Kinetic parameter identification method and device, equipment, storage medium and program product

Abstract

The invention provides a kinetic parameter identification method and device, equipment, a storage medium and a program product, and relates to the technical field of motor control, and the method comprises the steps: constructing an adjustable model; the adjustable model is a mathematical model constructed on the basis of to-be-identified parameters of the direct-drive rotary table, and the to-be-identified parameters are determined on the basis of kinetic parameters of the direct-drive rotary table; constructing a target function based on the adjustable model; on the basis of a gradient optimization method, updating the to-be-identified parameters in the adjustable model until the target function reaches the minimum value, and obtaining the converged to-be-identified parameters; and on the basis of the converged to-be-identified parameters, identifying the kinetic parameters of the direct-driven rotary table. By means of the method, the identification precision and the identification speed can be effectively balanced in the identification process, robustness is high, and real-time identification of the kinetic parameters is better facilitated.

Description

Technical Field

[0001] The present invention relates to the field of motor control technology, and in particular to a dynamic parameter identification method, device, equipment, storage medium and program product. Background Art

[0002] A direct-drive turntable is a complex optomechanical and electrical device with applications in precision machining, precision testing, inertial navigation simulation, and other fields. It consists of an internal frameless permanent magnet synchronous rotating motor and an external worktable assembly. Its relevant dynamic parameters, such as the turntable's moment of inertia and viscous friction damping coefficient, are unknown. These dynamic parameters are key to the mechanical response of the direct-drive turntable and are crucial for the design of the velocity loop controller in motion control systems. Therefore, identifying the relevant dynamic parameters of the direct-drive turntable is essential when designing a motion control system.

[0003] Existing dynamic parameter identification methods are mainly divided into offline identification methods and online identification methods: the offline identification method refers to a one-time estimation of system parameters or models after collecting sufficient data when the system stops running or under experimental conditions. This method does not consider the real-time changes of the system during motion; the online identification method refers to the real-time estimation and update of system parameters or models during the operation of the system. This method can capture changes in system parameters in a timely manner and adapt to changes in the dynamic characteristics of the system.

[0004] For precision direct-drive turntables, online identification methods are more suitable. Existing online identification methods mainly include least squares method, model reference adaptive method, Kalman filter method, etc.

[0005] However, these methods usually require fixed or linear adaptive coefficients for identification, and require the use of large amounts of data for complex calculations during the calculation process. The algorithm is highly complex, resulting in the inability to effectively balance identification accuracy and identification speed during the identification process, and poor robustness. Summary of the Invention

[0006] The present invention provides a kinetic parameter identification method, apparatus, device, storage medium and program product to solve the defects of the existing kinetic parameter identification method, such as high algorithm complexity, which results in the inability to effectively balance the identification accuracy and identification speed during the identification process and poor robustness.

[0007] The present invention provides a method for identifying dynamic parameters, comprising: constructing an adjustable model; the adjustable model is a mathematical model constructed based on the parameters to be identified of a direct-drive turntable, and the parameters to be identified are determined based on the dynamic parameters of the direct-drive turntable; based on the adjustable model, an objective function is constructed; based on a gradient optimization method, the parameters to be identified in the adjustable model are updated until the objective function reaches a minimum value, thereby obtaining the converged parameters to be identified; based on the converged parameters to be identified, the dynamic parameters of the direct-drive turntable are identified.

[0008] According to a dynamic parameter identification method provided by the present invention, the dynamic parameters include turntable moment of inertia, viscous friction damping coefficient and disturbance torque, and the parameters to be identified include a first parameter to be identified, a second parameter to be identified and a third parameter to be identified; wherein, the first parameter to be identified is determined based on the signal sampling period and the turntable moment of inertia; the second parameter to be identified is determined based on the signal sampling period, the turntable moment of inertia and the disturbance torque; and the third parameter to be identified is determined based on the signal sampling period, the viscous friction damping coefficient and the turntable moment of inertia.

[0009] According to a dynamic parameter identification method provided by the present invention, based on the gradient optimization method, the parameters to be identified in the adjustable model are updated until the objective function reaches a minimum value, and the parameters to be identified after convergence are obtained, including: determining a first gradient, a second gradient, and a third gradient; the first gradient is the gradient of the objective function with respect to the first parameter to be identified, the second gradient is the gradient of the objective function with respect to the second parameter to be identified, and the third gradient is the gradient of the objective function with respect to the third parameter to be identified; determining a first parameter update amount of the first parameter to be identified, a second parameter update amount of the second parameter to be identified, and a third parameter update amount of the third parameter to be identified; determining a learning rate positive definite matrix of the gradient optimization method based on the first gradient, the second gradient, the third gradient, the first parameter update amount, the second parameter update amount, and the third parameter update amount; based on the learning rate positive definite matrix, updating the parameters to be identified in the adjustable model according to the gradient direction until the objective function reaches a minimum value, and the parameters to be identified after convergence are obtained.

