Robot servo parameter self-adaption method and system based on parameter set

Abstract

The invention relates to the field of robot servo parameter self-adaption, in particular to a robot servo parameter self-adaption method and a robot servo parameter self-adaption system based on a parameter set, and the method comprises the following specific steps: S1, for a basic mode, a standard control mode, a low-speed precision mode and a high-speed positioning mode in a parameter set concept, obtaining a parameter set; a robot complete machine dynamic model based on a current real load needs to be obtained in advance, and dynamic parameters are calculated based on the dynamic model; s2, performing corresponding servo parameter adaptive strategy switching according to the selected mode; the system comprises a control system with the built-in method, factory preset fixed servo parameters are adopted in a basic mode, and servo parameters in other three modes adopt a self-adaptive adjustment mode combining kinetic parameters and servo loops. According to the working condition type of the robot, the corresponding mode is determined to carry out servo parameter self-adaption, and therefore the problem that the whole machine performance is reduced due to the fact that the consistency of the servo parameters becomes poor is solved.

Description

Technical Field

[0001] This invention relates to the field of robot servo parameter adaptation, specifically a robot servo parameter adaptation method and system based on a parameter set. Background Technology

[0002] In the patent "Adaptive Method and System for Robot Servo Parameters Based on Dynamic Model", adaptive adjustment of servo loop parameters is achieved based on robot dynamic parameters. However, a single parameter tuning strategy is used under any working condition, without fully considering the adaptation and optimization of parameters under specific working conditions. The parameter tuning strategy is not tied to the robot working condition, thus hindering the upper limit of servo control.

[0003] With the development and application of robots in various industries, the types of working conditions and loads that robots undertake are also increasing. Under different working conditions and loads, robots use uniform servo loop parameters, which can easily lead to a decrease in the consistency of servo parameters, thereby affecting the overall performance of the robot. Therefore, it is of great significance to realize a parameter adaptive strategy for robots for different working conditions and loads.

[0004] Currently, in practical applications, servo parameters are tuned based on standard load blocks before robots leave the factory. However, the types of loads at customer sites are complex, and sometimes large load biases can cause a decrease in the consistency of servo parameters, resulting in robot jitter. Therefore, the servo parameters preset at the factory cannot widely meet the application needs of customers at their sites. Summary of the Invention

[0005] To address the aforementioned problems, this invention proposes a robot servo parameter adaptive method and system based on a parameter set.

[0006] A method for adaptive robot servo parameters based on a parameter set, the specific steps of which are as follows: S1. For the basic mode, standard control mode, low-speed precision mode and high-speed positioning mode in the parameter set concept, it is necessary to obtain the robot's overall dynamic model based on the current real load in advance, and calculate the dynamic parameters based on this dynamic model. S2. Switch the corresponding servo parameter adaptive strategy according to the selected mode.

[0007] Step S1 specifically includes the following steps: S11. Based on the 3D CAD model of the robot body, obtain the theoretical dynamic parameters of each link; S12. Drive the robot to execute the preset trajectory and simultaneously collect joint position, velocity, acceleration and torque data. Then, calibrate the aforementioned theoretical dynamic parameters through robot dynamic parameter identification. S13. Based on the actual dynamic parameters obtained in step S12, the robot is instructed to execute a specific trajectory under the current load, and the mass, center of mass position and inertia tensor of the current real load are obtained through the load identification module, i.e., the load dynamic parameters. S14. Integrate the theoretical dynamic parameters obtained in step S11, the actual dynamic parameters identified in step S12, and the load dynamic parameters obtained in step S13, and import them into the robot dynamic model to obtain the robot's feedforward inertia and feedforward torque under the current real load conditions online. S15. Based on the joint equivalent inertia obtained in real time in S14, the servo parameter tuning module dynamically adjusts the control parameters of the position loop and the speed loop. The servo control algorithm generates a basic torque command based on this. This command is superimposed on the feedforward torque provided in step S14 in the servo control loop and finally synthesized into a total torque command for driving the motor. S16. To construct a parameter adaptive mechanism, the robot motion is driven by the total torque command synthesized in step S15, and the joint state during operation is fed back to the dynamic model in real time.

[0008] In step S11, the theoretical parameters include mass, center of mass, inertia, and tensor types.

