A method for mechanical parameter identification of robot joint servo system

Abstract

The present invention discloses a method for identifying mechanical parameters of a robot joint servo system. The method models the robot joint servo system as a whole, considers the influence of transmission elements represented by harmonic reducers on motion and torque, performs equivalent conversion on the torque and moment of inertia on the output side of the joint, and establishes a new integrated joint model on the side of the joint servo motor; an expanded sliding mode observer is designed according to the integrated joint model, and the moment of inertia, load torque and viscous friction parameters of the system are expanded into new state variables. The three mechanical parameters can be identified simultaneously through an observer, and the Coulomb friction component of the system can be identified when unloaded. In addition, the present invention designs a sliding mode gain self-correction coefficient based on the rate of change of the observed quantity, which can increase the convergence speed of the algorithm while obtaining higher identification accuracy.

Description

Technical Field

[0001] The invention belongs to the technical field of servo motor system control, and in particular relates to a method for identifying mechanical parameters of a robot joint servo system. Background Art

[0002] The bottom layer of robot motion control is the control of joint servo motor system. For robot joint servo system, field oriented vector control (FOC) algorithm is often used. The mechanical parameters of the system are of great significance to its control effect: the system moment of inertia is a necessary parameter for the design of speed loop PI controller, and accurate load torque and joint friction information are the basis of joint torque feedback control and robot dynamics model. However, during the movement of the robot, the equivalent inertia and load torque of each joint will change with the working conditions and the posture of the robot arm, and the friction torque will also change with the joint speed; the change of mechanical parameters represented by the moment of inertia, load torque and friction parameters has an important impact on the performance of the entire joint control. Therefore, real-time identification of these three mechanical parameters is crucial to optimize the control of the robot joint motor system.

[0003] In current research, the commonly used algorithms for online identification of mechanical parameters of motor systems mainly include extended Kalman filter method, model reference adaptive system method, neural network and observer-based method. The extended Kalman filter method has complex matrix operations, and its covariance calculation and design are difficult when identifying multiple parameters; when the model reference adaptive system method is applied to multiple parameter estimation, the design of its adaptive mechanism will greatly increase the complexity and calculation amount of the algorithm, and its robustness is poor; a series of intelligent algorithms represented by neural networks have the highest accuracy and are suitable for complex nonlinear systems, but the calculation amount is extremely large and is not suitable for fast online identification algorithms; the defects of the above algorithms limit their application in online multi-parameter identification.

[0004] In recent years, due to its better robustness and insensitivity to parameter changes, the extended sliding mode observer (ESMO) has been more widely used in online multi-parameter identification, such as [Liang Ge, Huang Shoudao, Li Mengdi, et al. Mechanical parameter identification of permanent magnet synchronous motor based on high-order fast terminal sliding mode disturbance observer [J]. Transactions of China Electrotechnical Society, 2020, 35(S2): 395-403] and [Wang Tao. Research on position sensorless control of permanent magnet synchronous motor based on online parameter identification [D]. Hefei University of Technology, 2023]. However, these existing estimation methods either only identify one mechanical parameter or identify multiple parameters through multiple observers, resulting in increased algorithm complexity. In addition, the gain coefficient in existing ESMO research is usually set as a fixed constant, and the convergence speed and accuracy of the error equation are also determined accordingly. In actual use, multiple parameter debugging experiments are required to obtain a gain coefficient that compromises between speed and accuracy. Such a value-taking method requires re-adjustment when facing different systems, and cannot take into account both the speed and accuracy of identification.

[0005] It can be seen that the current parameter identification algorithm still exhibits problems of algorithm complexity and difficulty in balancing identification speed and accuracy when dealing with the needs of simultaneous identification of multiple parameters. How to design an online simultaneous identification method for multiple parameters that can achieve rapid identification of parameters while achieving high identification accuracy is still quite challenging. Summary of the invention

[0006] In view of the above, the present invention provides a method for identifying the mechanical parameters of a robot joint servo system, which utilizes an extended sliding mode observer to realize the simultaneous identification of three mechanical parameters of the system's moment of inertia, viscous friction coefficient, and load torque, and introduces a self-correction coefficient to enable adaptive correction of the sliding mode gain parameters, thereby being able to obtain high-precision identification results while improving the identification speed.