[0010] According to a dynamic parameter identification method provided by the present invention, a learning rate positive definite matrix of a gradient optimization method is determined based on a first gradient, a second gradient, a third gradient, a first parameter update amount, a second parameter update amount, and a third parameter update amount. The method includes: determining a first learning rate of a first parameter to be identified based on the first gradient and the first parameter update amount; determining a second learning rate of a second parameter to be identified based on the second gradient and the second parameter update amount; determining a third learning rate of a third parameter to be identified based on the third gradient and the third parameter update amount; and constructing a learning rate positive definite matrix of the gradient optimization method based on the first learning rate, the second learning rate, and the third learning rate.

[0011] According to a dynamic parameter identification method provided by the present invention, a first learning rate is determined based on the exponentially decaying average of the square of the first gradient and the exponentially decaying average of the square of the first parameter update amount; the second learning rate is determined based on the exponentially decaying average of the square of the second gradient and the exponentially decaying average of the square of the second parameter update amount; and the third learning rate is determined based on the exponentially decaying average of the square of the third gradient and the exponentially decaying average of the square of the third parameter update amount.

[0012] According to a dynamic parameter identification method provided by the present invention, an adjustable model is constructed, including: establishing a dynamic equation of a direct-drive turntable; the dynamic equation of the direct-drive turntable is a mathematical model determined based on the dynamic parameters of the direct-drive turntable; the dynamic equation of the direct-drive turntable is discretized to obtain a discrete dynamic equation; and based on the discrete dynamic equation, an adjustable model is constructed.

[0013] The present invention also provides a dynamic parameter identification device, including: a construction module for constructing an adjustable model; the adjustable model is a mathematical model constructed based on the parameters to be identified of the direct-drive turntable, and the parameters to be identified are determined based on the dynamic parameters of the direct-drive turntable; an objective function determination module for constructing an objective function based on the adjustable model; a first identification module for updating the parameters to be identified in the adjustable model based on the gradient optimization method until the objective function reaches a minimum value, thereby obtaining the converged parameters to be identified; a second identification module for identifying the dynamic parameters of the direct-drive turntable based on the converged parameters to be identified.

[0014] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, any one of the above-mentioned dynamic parameter identification methods is implemented.

[0015] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which implements any of the above-mentioned kinetic parameter identification methods when executed by a processor.

[0016] The present invention also provides a computer program product, comprising a computer program, which implements any of the above-mentioned kinetic parameter identification methods when executed by a processor.

[0017] The dynamic parameter identification method, device, equipment, storage medium and program product provided by the present invention first construct an adjustable model based on the dynamic parameters of the direct-drive turntable, then construct an objective function based on the adjustable model, and finally use the gradient optimization method to update the parameters to be identified in the adjustable model until the objective function reaches a minimum value, thereby obtaining the converged parameters to be identified, and then identifying the dynamic parameters of the direct-drive turntable based on the converged parameters to be identified. In the above manner, since the gradient optimization method has the advantages of simple structure and small amount of calculation, it can significantly reduce the complexity of the algorithm while ensuring the identification accuracy, thereby improving the identification speed, so that the identification accuracy and identification speed can be effectively balanced during the identification process, with high robustness, which is more conducive to the real-time identification of dynamic parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the present invention or the prior art, a brief introduction is given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 This is one of the flow charts of the kinetic parameter identification method provided by the present invention.

[0020] Figure 2 This is the second flow chart of the kinetic parameter identification method provided by the present invention.

[0021] Figure 3 It is a structural schematic diagram of the direct-drive turntable system provided by the present invention.

[0022] Figure 4 This is the third flow chart of the kinetic parameter identification method provided by the present invention.

[0023] Figure 5 This is one of the comparison diagrams of the experimental results between the kinetic parameter identification method provided by the present invention and the traditional algorithm.

[0024] Figure 6 This is the second comparison chart of the experimental results between the kinetic parameter identification method provided by the present invention and the traditional algorithm.

