7, characterized in that, The disturbance observer module converts the observed disturbance estimate into a current compensation amount, which satisfies the following: ， Among them, i qD (k) represents the compensation current value of the disturbance observer module, k t i is the torque coefficient. q Here, J is the q-axis current feedback value, and J is the equivalent moment of inertia referred to the motor side. This is the rate of change of the angular velocity of the motor side, i.e., the angular acceleration of the motor side.

9. The adaptive sliding mode position control method for a motor reducer system according to claim 8, characterized in that, The speed loop controller outputs a q-axis current setpoint that satisfies the following: ， in, This is the speed loop output current command, i.e., the compensated q-axis current command. qs (k) is the basic q-axis current command generated by the speed loop PI controller based on the speed error.

10. The adaptive sliding mode position control method for a motor reducer system according to claim 9, characterized in that, The motor reducer system includes a permanent magnet synchronous motor, a reducer, a motor-side position acquisition module, and a reducer output position acquisition module. The motor-side position acquisition module is used to acquire the motor-side position, and the reducer output position acquisition module is used to acquire the actual position of the reducer output.

Images