[0007] A method for identifying mechanical parameters of a robot joint servo system comprises the following steps:

[0008] (1) Considering the influence of the transmission elements represented by the harmonic reducer on the motion and torque in the system, a new integrated joint model is established on the servo motor side of the system;

[0009] (2) Designing an extended sliding mode observer based on the equivalent motion equation of the integrated joint model to identify the equivalent moment of inertia, equivalent load torque and equivalent viscous friction coefficient on the servo motor side;

[0010] (3) The gain coefficient of the extended sliding mode observer is optimized based on the rate of change of the observed quantity to improve the identification accuracy.

[0011] Furthermore, the expression of the integrated joint model in step (1) is as follows:

[0012]

[0013] in: T e is the electromagnetic torque of the servo motor, J m is the moment of inertia of the servo motor, J HD is the moment of inertia of the harmonic reducer, N is the reduction ratio of the harmonic reducer, J L is the moment of inertia of the load, oh m is the speed of the servo motor, for oh m The first derivative of T fm is the friction torque on the servo motor side, T sb is the friction torque on the load side, T L is the load torque.

[0014] Furthermore, the friction torque T fm and T sb The expression is as follows:

[0015]

[0016] in: T c is the Coulomb friction coefficient on the servo motor side, sgn( ) is the sign function, B m is the viscous friction coefficient on the servo motor side, T r is the Coulomb friction torque of the Stribeck model, s r is the viscous friction coefficient of the Stribeck model, oh s is the speed of the Stribeck model (model constant), T s is the maximum static friction torque of the Stribeck model, oh out is the output speed of the harmonic reducer, oh out =- oh m / N .

[0017] Furthermore, the expression of the equivalent motion equation in step (2) is as follows:

[0018]

[0019]

[0020]

[0021]

[0022] in: J eq is the equivalent moment of inertia of the servo motor side, B is the equivalent viscous friction coefficient on the servo motor side, T LC is the equivalent load torque on the servo motor side.

[0023] Furthermore, the expression of the extended sliding mode observer in step (2) is as follows:

[0024]

[0025] in: l =1 / J eq , , , , They are oh m , l , B , T LC The observed value of , , , They are , , , The first derivative of S The sliding surface is , k is the switching gain coefficient of the extended sliding mode observer, g 1~ g 3 is the feedback gain coefficient of the extended sliding mode observer.

[0026] Furthermore, in step (3), the gain coefficient of the extended sliding mode observer is optimized and designed, and the specific expression is as follows:

[0027]

[0028]

[0029]

[0030]

[0031] in: a 1~ a 3 is a given smaller constant, A is a constant greater than or equal to 1, b 1~ b 3 is the self-correction coefficient.

[0032] Furthermore, the self-correction coefficient b 1~ b The expression of 3 is as follows:

[0033]

[0034]

[0035]

[0036] in: d is the correction intensity coefficient (a constant greater than 0), err 1. err 2. err 3 corresponds to , , The rate of change.

[0037] Furthermore, the rate of change err 1. err 2. err The expression of 3 is as follows:

[0038]

[0039]

[0040]

[0041] in: Corresponding to the last ten cycles in the last twenty cycles , , The average value of Corresponding to the first ten cycles in the last twenty cycles , , The average value of .

[0042] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and the processor is used to execute the computer program to implement the above-mentioned method for identifying mechanical parameters of a robot joint servo system.

[0043] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for identifying mechanical parameters of a robot joint servo system is implemented.

[0044] Based on the above technical solution, the present invention has the following beneficial technical effects:

[0045] 1. The present invention takes into account the influence of transmission elements represented by harmonic reducers on motion and torque, performs equivalent conversion of the torque and moment of inertia on the output side of the joint, and establishes a new integrated joint model on the joint servo motor side, which is suitable for the technical characteristics of modern joint module integration.

[0046] 2. The parameter identification method based on the expanded sliding mode observer of the present invention has a simple structure, can realize the online identification of mechanical parameters of multiple systems at the same time, improve the efficiency of the algorithm, and is easy to be deployed on various low-computing power control platforms, thereby realizing the online identification of mechanical parameters of the robot joint servo system.

[0047] 3. The observer gain parameter used in the present invention is self-corrected in real time based on the observed quantity change rate, which can increase the algorithm convergence speed while obtaining higher identification accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 Schematic diagram of the control structure of the robot joint servo system in an embodiment of the present invention.

[0049] Figure 2 Schematic diagram of the structure of the extended sliding mode observer in an embodiment of the present invention, where 1 / s represents the differential operator.

[0050] Figure 3 Schematic diagram of the waveform of the square wave speed instruction in the embodiment of the present invention.

[0051] Figure 4 Schematic diagram of parameter identification results in an embodiment of the present invention.