[0025] Figure 7 This is the third comparison chart of experimental results between the kinetic parameter identification method provided by the present invention and the traditional algorithm.

[0026] Figure 8 It is a structural schematic diagram of the dynamic parameter identification device provided by the present invention.

[0027] Figure 9It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0028] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0029] See also Figure 1 , Figure 1 This is one of the flow charts of the kinetic parameter identification method provided by the present invention. In this embodiment, the kinetic parameter identification method includes steps S110 to S140, each of which is as follows: S110: Build an adjustable model.

[0030] The adjustable model is a mathematical model constructed based on the parameters to be identified of the direct-drive turntable, and the parameters to be identified are determined based on the dynamic parameters of the direct-drive turntable.

[0031] Specifically, a dynamic equation of a direct-drive turntable is first established. The dynamic equation of a direct-drive turntable is a mathematical model determined based on the dynamic parameters of the direct-drive turntable and can be used as a reference model in subsequent steps.

[0032] The dynamic equation of the direct-drive turntable is expressed as follows: ; in, is the angular velocity of the direct-drive turntable; is the moment of inertia of the turntable; is the viscous friction damping coefficient; is the disturbance torque, which is the sum of the disturbance torque caused by load torque, Coulomb friction and other factors; is the electromagnetic torque output by the motor.

[0033] Furthermore, the dynamic equation of the direct-drive turntable is discretized to obtain a discrete dynamic equation.

[0034] The expression of the discrete dynamics equation is as follows: ; in, is the signal sampling period; Indicates the time, Indicates the time; Indicates the direct drive rotary table Angular velocity of motion at any moment; Indicates the direct drive rotary table The angular velocity of motion at any moment.

[0035] Furthermore, based on the discrete dynamic equations, an adjustable model of the adaptive system is constructed.

[0036] Among them, the expression of the adjustable model of the adaptive system is as follows: ; ; ; ; in, represents the estimated value of the turntable speed by the adjustable model; 、 、 These are the three parameters to be identified of the direct-drive turntable.

[0037] From the above formula, we can see that the first parameter to be identified is Based on the signal sampling period and turntable moment of inertia Determined, the second parameter to be identified Based on the signal sampling period , turntable moment of inertia and disturbance torque Determined, the third parameter to be identified Based on the signal sampling period , viscous friction damping coefficient and turntable moment of inertia Sure.

[0038] For direct drive turntable, due to the signal sampling period The value of is usually known, so only the 、 、 The identification of these three parameters can be based on 、 、 The value of the turntable moment of inertia of the direct drive turntable is calculated , viscous friction damping coefficient and disturbance torque , thereby completing the turntable moment of inertia of the direct drive turntable , viscous friction damping coefficient and disturbance torque Identification of these three kinetic parameters.

[0039] It should be noted that the existing technology does not include the damping term (i.e., the viscous friction damping coefficient) ) is added to the adjustable model of the adaptive system. Since the damping term proportional to the motor speed is not added, the damping term proportional to the motor speed is mistakenly added to other inertia identification terms during the identification process, resulting in fluctuations and errors in the final identification results, resulting in poor overall identification results. Based on this, this embodiment introduces the damping term (i.e., the viscous friction damping coefficient) into the adjustable model. ) can not only reduce the fluctuation and error of the identification results of other dynamic parameters, but also enable more accurate modeling of the direct-drive turntable.

[0040] S120: Construct an objective function based on the adjustable model.

[0041] Specifically, after constructing the adjustable model of the adaptive system, it is necessary to determine the parameter matrix to be identified and the input matrix .

[0042] Among them, the parameter matrix to be identified is and the input matrix The expressions are as follows: ; ; Among them, the "The superscript values ​​are estimates. The first parameter to be identified In the The estimated value of the time, Represents the second parameter to be identified In the The estimated value of the time, The third parameter to be identified In the Estimated value of the moment; Indicates that the motor is The electromagnetic torque output at any moment.

[0043] Furthermore, according to the above formula, the prior adjustable model corresponding to the adjustable model of the adaptive system is constructed and posterior adjustable models .

[0044] Among them, the prior adjustable model and posterior adjustable models The expressions are as follows: ; ; in, Indicates that the adjustable model is The prior output at the moment, i.e. The predicted value for the next moment after the moment parameter update is completed; Indicates that the adjustable model is The posterior output at the moment evaluates the immediate impact of parameter updates on the system.