[0009] In step S12, the robot's dynamic parameter identification module consists of two parts: robot dynamics calculation and robot dynamic parameter identification. The robot dynamics calculations are performed using the following steps: Step 1: Calculate the driving torque of each axis of the robot. For a typical serial robot, its dynamic equations can be expressed in the following matrix form: ; In the formula, M is the inertia matrix; N is collectively referred to as the nonlinear force matrix of the system, which includes Coriolis force, centripetal force, gravity and friction. Step 2: By linearizing equation (1), we can obtain: ; In the formula, Y is the motion matrix; P is the parameter matrix; In the above formula, P is a parameter matrix containing the connecting rod mass m, the moment of mass s, the inertia tensor I, and the friction force f, which can be expressed as follows: ; Step 3: By eliminating By identifying the linearly dependent terms in P and removing their corresponding elements, we can deduce... A set of maximally linearly independent sets is obtained, and the corresponding minimum dynamic parameter matrix is ​​obtained: ; In the formula, Y^ is the motion matrix to be identified; P^ is the parameter matrix to be identified.

[0010] The steps for identifying the robot's dynamic parameters are as follows: Step 1: Run the robot and collect actual motion information and torque data for each axis. Based on the data, construct the following expression: ; In the formula, τ i (i=1...k) is the robot driving torque matrix; Step 2: Solve equation (5) using the least squares method to obtain the optimal estimate of the robot's dynamic parameters P^, and then complete the entire identification process.

[0011] In step S13, the load parameter identification module consists of two parts: load identification model establishment and robot load parameter identification. The steps for the load identification model are as follows: Step 1: First, obtain the reference torque of the body joints without load. Then, treat the additional torque generated by the load as a linear superposition component to solve for the load parameters. The specific process is as follows: ; In the formula, τ Total τ is the torque required for the robot to operate under the current load; Base The basic torque required for the robot body to operate (unloaded); Δτ L This refers to the incremental torque generated under the current load. Step 2: The transpose of the Jacobian matrix provides a linear mapping relationship; based on this relationship, the torque contribution of the end load at each joint can be analyzed and decoupled using the transpose of the Jacobian matrix. The corresponding load identification model can be expressed as:

[0012] In the formula, J is the Jacobian matrix; Y L P is the observation matrix to be identified; L The load parameter matrix to be identified; The steps for identifying robot load parameters are as follows: Using the same formula (5), the robot is run and the actual motion information and torque data of each axis are collected to construct the following expression: ; In the formula, τ L_i (i=1...k) is the load torque matrix; Similarly, the least squares method is used to perform P on equation (8). LThe optimal estimate is then obtained, thus completing the entire identification process.

[0013] The specific servo parameter tuning strategy for step S15 is as follows: The ideal speed open-loop transfer function based on a series PI converter is: ; In the formula, K pS τ is the proportional gain of the speed loop PI controller. i The integral time constant of the speed loop PI controller is J; the joint inertia is T. c The current loop time constant; To eliminate the dependence of the speed loop control performance on the inertia of the controlled object, the proportional gain in equation (9) is set to K. pS = K speed *J, then: ; Obviously, equation (10) is a typical type II system, and the open-loop cutoff frequency ω can be approximated as the velocity. cS = K speed ; The open-loop transfer function for the ideal position based on P is: ; In the formula, K pP For position loop proportional gain; T s The velocity loop time constant; Equation (11) is a typical Type I system. According to control theory, the open-loop cutoff frequency ω is known. cP = K pP Furthermore, the intermediate frequency bandwidth h = τ of the velocity loop is defined by equation (11). i / T c And when ω c 2 = 1 / (τ i T c The velocity loop can achieve the maximum stability margin. By determining the position loop bandwidth, velocity loop bandwidth, and intermediate frequency bandwidth corresponding to different rigidity levels using the established rigidity level table, and based on the feedforward inertia calculated from dynamics in step S15, the following can be obtained: ; Then, based on the result of the tuning in equation (12), the speed command can be obtained: ; In the formula, V Ref For speed command value; e pos This is the positional deviation value; Then, by passing through the speed loop and combining the joint feedforward torque from step S15, the total torque command T can be obtained. Ref_Total : ; In the formula, T Ref_Total This is the total torque command value; e spd T represents the speed deviation value. FFW This is the joint feedforward torque.