[0052] Figure 5 Schematic diagram of parameter identification results without correcting the gain coefficient in an embodiment of the present invention. DETAILED DESCRIPTION

[0053] In order to describe the present invention more specifically, the technical solution of the present invention is described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0054] In this embodiment, a rated power of 0.5 kW, a rated torque of 56 Nm, and a moment of inertia of 4.09×10 -4 kg×m 2 Taking the joint module with a viscous friction coefficient of 0.0035, a rated speed of 1500r / min and a reduction ratio of 100 as an example, the proposed parameter identification algorithm is verified under the working condition of a load torque of 15Nm (converted to the load torque of the motor side is 0.15Nm).

[0055] like Figure 1 As shown, in this embodiment, the servo system adopts a dual closed-loop control framework based on a magnetic field oriented control strategy, which generally includes a speed loop and a current loop dual-loop control structure; the system first obtains the three-phase current through a sampling circuit i a and i b , which undergo Clarke transformation to obtain the current components of the αβ axis i α and i β , and then the current components of the dq axis are obtained through Park transformation i d and i q ; In the speed loop structure, the speed reference value is given And the actual speed obtained by the encoder oh m As the input of the speed loop PI controller, the speed loop PI controller outputs the electromagnetic torque reference signal, and the electromagnetic torque reference passes through the motor current-torque coefficient k t Convert the q current reference command to the actual sampling i q Input to the current loop PI controller; in the current loop structure, The control means that the d-axis current reference is 0, which is different from the actual i d The two PI controllers of the current loop output the d-axis and q-axis reference voltages respectively. U d and U q , and then use the rotor position angle i e The reference voltage is obtained by inverse Park transformation U α and U β The current loop outputs the three-phase duty cycle, which is then calculated by the SVPWM (space vector pulse width modulation) algorithm and given to the three-phase driver. The DC voltage of the driver is Udc , achieving dual closed-loop high-precision control. To implement the online parameter identification algorithm, the joint needs to be operated in the speed closed-loop mode, and the motor electromagnetic torque is converted according to the converted integrated joint model and the corresponding expanded sliding mode observer. T e With speed oh m As input signals, three mechanical parameters can be identified online. The specific implementation process is as follows:

[0056] (1) Establish an integrated joint model.

[0057] In mathematical modeling, the influence of transmission elements such as harmonic reducers on motion and torque is considered. As the core element of the joint, the harmonic reducer plays the role of reducing the speed and amplifying the torque. The expressions for converting the speed, torque and inertia on both sides of the harmonic reducer are as follows:

[0058]

[0059]

[0060]

[0061] in: N is the reduction ratio (taken as 100 in this embodiment); oh in , T in , J in Represents the speed, torque and equivalent moment of inertia of the input side of the harmonic reducer; oh out , T out , J out are the speed, torque and equivalent moment of inertia on the output side of the harmonic reducer.

[0062] For the establishment of the friction model, the Stribeck model is used for the friction on the output side of the low-speed joint; the Coulomb plus viscous friction model is used on the input side of the high-speed harmonic reducer (i.e., the motor output side). The expressions of each friction component are as follows:

[0063]

[0064] in: T fm is the friction torque on the servo motor side, T sb is the friction torque on the load side, T c is the Coulomb friction coefficient on the servo motor side, sgn( ) is the sign function,B m is the viscous friction coefficient on the servo motor side, T r is the Coulomb friction torque of the Stribeck model, s r is the viscous friction coefficient of the Stribeck model, oh s is the speed of the Stribeck model (model constant), T s is the maximum static friction torque of the Stribeck model.

[0065] Combining friction with the motion equation of the servo motor, we can get:

[0066]

[0067] in: T e is the electromagnetic torque of the servo motor, J m is the moment of inertia of the servo motor, J HD is the moment of inertia of the harmonic reducer, J L is the moment of inertia of the load, oh m The speed of the servo motor is equal to oh in , T L is the load torque.

[0068] Since the nonlinear component of the Stribeck model accounts for a small proportion, and after the reduction ratio is converted, the proportion in the servo motor side torque is greatly reduced to less than one percent, the nonlinear exponential term of the Stribeck model on the joint output side can be ignored when analyzing the motor side motion equation. At this time, the expression of the joint equivalent motion equation is as follows:

[0069]

[0070]

[0071]

[0072]

[0073] in: J eq is the equivalent moment of inertia of the servo motor side, B is the equivalent viscous friction coefficient on the servo motor side, T LCis the equivalent load torque on the servo motor side (when T L =0 corresponds to no-load, T LC is the Coulomb friction torque).