[0045] Correspondingly, in The a priori error between the output of the reference model and the output of the adjustable model at time and posterior error The expressions are as follows: ; .

[0046] Furthermore, the following update formula for the parameters to be identified is defined: ; in, Indicates the The parameter matrix to be identified at time The update should be affected by the last moment value (i.e. The parameter matrix to be identified at time ), input value (i.e. Input value at time ), the current error (i.e. The posterior error at time ) and a series of factors.

[0047] On the basis of the above definition, an adaptive law algorithm is designed according to the gradient optimization method, that is, the objective function of the gradient optimization method iteration is defined according to the adjustable model of the adaptive system.

[0048] The objective function is expressed as follows: ; Among them, the objective function The meaning is: by adjusting the parameters to be identified, the objective function value is minimized.

[0049] S130: Based on the gradient optimization method, the parameters to be identified in the adjustable model are updated until the objective function reaches a minimum value, thereby obtaining the converged parameters to be identified.

[0050] The parameters to be identified after convergence include the first parameters to be identified after convergence , the second parameter to be identified after convergence and the third parameter to be identified after convergence .

[0051] S140: Based on the converged parameters to be identified, the dynamic parameters of the direct-drive turntable are identified.

[0052] Specifically, according to the first parameter to be identified after convergence , the second parameter to be identified after convergence and the third parameter to be identified after convergence Calculate the moment of inertia of the direct drive rotary table , viscous friction damping coefficient and disturbance torque , thereby completing the turntable moment of inertia of the direct drive turntable , viscous friction damping coefficient and disturbance torque Identification of these three kinetic parameters.

[0053] The dynamic parameter identification method provided in this embodiment first constructs an adjustable model based on the dynamic parameters of the direct-drive turntable, then constructs an objective function based on the adjustable model, and finally uses the gradient optimization method to update the parameters to be identified in the adjustable model until the objective function reaches a minimum value, thereby obtaining the converged parameters to be identified, and then identifying the dynamic parameters of the direct-drive turntable based on the converged parameters to be identified. In this way, due to the advantages of the gradient optimization method of simple structure and small computational complexity, it can significantly reduce the complexity of the algorithm while ensuring the identification accuracy, thereby improving the identification speed, thereby effectively balancing the identification accuracy and identification speed during the identification process, with high robustness, and more conducive to the real-time identification of dynamic parameters.

[0054] In some embodiments, the dynamic parameters include the turntable moment of inertia , viscous friction damping coefficient and disturbance torque , the parameters to be identified include the first parameter to be identified , the second parameter to be identified and the third parameter to be identified .

[0055] Among them, the first parameter to be identified Based on the signal sampling period and turntable moment of inertia Determined, its expression is as follows: .

[0056] The second parameter to be identified Based on the signal sampling period , turntable moment of inertia and disturbance torque Determined, its expression is as follows: .

[0057] The third parameter to be identified Based on the signal sampling period , viscous friction damping coefficient and turntable moment of inertia Determined, its expression is as follows: .

[0058] In some embodiments, based on the gradient optimization method, the parameters to be identified in the adjustable model are updated until the objective function reaches a minimum value, and the converged parameters to be identified are obtained, including: determining a first gradient, a second gradient, and a third gradient; the first gradient is the gradient of the objective function with respect to the first parameter to be identified, the second gradient is the gradient of the objective function with respect to the second parameter to be identified, and the third gradient is the gradient of the objective function with respect to the third parameter to be identified; determining a first parameter update amount of the first parameter to be identified, a second parameter update amount of the second parameter to be identified, and a third parameter update amount of the third parameter to be identified; determining a learning rate positive definite matrix of the gradient optimization method based on the first gradient, the second gradient, the third gradient, the first parameter update amount, the second parameter update amount, and the third parameter update amount; based on the learning rate positive definite matrix, the parameters to be identified in the adjustable model are updated according to the gradient direction until the objective function reaches a minimum value, and the converged parameters to be identified are obtained.

[0059] Specifically, the process of updating the parameters to be identified in the adjustable model based on the gradient optimization method can be expressed by the following formula: ; ; in, is the learning rate positive definite matrix of the gradient optimization method. From the above formula, we can see that The first parameter to be identified is The first learning rate , the second parameter to be identified The second learning rate and the third parameter to be identified The third learning rate The matrix composed of .

[0060] Among them, the expression of the objective function gradient is as follows: ; Based on the above formula, the update formula of the parameters to be identified can be organized as: .