[0014] The specific classification of step S2 is as follows: S21, Basic Mode: In this mode, the robot uses the factory-preset fixed servo control parameters, which can meet the basic motion control performance requirements. S22, Standard Control Mode: This mode is applicable to most typical working conditions and has wide applicability, meeting the needs of common industrial applications such as palletizing and handling. In the standard control mode, the servo parameter adaptive strategy in step S16 is followed, namely equations (16) and (17). S23, Low-speed precision mode: This mode is suitable for situations where the robot operates at low speed and there are certain requirements for trajectory accuracy, such as arc welding, low-speed glue application, and low-speed cutting. S24. High-speed positioning mode: This mode is suitable for situations where the robot operates at high speed and a certain positioning accuracy is required, such as spot welding, assembly, and loading / unloading.

[0015] In the low-speed precision mode of step S23, adjustments are made based on the servo parameter adaptive strategy in step S16. The specific adaptive servo parameter strategy is as follows: From equation (11), we can see that the closed-loop transfer function based on the ideal position loop of P is: ; In the formula, K pP For position loop proportional gain; T s The velocity loop time constant; Based on the comparison and analysis of the closed-loop transfer function of a typical Type I system with Equation (15), its damping ratio ξ can be obtained as: ; If the position loop is not expected to overshoot, it will be beneficial to the trajectory accuracy of the robot in low-speed precision mode; therefore, in low-speed precision mode, the position loop is designed to be in an overdamped state, that is, equation (16) is greater than 1, and thus we can obtain: ; Therefore, in low-speed precision mode, by additionally adjusting the rigidity level table, the speed loop bandwidth and position loop bandwidth can satisfy equation (17). Then, the parameters are self-tuned in the same form as Equation (12), and the loop commands are generated in the forms of Equations (13) and (14) to drive the robot body to move.

[0016] In the high-speed positioning mode of step S24, adjustments are made based on the servo parameter adaptive strategy in step S16. The specific adaptive servo parameter strategy is as follows: The intermediate frequency bandwidth is defined as h = τ given in step S16. i / T c According to control theory, a smaller h results in a faster response speed but more oscillations and a larger overshoot; a larger h results in a slower response but less oscillations and a smaller overshoot. Therefore, in high-speed positioning mode, by additionally adjusting the stiffness level table, while maintaining T... c While keeping them unchanged, appropriately reduce the intermediate frequency bandwidth h of the speed loop and the proportional gain K of the position loop. pP ; Then, the parameters are self-tuned in the same form as Equation (12), and the loop commands are generated in the forms of Equations (13) and (14) to drive the robot body to move.

[0017] A system based on a parameter set-based adaptive method for robot servo parameters includes a control system that incorporates the aforementioned method.

[0018] The beneficial effects of this invention are as follows: In the basic mode, fixed servo parameters are preset by the factory, while the servo parameters of the other three modes adopt an adaptive adjustment method that combines dynamic parameters and servo loop. The corresponding mode is determined according to the robot's working condition to adapt the servo parameters. This method has the advantages of being simple to implement, highly adaptable, and easy to use, thereby solving the problem of overall machine performance degradation caused by poor consistency of servo parameters. Attached Figure Description

[0019] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0020] Figure 1 This is a schematic diagram of the process of the present invention; Figure 2 This is the velocity loop Bode plot for the low-speed precision mode of this invention; Figure 3 This is the position loop Bode diagram in the low-speed accuracy mode of the present invention; Figure 4 This is a velocity loop Bode plot for the high-speed positioning mode of the present invention; Figure 5 This is a Bode diagram of the position loop in the high-speed positioning mode of the present invention. Detailed Implementation

[0021] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below.

[0022] Example 1: like Figures 1 to 5 As shown, a robot servo parameter adaptive method based on a parameter set has the following specific steps: S1. For the basic mode, standard control mode, low-speed precision mode and high-speed positioning mode in the parameter set concept, in order to improve the adaptability of the robot under different loads, it is necessary to obtain the robot dynamic model based on the current real load in advance, and calculate the dynamic parameters based on this dynamic model. S2. Switch the corresponding servo parameter adaptive strategy according to the selected mode.