[0074] (2) Design an extended sliding mode observer for synchronously identifying three mechanical parameters.

[0075] Based on the joint equivalent motion equation of the integrated model, the system equivalent moment of inertia, equivalent viscous friction coefficient, and equivalent load torque are expanded as system state variables, and it is assumed that these mechanical parameters are constant in a very short time within a control cycle. On this basis, the following design is made: Figure 2 The extended sliding mode observer shown in Figure 1 has the following expression:

[0076]

[0077] in: l =1 / J eq , , , , They are oh m , l , B , T LC The observed value of S The sliding surface is , k is the switching gain coefficient of the extended sliding mode observer, g 1~ g 3 is the feedback gain coefficient of the extended sliding mode observer.

[0078] To ensure the stability and convergence of the observer, the setting of the gain parameter needs to meet the stability and convergence requirements. For stability analysis, the Lyapunov function is taken V =0.5 S 2 , the expression of the system stability criterion is:

[0079]

[0080] in: , from which we can get the expression of the value range of the observer switching gain:

[0081]

[0082] In this embodiment, k=-5500; For the convergence analysis of the observer, the solution of the error equation can essentially be transformed into the form of a differential equation. When the observer converges to the vicinity of the sliding surface, , the error equations of the three mechanical parameter variables can be rewritten as follows:

[0083]

[0084]

[0085]

[0086] in: R 1. R 2. R 3 is the gain coefficient, that is , , ; , , . At this time, if the non-homogeneous terms of the above equations , , In steady state, it can be controlled near 0, that is, the equation is a sub-differential equation, and the parameter estimation error can converge exponentially to 0. That is, in steady state, if it satisfies , , , the expression of the solution of the error equation is as follows:

[0087]

[0088]

[0089]

[0090] Set a reasonable feedback gain g 1. g 2. g 3 makes the exponential term greater than 0, which means that the observation error can converge exponentially to 0. At this time, the expression that should be taken is:

[0091]

[0092]

[0093]

[0094] in: a 1. a 2. a 3 is a constant greater than 0. The specific coefficient of the feedback gain used in this embodiment is a1=10, a 2=1, a 3=0.2.

[0095] (3) Design the sliding mode gain self-correction coefficient based on the rate of change of the observed quantity.

[0096] The gain parameters set in step (2) are self-corrected. In the early stage of identification, a larger gain coefficient is used to accelerate the convergence of the observer, so that the observed value converges quickly to a range with a smaller error. After that, a smaller gain coefficient is used to obtain an observation result with higher accuracy and less jitter, which can improve the identification speed while ensuring high identification accuracy. The size of the correction value is based on the rate of change of the observed value. Since the variables of the sliding mode observer jitter along the sliding surface, it is difficult to judge the rise or fall of the observed value between two consecutive observations. Therefore, the average value change of the observed value of multiple cycles is used to judge the change rate of the observed value. The expression for calculating the rate of change is as follows:

[0097]

[0098]

[0099]

[0100] in: err 1. err 2. err 3 corresponds to , , The rate of change, Corresponding to the last ten cycles in the last twenty cycles , , The average value of Corresponding to the first ten cycles in the last twenty cycles , , The average value of .

[0101] The expression for the self-correction coefficient based on the rate of change is as follows:

[0102]

[0103]

[0104]

[0105] in: b 1~ b 3 is the self-correction coefficient, d is the correction strength coefficient (a constant greater than 0), the larger the value, the stronger the correction effect.d Take 1.

[0106] After the self-correction coefficient is introduced, the expression of the sliding mode gain parameter is as follows:

[0107]

[0108]

[0109]

[0110]

[0111] (4) Given a speed command, parameter identification is achieved.

[0112] In this embodiment, a periodic square wave speed signal is selected to simulate the speed fluctuation and speed command change during the operation of the motor. The expression of the given speed signal is as follows:

[0113]

[0114] in: oh 1. oh 2 is the low speed value and high speed value of the square wave (rpm), T Indicates the period of the square wave in seconds. The duty cycle of the square wave instruction is 50%. The value in this embodiment is oh 1=200, oh 2=500, T = 1 second.