[0061] It should be noted that in the conventional gradient optimization method, the learning rate positive definite matrix is a preset fixed value (i.e. 、 、 (are all fixed values). However, in this embodiment, in order to achieve dynamic balance and debugging-free, it is necessary to replace the fixed learning rate with a learning rate that can be adaptively and dynamically adjusted.

[0062] Specifically, define the calculation formula of the learning rate, The learning rate at each moment The calculation formula is as follows: ; ; ; in, (Exponential Moving Average) means exponential decay average; Indicates the The parameter update amount of the parameters to be identified, For the Moment The exponentially decaying average of the squares of the parameter updates of the parameters to be identified; It represents the objective function for The gradient of the parameter to be identified, For the The objective function at the moment The exponentially decaying average of the squares of the gradients of the parameters to be identified; 、 Both are forgetting factors in exponential decay averaging, which makes the weight of early data in exponential decay averaging decrease exponentially; It is a numerical stability term, which is used to prevent the denominator from being zero. It is usually a very small constant (for example ).

[0063] By maintaining the exponentially decaying average of the square of the current gradient and the exponentially decaying average of the square of the parameter update, the learning rate can be adjusted dynamically in real time. If the current gradient changes slowly, a higher learning rate can be maintained to accelerate convergence. If the current gradient fluctuates frequently, the dynamic adjustment process reduces the learning rate to enhance stability. The exponentially decaying averaging mechanism smooths single-step noise and improves the interference resistance of the identification process. Furthermore, the real-time dynamic adjustment of the learning rate eliminates the need for users to pre-set a fixed learning rate to identify dynamic parameters, reducing debugging complexity.

[0064] Based on the above design of dynamic learning rate, in this embodiment, after constructing the objective function, it is necessary to calculate the first gradient , second gradient and the third gradient .

[0065] Among them, the first gradient The first parameter to be identified for the objective function The gradient, the second gradient The second parameter to be identified is the objective function Gradient, third gradient The third parameter to be identified is the objective function gradient.

[0066] Further, determine the first parameter to be identified The first parameter update amount , the second parameter to be identified The second parameter update amount and the third parameter to be identified The third parameter update amount . Furthermore, based on the first gradient , second gradient , the third level , the first parameter update amount , the second parameter update amount and the third parameter update amount , determine the learning rate positive definite matrix of the gradient optimization method , and based on the learning rate positive definite matrix , all the parameters to be identified in the adjustable model are updated according to the gradient direction. At this time, the model can start iterative calculation according to the gradient optimal adaptive law until the objective function gradually reaches the minimum value, obtain all the converged parameters to be identified, and then complete the identification of the dynamic parameters of the direct-drive turntable.

[0067] In some embodiments, based on the first gradient, the second gradient, the third gradient, the first parameter update amount, the second parameter update amount and the third parameter update amount, determining the learning rate positive definite matrix of the gradient optimization method includes: determining a first learning rate of the first parameter to be identified based on the first gradient and the first parameter update amount; determining a second learning rate of the second parameter to be identified based on the second gradient and the second parameter update amount; determining a third learning rate of the third parameter to be identified based on the third gradient and the third parameter update amount; and constructing a learning rate positive definite matrix of the gradient optimization method based on the first learning rate, the second learning rate and the third learning rate.

[0068] In some embodiments, the first learning rate is determined based on an exponentially decaying average of the square of the first gradient and an exponentially decaying average of the square of the first parameter update amount; the second learning rate is determined based on an exponentially decaying average of the square of the second gradient and an exponentially decaying average of the square of the second parameter update amount; and the third learning rate is determined based on an exponentially decaying average of the square of the third gradient and an exponentially decaying average of the square of the third parameter update amount.

[0069] Specifically, the first parameter to be identified is The first learning rate The expression is as follows: ; ; .

[0070] The second parameter to be identified The second learning rate The expression is as follows: ; ; .

[0071] The third parameter to be identified The third learning rate The expression is as follows: ; ; .

[0072] In existing identification methods, adaptive parameters usually use fixed values ​​or simple linear gains, which makes it difficult to balance identification accuracy and identification speed. The dynamic parameter identification method provided in this embodiment can dynamically adjust the learning rate according to data changes in real time through a dynamic adaptive algorithm. If the gradient changes slowly, a high learning rate can be maintained to accelerate convergence. If the gradient fluctuates frequently, the dynamic adjustment process will reduce the learning rate to enhance stability, achieve real-time feedback and update of data, and at the same time enhance identification accuracy and improve identification speed. In addition, this method can complete the identification work by simply configuring general parameters, without the need for repeated manual debugging of the learning rate parameters. The debugging time is greatly reduced, and the adaptability of the algorithm engineering is improved. The computational complexity of the gradient optimization method used is not high, and the online update cycle can be implemented in a low-cost MCU to meet the needs of online parameter identification, thereby reducing the debugging cost while improving the practicality and scalability of the algorithm engineering.