[0023] The dynamic parameters in step S1 are feedforward inertia and feedforward torque. In this invention, the basic mode uses factory-preset fixed servo parameters, while the servo parameters of the other three modes adopt an adaptive adjustment method that combines dynamic parameters and servo loop. The appropriate mode is determined according to the robot's working condition to adapt the servo parameters. This method has the advantages of being simple to implement, highly adaptable, and easy to use, thereby solving the problem of overall machine performance degradation caused by poor consistency of servo parameters.

[0024] Step S1 specifically includes the following steps: S11. Based on the 3D CAD model of the robot body, obtain the theoretical dynamic parameters of each link; S12. Drive the robot to execute the preset trajectory and simultaneously collect joint position, velocity, acceleration and torque data; use this data to calibrate the aforementioned theoretical dynamic parameters through robot dynamic parameter identification in order to obtain parameter values ​​that are more consistent with the actual physical system. S13. Based on the actual dynamic parameters obtained in step S12, the robot is instructed to execute a specific trajectory under the current load, and the mass, center of mass position and inertia tensor of the current real load are obtained through the load identification module, i.e., the load dynamic parameters. S14. Integrate the theoretical dynamic parameters obtained in step S11, the actual dynamic parameters identified in step S12, and the load dynamic parameters obtained in step S13, and import them into the robot dynamic model. In addition, combine the theoretical parameters in step S11 with the actual parameters in step S12 and the integrated dynamic model established in step S14 to obtain the robot's feedforward inertia and feedforward torque under the current real load conditions online. S15. Based on the joint equivalent inertia obtained in real time in S14, the servo parameter tuning module dynamically adjusts the control parameters of the position loop and the speed loop. The servo control algorithm generates a basic torque command based on this, which is superimposed with the feedforward torque provided in S14 in the servo control loop to finally synthesize the total torque command used to drive the motor. S16. To construct a parameter adaptive mechanism, the robot is driven to move by the total torque command synthesized in step S15, and the joint state during operation is fed back to the dynamic model in real time. This process forms a closed-loop adjustment loop starting from step S14, thereby dynamically updating the servo loop parameters in real time.

[0025] In step S11, the theoretical parameters include mass, center of mass, inertia, and tensor types.

[0026] In step S12, the robot's dynamic parameter identification module consists of two parts: robot dynamics calculation and robot dynamic parameter identification.

[0027] The robot dynamics calculations are performed using the following steps: Step 1: Calculate the driving torque of each axis of the robot. For a typical serial robot, its dynamic equations can be expressed in the following matrix form: ; In the formula, M is the inertia matrix; N is collectively referred to as the nonlinear force matrix of the system, which includes Coriolis force, centripetal force, gravity and friction. Step 2: By linearizing equation (1), we can obtain: ; In the formula, Y is the motion matrix; P is the parameter matrix; In the above formula, P is a parameter matrix containing the connecting rod mass m, the moment of mass s, the inertia tensor I, and the friction force f, which can be expressed as follows: ; Step 3: By eliminating By identifying the linearly dependent terms in P and removing their corresponding elements, we can deduce... A set of maximally linearly independent sets is obtained, and the corresponding minimum dynamic parameter matrix is ​​obtained: ; In the formula, Y^ is the motion matrix to be identified; P^ is the parameter matrix to be identified.

[0028] The steps for identifying the robot's dynamic parameters are as follows: Step 1: Run the robot and collect actual motion information and torque data for each axis. Based on the data, construct the following expression: ; In the formula, τ i (i=1...k) is the robot driving torque matrix; Step 2: Solve equation (5) using the least squares method to obtain the optimal estimate of the robot's dynamic parameters P^, and then complete the entire identification process.

[0029] In step S13, the load parameter identification module consists of two parts: load identification model establishment and robot load parameter identification.