[0115] In this embodiment, the speed command waveform is as follows: Figure 3 As shown, under this working condition, the identification method of the present invention is applied to test the embodiment, and the identification results are as follows Figure 4 As shown in the figure, the identification results without introducing the self-correction coefficient are as follows Figure 5 As shown, , , Corresponding to , , The results show that the expanded sliding mode observer designed in the present invention can identify multiple mechanical parameters at the same time, and on this basis, the self-correction coefficient is used to further improve the identification speed and accuracy of the system. After the self-correction coefficient is introduced, the time required for the identification of the moment of inertia of the system is shortened by 57%, the steady-state jitter is reduced by 74%, the time required for the identification of the viscous friction coefficient is shortened by 55%, the steady-state jitter is reduced by 62.5%, the time required for the load torque identification is shortened by 60%, and the steady-state jitter is reduced by 50%. It can be seen that the improvements in speed and accuracy are quite obvious.

[0116] The above description of the embodiments is to facilitate the understanding and application of the present invention by those skilled in the art. It is obvious that those skilled in the art can easily make various modifications to the above embodiments and apply the general principles described herein to other embodiments without creative work. Therefore, the present invention is not limited to the above embodiments. Improvements and modifications made by those skilled in the art to the present invention based on the disclosure of the present invention should be within the protection scope of the present invention.

Claims

1. A method for identifying mechanical parameters of a robot joint servo system, comprising the following steps: (1) Considering the influence of the transmission elements on the motion and torque in the system, an integrated joint model is established on the servo motor side of the system; (2) According to the equivalent motion equation of the integrated joint model, an extended sliding mode observer is designed to identify the equivalent moment of inertia, equivalent load torque and equivalent viscous friction coefficient on the servo motor side; the expression of the extended sliding mode observer is as follows: in: They are The first derivative of They are ω m ,λ,B,T LC The observed value, ω m is the speed of the servo motor, λ=1 / J eq , J eq is the equivalent moment of inertia on the servo motor side, B is the equivalent viscous friction coefficient on the servo motor side, T LC is the equivalent load torque on the servo motor side, T e is the electromagnetic torque of the servo motor, k is the switch gain coefficient of the extended sliding mode observer, sgn() is the sign function, g1~g3 are the feedback gain coefficients of the extended sliding mode observer, S is the sliding surface (3) optimizing the gain coefficient of the extended sliding mode observer based on the observed quantity change rate to improve the identification accuracy; The specific expression of the gain coefficient after optimization design is as follows: Among them: a1~a3 are given constants, A is a constant greater than or equal to 1, and b1~b3 are self-correction coefficients.

2. The method for identifying mechanical parameters of a robot joint servo system according to claim 1, characterized in that: The expression of the integrated joint model in step (1) is as follows: Among them: J m is the moment of inertia of the servo motor, J HD is the moment of inertia of the harmonic reducer, N is the reduction ratio of the harmonic reducer, J L is the moment of inertia of the load, ω m The first derivative of fm is the friction torque on the servo motor side, T sb is the friction torque on the load side, T L is the load torque.

3. The method for identifying mechanical parameters of a robot joint servo system according to claim 2, characterized in that: The friction torque T fm and T sb The expression is as follows: Where: T c is the Coulomb friction coefficient on the servo motor side, B m is the viscous friction coefficient on the servo motor side, T r is the Coulomb friction torque of the Stribeck model, σ r is the viscous friction coefficient of the Stribeck model, ω s is the speed of the Stribeck model, T s is the maximum static friction torque of the Stribeck model, ω out is the speed ω output by the harmonic reducer out =-ω m / N.

4. The method for identifying mechanical parameters of a robot joint servo system according to claim 3, characterized in that: The expression of the equivalent motion equation in step (2) is as follows:

5. The method for identifying mechanical parameters of a robot joint servo system according to claim 1, characterized in that: The expressions of the self-correction coefficients b1 to b3 are as follows: b1=1+d·err1 b2=1+d·err2 b3=1+d·err3 Where: d is the correction intensity coefficient, err1, err2, err3 correspond to The rate of change.

6. The method for identifying mechanical parameters of a robot joint servo system according to claim 5, characterized in that: The expressions of the change rates err1, err2, and err3 are as follows: in: Corresponding to the last ten cycles in the last twenty cycles The average value of Corresponding to the first ten cycles in the last twenty cycles The average value of .

7. A computer device comprising a memory and a processor, wherein a computer program is stored in the memory, wherein: The processor is used to execute the computer program to implement the method for identifying mechanical parameters of a robot joint servo system as claimed in any one of claims 1 to 6.

8. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method for identifying mechanical parameters of a robot joint servo system as claimed in any one of claims 1 to 6 is implemented.

Images