[0073] In some embodiments, constructing an adjustable model includes: establishing a dynamic equation of a direct-drive turntable; the dynamic equation of the direct-drive turntable is a mathematical model determined based on the dynamic parameters of the direct-drive turntable; discretizing the dynamic equation of the direct-drive turntable to obtain a discrete dynamic equation; and constructing an adjustable model based on the discrete dynamic equation.

[0074] Specifically, a dynamic equation of a direct-drive turntable is first established. The dynamic equation of a direct-drive turntable is a mathematical model determined based on the dynamic parameters of the direct-drive turntable and can be used as a reference model in subsequent steps.

[0075] The dynamic equation of the direct-drive turntable is expressed as follows: ; in, is the angular velocity of the direct-drive turntable; is the moment of inertia of the turntable; is the viscous friction damping coefficient; is the disturbance torque, which is the sum of the disturbance torque caused by load torque, Coulomb friction and other factors; is the electromagnetic torque output by the motor.

[0076] Furthermore, the dynamic equation of the direct-drive turntable is discretized to obtain a discrete dynamic equation.

[0077] The expression of the discrete dynamics equation is as follows: ; in, is the signal sampling period; Indicates the time, Indicates the time; Indicates the direct drive rotary table Angular velocity of motion at any moment; Indicates the direct drive rotary table The angular velocity of motion at any moment.

[0078] Furthermore, based on the discrete dynamic equations, an adjustable model of the adaptive system is constructed.

[0079] Among them, the expression of the adjustable model of the adaptive system is as follows: ; ; ; ; in, represents the estimated value of the turntable speed by the adjustable model; 、 、 These are the three parameters to be identified of the direct-drive turntable.

[0080] The identification results of existing identification algorithms often exhibit fluctuations and errors. Based on this, the dynamic parameter identification method provided in this embodiment analyzes the causes of these fluctuations from the perspective of model building and takes targeted measures. By optimizing the algorithm structure, improving the design of the reference model and adjustable model, and incorporating the identification of the damping term, this method effectively addresses the issue of large fluctuations in identification results. This not only reduces the fluctuations and errors in the identification results of the remaining dynamic parameters, but also enables more accurate modeling of the direct-drive turntable.

[0081] The present invention also provides a specific example of a method for identifying kinetic parameters. Figures 2 to 7 , Figure 2 This is the second flow chart of the kinetic parameter identification method provided by the present invention. Figure 3 is a structural diagram of the direct-drive turntable system provided by the present invention, Figure 4 This is the third flow chart of the kinetic parameter identification method provided by the present invention. Figure 5 This is one of the comparison charts of the experimental results between the kinetic parameter identification method provided by the present invention and the traditional algorithm. Figure 6 This is the second comparison chart of the experimental results between the kinetic parameter identification method provided by the present invention and the traditional algorithm. Figure 7 This is the third comparison chart of experimental results between the kinetic parameter identification method provided by the present invention and the traditional algorithm.

[0082] Figure 2 and Figure 4 The kinetic parameter identification method is applied to Figure 3 Direct drive rotary table system.

[0083] like Figure 3 As shown in the figure, a direct-drive turntable speed-current closed-loop servo system (abbreviated as a direct-drive turntable system) is first built to obtain the angular velocity, torque current and other parameters required for inertia identification. The system includes a speed controller 1, a current controller 2, a coordinate transformation module 3, an SVPWM modulation module 4, a three-phase inverter 5, and a frameless permanent magnet synchronous rotating motor 6 inside the turntable ( Figure 3 denoted as PMSM), high-resolution photoelectric encoder 7, current sensor 8 and adaptive identification observer 9.