[0030] The steps for the load identification model are as follows: Step 1: The dynamic characteristics of each axis of the robot exhibit a linear superposition relationship. Based on this principle, the dynamic parameters of the load can be identified as follows: First, obtain the reference torque of the body joints when there is no load. Then, consider the additional torque generated by the load as a linear superposition component, and then solve for the load parameters. The specific process is as follows: ; In the formula, τ Total τ is the torque required for the robot to operate under the current load; Base The basic torque required for the robot body to operate (unloaded); Δτ L This refers to the incremental torque generated under the current load. Step 2: The transpose of the Jacobian matrix provides a linear mapping relationship that can map the generalized force vector acting on the robot end effector to the generalized force vector corresponding to each joint; this mapping maintains the linearity while ensuring the applicability of the torque superposition principle.

[0031] Based on this relationship, the transpose of the Jacobian matrix can be used to analyze and decouple the torque contribution of the end load at each joint. The corresponding load identification model can be expressed as: ; In the formula, J is the Jacobian matrix; Y L P is the observation matrix to be identified; L The load parameter matrix to be identified.

[0032] The steps for identifying robot load parameters are as follows: Using the same formula (5), the robot is run and the actual motion information and torque data of each axis are collected to construct the following expression: ; In the formula, τ L_i (i=1...k) is the load torque matrix; Similarly, the least squares method is used to perform P on equation (8). L The optimal estimate is then obtained, thus completing the entire identification process.

[0033] The specific servo parameter tuning strategy for step S15 is as follows: The ideal speed open-loop transfer function based on a series PI converter is: ; In the formula, K pS τ is the proportional gain of the speed loop PI controller. i The integral time constant of the speed loop PI controller is J; the joint inertia is T. c The current loop time constant; To eliminate the dependence of the speed loop control performance on the inertia of the controlled object, the proportional gain in equation (9) is set to K. pS = K speed *J, then: ; Obviously, equation (10) is a typical type II system, and the open-loop cutoff frequency ω can be approximated as the velocity. cS = K speed ; The open-loop transfer function for the ideal position based on P is: ; In the formula, K pP For position loop proportional gain; T s The velocity loop time constant; Equation (11) is a typical Type I system. According to control theory, the open-loop cutoff frequency ω is known. cP = K pP Furthermore, the intermediate frequency bandwidth h = τ of the velocity loop is defined by equation (11). i / T c And when ω c 2 = 1 / (τ i T c The velocity loop can obtain the maximum stability margin, and this invention will use this as the adaptive principle for the loop parameters; In summary, by determining the position loop bandwidth, velocity loop bandwidth, and intermediate frequency bandwidth corresponding to different rigidity levels using the established rigidity level table, and based on the feedforward inertia calculated from dynamics using S15, the following can be obtained: ; Then, based on the result of the tuning in equation (12), the speed command can be obtained: ; In the formula, V Ref For speed command value; e pos This is the positional deviation value; Then, by combining the speed loop with the joint feedforward torque of S15, the total torque command T can be obtained. Ref_Total : ; In the formula, T Ref_Total This is the total torque command value; e spd T represents the speed deviation value. FFW This is the joint feedforward torque.

[0034] The specific classification of step S2 is as follows: S21, Basic Mode: In this mode, the robot uses the factory-preset fixed servo control parameters, which can meet the basic motion control performance requirements. S22, Standard Control Mode: This mode is applicable to most typical working conditions and has wide applicability, meeting the needs of common industrial applications such as palletizing and handling. In the standard control mode, the servo parameter adaptive strategy in step S16 is followed, namely equations (16) and (17). S23, Low-speed precision mode: This mode is suitable for situations where the robot operates at low speed and there are certain requirements for trajectory accuracy, such as arc welding, low-speed glue application, and low-speed cutting. S24. High-speed positioning mode: This mode is suitable for situations where the robot operates at high speed and a certain positioning accuracy is required, such as spot welding, assembly, and loading / unloading.