[0084] Among them, the photoelectric encoder 7 is used to feedback the mechanical angle of the turntable, the coordinate transformation module 3 is used for coordinate transformation and speed calculation, the current sensor 8 is used to measure the turntable phase current, and the difference between the command speed and the feedback speed is input to the speed controller 1 and the q-axis command current is calculated. , d-axis command current During the control process, it is always set to 0; the difference between the command current and the feedback current is input to the current controller 2 and the given voltage is calculated. 、 The given voltage is calculated after the coordinate transformation module 3 and the SVPWM modulation module 4 to obtain the duty cycle, which is used to control the output voltage of the three-phase inverter 5 、 、 , thereby controlling the rotational motion of the direct-drive turntable through the current sensor 8.

[0085] The mechanical angle of the turntable obtained by the photoelectric encoder 7 and the phase current obtained by the current sensor 8 are calculated by the servo driver embedded platform to obtain the actual angular velocity of the motor (i.e. the angular velocity of the direct-drive turntable movement). and motor torque current The servo drive transmits the signal to the host computer through serial communication. The communication cycle (i.e. signal sampling cycle) =0.001s. A kinetic parameter identification method based on the gradient optimization method is used on the host computer to achieve online identification of kinetic parameters.

[0086] Specifically, the direct-drive turntable dynamic equation of the direct-drive turntable speed-current closed-loop servo system is Discretization processing is performed to obtain discrete dynamic equations.

[0087] The expression of the discrete dynamics equation is as follows: ; in, =0.001s; Calculated from the collected torque current, , is the torque coefficient.

[0088] Furthermore, an adjustable model of the adaptive system is designed based on the discrete dynamic equations.

[0089] Among them, the expression of the adjustable model of the adaptive system is as follows: ; ; ; ; in, Indicates the adjustable model turntable speed An estimate of ; in the initialization setting, , , .

[0090] Furthermore, the output error is calculated .

[0091] Objective function Set to the square of the output error between the reference model and the adjustable model, that is, , the adjustment target is .

[0092] Furthermore, according to the gradient optimization method, the gradient of the objective function to each parameter to be identified is calculated, that is, the first gradient , second gradient and the third gradient , then: ; ; .

[0093] Furthermore, according to the current gradient values ​​(i.e. the first gradient , second gradient and the third gradient ), the exponential decay average of the square of each gradient is calculated using the following formula ( ): .

[0094] Furthermore, according to the current parameter update amount of each parameter to be identified, the exponential decay average value of the square of the parameter update amount of each parameter to be identified is calculated using the following formula ( ): .

[0095] Furthermore, the dynamic learning rate of parameter gradient descent is calculated using the following formula: ; Among them, the numerical stability term Set to .

[0096] Furthermore, each parameter to be identified is updated in the direction of the gradient. According to the current gradient value and the learning rate, each parameter to be identified can be iteratively updated, that is, the following formula is satisfied: .

[0097] The above model error calculation, gradient calculation, dynamic learning rate calculation and update process of each parameter to be identified are cycled to realize the online parameter identification process.

[0098] Among them, such as Figure 4As shown in the figure, the objective function value is used to judge its convergence. When the objective function reaches and continues to remain in the range of [-0.001, 0.001], it can be considered that all the parameters to be identified have converged.

[0099] Following the above steps, the accurate dynamic parameters of the direct-drive turntable system can be obtained. To further illustrate the technical effects of the present invention, this embodiment will experimentally test three identification algorithms: (1) the traditional Landau discrete adaptive identification algorithm; (2) an identification algorithm based on the gradient optimization method, but using a fixed learning rate; and (3) the dynamic parameter identification method of the present invention, i.e., an identification algorithm based on the gradient optimization method, using a dynamic learning rate.

[0100] The parameters of the direct-drive turntable used in the experiment are shown in Table 1.

[0101] Table 1

[0102] The turntable's moment of inertia and viscous friction damping coefficient were calculated using an offline least-squares fitting method and served as reference values ​​for subsequent identification results. The turntable was operating without load, with an S-shaped speed command curve in the speed range of [0, 40] rpm and a frequency of 3 Hz. The entire identification process lasted 25 seconds.

[0103] The experimental results of the traditional Landau discrete adaptive identification algorithm are as follows Figure 5 As shown by Figure 5 It can be seen that the traditional Landau discrete adaptive identification algorithm converges quickly, with a convergence time of less than 2 seconds, but its identification results are between 0.1430 and 0.1435. The traditional Landau discrete adaptive identification algorithm is not accurate enough in building the adjustable model, which leads to fluctuations and errors in the identification results.