[0035] In the low-speed precision mode of step S23, adjustments are made based on the servo parameter adaptive strategy in step S16. The specific adaptive servo parameter strategy is as follows: From equation (11), we can see that the closed-loop transfer function based on the ideal position loop of P is: ; In the formula, K pP For position loop proportional gain; T s The velocity loop time constant; Based on the comparison and analysis of the closed-loop transfer function of a typical Type I system with Equation (15), its damping ratio ξ can be obtained as: ; If the position loop is not expected to overshoot, it will be beneficial to the trajectory accuracy of the robot in low-speed precision mode; therefore, in low-speed precision mode, the position loop is designed to be in an overdamped state, that is, equation (16) is greater than 1, and thus we can obtain: ; Therefore, in low-speed precision mode, by additionally adjusting the rigidity level table, the speed loop bandwidth and position loop bandwidth can satisfy equation (17). Then, the parameters are self-tuned in the same form as Equation (12), and the loop commands are generated in the forms of Equations (13) and (14) to drive the robot body to move. Finally, to further improve the ease of use of the present invention, a "mode depth" interface is provided on the robot teach pendant, with a range of 0.0 to 1.0. This parameter takes effect immediately, and the larger the value set, the more the parameter adapts to the set mode, and the better the corresponding performance indicators are improved. In low-speed precision mode, adjusting the "mode depth" increases the position loop stiffness level, while the velocity loop stiffness must still satisfy equation (18); during the adjustment process, the corresponding velocity loop and position loop Bode diagrams are as follows: Figure 2 and Figure 3 As shown; based on this adjustment strategy, the stability margin of the speed loop varies from 64.1 deg to 66.2 deg, and the stability margin of the position loop varies from 84.6 to 84.8 deg, which can meet the stability requirements of the servo loop.

[0036] In the high-speed positioning mode of step S24, adjustments are made based on the servo parameter adaptive strategy in step S16. The specific adaptive servo parameter strategy is as follows: The intermediate frequency bandwidth h = τ is defined in step S16. i / T c According to control theory, a smaller h results in a faster response speed but more oscillations and a larger overshoot; a larger h results in a slower response but less oscillation and a smaller overshoot. Furthermore, the high-speed positioning mode designed in this invention is mainly applied to spot welding, assembly, and loading / unloading operations, which not only require a certain level of positioning accuracy but also a certain operating cycle time. Therefore, this invention reduces the position loop proportional gain K in this mode. pP This is to increase the damping of the position loop, aiming to improve positioning accuracy, reduce positioning overshoot, and at the same time ensure sufficient speed loop bandwidth requirements. Therefore, in high-speed positioning mode, by additionally adjusting the rigidity level table, while maintaining T... c While keeping them unchanged, appropriately reduce the intermediate frequency bandwidth h of the speed loop and the proportional gain K of the position loop. pP ; Then, the parameters are self-tuned in the same form as Equation (12), and the loop commands are generated in the forms of Equations (13) and (14) to drive the robot body to move.

[0037] Finally, to further improve the ease of use of the present invention, a "mode depth" interface is provided on the robot teach pendant, with a range of 0.0 to 1.0. This parameter takes effect immediately, and the larger the value set, the more the parameter adapts to the set mode, and the better the corresponding performance indicators are improved. In high-speed positioning mode, adjusting the "mode depth" is equivalent to adjusting the decrease in the velocity loop intermediate frequency bandwidth h and the position loop proportional gain KpP. During the adjustment process, the corresponding velocity loop and position loop Bode plots are as follows: Figure 4 and Figure 5 As shown; based on this adjustment strategy, the stability margin of the speed loop varies from 64.2 deg to 69.3 deg, and the stability margin of the position loop varies from 84.8 to 87.1 deg, which can meet the stability requirements of the servo loop.

[0038] Example 2: A system based on a parameter set-based adaptive method for robot servo parameters includes a control system that incorporates the aforementioned method.

[0039] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely prisms of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A robot servo parameter self-adaptive method based on a parameter set, characterized in that: The specific steps are as follows: S1, for the basic mode, standard control mode, low speed precision mode and high speed positioning mode in the parameter set concept, the robot whole dynamics model under the current real load needs to be obtained in advance, and the dynamics parameters are calculated based on the dynamics model; S2, according to the selected mode, the corresponding servo parameter adaptive strategy switching is carried out.

2. The robot servo parameter self-adaptive method based on parameter set according to claim 1, characterized in that: The step S1 specifically includes the following steps: S11, based on the three-dimensional CAD model of the robot body, the theoretical dynamics parameters of each connecting rod are obtained; S12, the robot is driven to execute the preset trajectory and synchronously collect the joint position, speed, acceleration and torque data, and the theoretical dynamics parameters are calibrated through the robot dynamics parameter identification; S13, based on the actual dynamics parameters obtained in step S12, the robot executes a specific trajectory under the current load, and the mass, center of mass position and inertia tensor of the current real load, i.e. the load dynamics parameters, are obtained through the load identification module; S14, the theoretical dynamics parameters obtained in step S11, the actual dynamics parameters identified in step S12 and the load dynamics parameters obtained in step S13 are integrated and uniformly imported into the robot dynamics model, and the feedforward inertia and feedforward torque of the robot under the current real load working condition are obtained online; S15, based on the joint equivalent inertia obtained in S14, the servo parameter setting module dynamically adjusts the control parameters of the position loop and the speed loop, and the servo control algorithm generates a basic torque instruction according to the same, which is superimposed with the feedforward torque provided in step S14 in the servo control loop, and finally the total torque instruction for driving the motor is synthesized; S16, in order to construct the parameter adaptive mechanism, the total torque instruction synthesized in step S15 is used to drive the robot, and the joint state in the running process is fed back to the dynamics model in real time.

3. The robot servo parameter self-adaptive method based on parameter set according to claim 2, characterized in that: In the step S11, the theoretical parameters include mass, center of mass, inertia, tensor and the like.

4. The robot servo parameter self-adaptive method based on parameter set according to claim 2, characterized in that: In the step S12, the dynamics parameter identification module of the robot is composed of robot dynamics calculation and robot dynamics parameter identification; The robot dynamics calculation has the following steps: Step one: the driving torque of each axis of the robot is obtained, and for a typical serial robot, the dynamics equation can be expressed in the following matrix form: ； In the formula, M is the inertia matrix; N is collectively referred to as the system nonlinear force matrix, including Coriolis force, centripetal force, gravity and friction force; Step two: linearization approximation is carried out from formula (1) to obtain: ； In the formula, Y is the motion quantity matrix; P is the parameter matrix; In the above formula, P is the parameter matrix including connecting rod mass m, center of mass s, inertia tensor I and friction force f, which can be expressed as follows: ； Step three: by eliminating the linearly dependent terms in P and removing the corresponding elements in P, we can derive a set of maximal linearly independent groups of and obtain the corresponding minimal kinetic parameter matrix: ； In the formula, Y^ is the motion quantity matrix to be identified; P^ is the parameter matrix to be identified; The robot dynamics parameter identification has the following steps: Step one: the robot is run and the actual motion information and torque data of each axis are collected, and the following expression is constructed according to the data: ； where τ i (i = 1...k) is the robot drive torque matrix; Step two: the least square method is used to solve formula (5) to obtain the optimal estimated value P^ of the robot dynamics parameters, and the whole identification process is completed.

5. The robot servo parameter self-adaptive method based on parameter set according to claim 2, characterized in that: The load parameter identification module is composed of a load identification model and a robot load parameter identification; The load identification model has the following steps: Step one: first, obtain the body joint torque reference without load, then consider the additional torque generated by the load as a linear superposition component, and then solve the load parameter, the specific process is as follows: ； where τ Total is the torque required to run the robot based on the current load; τ Base is the base torque required to run the robot body (unladen); Δτ L is the incremental torque generated under the current load; Step two: the transpose of the Jacobian matrix provides a linear mapping relationship; based on this relationship, the torque contribution of the end load at each joint can be analyzed and decoupled using the transpose of the Jacobian matrix, and the corresponding load identification model can be expressed as: ； In the formula, J is a Jacobian matrix; Y L is an observation matrix to be identified; P L is a load parameter matrix to be identified; The robot load parameter identification has the following steps: The same as formula (5), run the robot and collect the actual motion information and torque data of each axis, and construct the following expression: ； where τ L_i (i = 1...k) is the load torque matrix; Similarly, the P L The optimal estimation is solved, and the whole identification process is completed.

6. The robot servo parameter self-adaptive method based on parameter set according to claim 2, characterized in that: The specific servo parameter setting strategy of step S15 is as follows: For the ideal speed open-loop transfer function based on series PI: ； where K pS is the proportional gain of the velocity loop PI regulator; τ i is the integral time constant of the velocity loop PI regulator; J is the joint inertia; T c is the current loop time constant; To eliminate the dependence of the speed loop control performance on the inertia of the controlled object, set the proportional gain in equation (9) to K pS = K speed *J, then we have: ； It is clear that equation (10) is a typical type II system, and the speed open-loop cutoff frequency ω cS = K speed ; For the ideal position open-loop transfer function based on P: ； In the formula, K pP is a position loop proportional gain; T s is a velocity loop time constant; Equation (11) is a typical type I system, according to control theory, the position open-loop cutoff frequency ω cP = K pP ; in addition, the frequency bandwidth h = τ i / T c in the speed loop is defined by equation (11), and the speed loop can obtain the maximum stability margin when ω c 2 = 1 / (τ i T c ) Through the formulated rigidity level table, the corresponding position loop bandwidth, speed loop bandwidth and medium frequency bandwidth under different rigidity levels are determined, and according to the feedforward inertia calculated by the dynamics in step S15, the following can be obtained: ； According to the results adjusted by formula (12), the speed command can be obtained: ； In the formula, V Ref is a velocity command value; e pos is a position error value; Then, via the speed loop and in combination with the joint feedforward torque of step S15, the total torque command T Ref_Total : ； In the formula, T Ref_Total is a total torque command value; e spd is a speed deviation value; T FFW is a joint feedforward torque.

7. The robot servo parameter self-adaptive method based on parameter set according to claim 1, characterized in that: The specific classification of step S2 is as follows: S21, basic mode: in this mode, the fixed servo control parameters preset by the robot factory are used, which can meet the basic motion control performance requirements; S22, standard control mode: this mode is suitable for most typical working conditions and has wide applicability, which can meet the requirements of common industrial applications such as stacking and carrying; In the standard control mode, the servo parameter adaptive strategy in step S16 is followed, that is, formula (16) and formula (17); S23, low-speed precision mode: this mode is suitable for situations where the robot runs at low speed and has certain requirements for trajectory precision, such as arc welding, low-speed gluing and low-speed cutting; S24, high-speed positioning mode: this mode is suitable for situations where the robot runs at high speed and has certain requirements for positioning precision, such as spot welding, assembly and feeding.

8. The robot servo parameter self-adaptive method based on parameter set according to claim 7, characterized in that: In the low-speed precision mode of step S23, the servo parameter adaptive strategy in step S16 is adjusted; The specific servo parameter adaptive strategy is as follows: According to formula (11), the ideal position loop closed-loop transfer function based on P is: ； In the formula, K pP is a position loop proportional gain; T s is a velocity loop time constant; According to the comparison and analysis of the closed-loop transfer function of the typical type I system and formula (15), the damping ratio ξ can be obtained: ； In order to avoid overshoot of the expected position loop, it is beneficial to improve the trajectory precision of the robot in the low-speed precision mode; therefore, the position loop is designed to be over-damped in the low-speed precision mode, that is, formula (16) is greater than 1, and the following can be obtained: ； Therefore, in the low-speed precision mode, the rigidity level table is adjusted to make the speed loop bandwidth and the position loop bandwidth satisfy formula (17); Then, the parameters are self-adjusted in the form of formula (12), the loop command is generated in the form of formula (13) and formula (14), and the robot body is driven to move.

9. The robot servo parameter self-adaptive method based on parameter set according to claim 7, characterized in that: In the high-speed positioning mode of step S24, the servo parameter adaptive strategy in step S16 is adjusted; The specific servo parameter adaptive strategy is as follows: The intermediate frequency bandwidth h = τ defined in step S16 i / T c According to control theory, the smaller h is, the faster the response speed is, but the more oscillation and the larger the overshoot are; the larger h is, the slower the response is, but the less oscillation and the smaller the overshoot are; therefore, in the high-speed positioning mode, the intermediate frequency bandwidth h and the proportional gain K c of the position loop are appropriately reduced by additionally adjusting the rigidity level table under the premise of keeping T pP unchanged. Then, the parameter self-tuning is performed in the form of formula (12), and the loop instruction generation is performed in the form of formula (13) and formula (14), so as to drive the robot body movement.

10. A system for using a robot servo parameter self-adaptive method based on a parameter set according to any one of claims 1 to 9, characterized in that: The control system comprising the above method is built-in.

Images