[0104] When a fixed learning rate is used, the experimental results of the identification algorithm based on the gradient optimization method are as follows: Figure 6 As shown by Figure 6 It can be seen that after considering the viscous friction damping coefficient, the reference model is more accurate, and its identification accuracy has been greatly improved compared with the traditional Landau discrete adaptive identification algorithm. The identification result is 0.14285±0.00003 , the fluctuation of the identification results is greatly reduced. However, due to the use of a fixed learning rate, its identification speed is slow, and it has obvious overshoot and oscillation in the early stage of identification.

[0105] The experimental results of the kinetic parameter identification method of the present invention, i.e. the gradient optimal adaptive identification algorithm using dynamic learning rate, are as follows: Figure 7 As shown by Figure 7It can be seen that when the learning rate is dynamically adjusted, compared with Figure 6 The experimental results show that the algorithm can achieve fast convergence in the initial stage, the convergence time is within 1s, there is no large overshoot and oscillation during the identification process, the identification result curve is smooth, and the identification result is 0.14283±0.00003 , the accuracy has been greatly improved compared with the traditional Landau discrete adaptive identification algorithm.

[0106] The dynamic parameter identification method of the present invention combines accurate model construction, an adaptive law based on the gradient optimization method, and a dynamic learning rate adjustment strategy. As can be seen from the above experimental results, the scheme of the present invention improves the accuracy of the identification results, accelerates the convergence speed of the identification process, and reduces the debugging complexity.

[0107] To sum up, the online adaptive identification method of the direct-drive turntable dynamic parameters based on the gradient optimization method of the present invention is based on the model reference adaptive theory, takes the system damping term into consideration, and constructs an adaptive identification system based on the gradient optimization method. In order to improve the identification speed and identification accuracy, reduce the debugging complexity, and improve the adaptability of the algorithm, the learning rate is dynamically adjusted when using the gradient optimization method, thereby improving the accuracy, speed and convenience of the online identification of the turntable dynamic parameters.

[0108] The present invention also provides a dynamic parameter identification device. Figure 8 , Figure 8 8 is a schematic diagram of the structure of the kinetic parameter identification device provided by the present invention. In this embodiment, the kinetic parameter identification device includes a construction module 810, an objective function determination module 820, a first identification module 830 and a second identification module 840.

[0109] The construction module 810 is used to construct an adjustable model.

[0110] The adjustable model is a mathematical model constructed based on the parameters to be identified of the direct-drive turntable, and the parameters to be identified are determined based on the dynamic parameters of the direct-drive turntable.

[0111] The objective function determination module 820 is used to construct an objective function based on the adjustable model.

[0112] The first identification module 830 is used to update the parameters to be identified in the adjustable model based on the gradient optimization method until the objective function reaches a minimum value, thereby obtaining the converged parameters to be identified.

[0113] The second identification module 840 is used to identify the dynamic parameters of the direct-drive turntable based on the converged parameters to be identified.

[0114] In some embodiments, the dynamic parameters include the turntable moment of inertia, the viscous friction damping coefficient and the disturbance torque, and the parameters to be identified include a first parameter to be identified, a second parameter to be identified and a third parameter to be identified; wherein, the first parameter to be identified is determined based on the signal sampling period and the turntable moment of inertia; the second parameter to be identified is determined based on the signal sampling period, the turntable moment of inertia and the disturbance torque; the third parameter to be identified is determined based on the signal sampling period, the viscous friction damping coefficient and the turntable moment of inertia.

[0115] In some embodiments, the first identification module 830 is used to determine a first gradient, a second gradient, and a third gradient; the first gradient is the gradient of the objective function with respect to the first parameter to be identified, the second gradient is the gradient of the objective function with respect to the second parameter to be identified, and the third gradient is the gradient of the objective function with respect to the third parameter to be identified; determine a first parameter update amount of the first parameter to be identified, a second parameter update amount of the second parameter to be identified, and a third parameter update amount of the third parameter to be identified; determine a learning rate positive definite matrix of the gradient optimization method based on the first gradient, the second gradient, the third gradient, the first parameter update amount, the second parameter update amount, and the third parameter update amount; based on the learning rate positive definite matrix, update the parameters to be identified in the adjustable model according to the gradient direction until the objective function reaches a minimum value, and obtain the converged parameters to be identified.

[0116] In some embodiments, the first identification module 830 is used to determine a first learning rate of a first parameter to be identified based on the first gradient and the first parameter update amount; determine a second learning rate of a second parameter to be identified based on the second gradient and the second parameter update amount; determine a third learning rate of a third parameter to be identified based on the third gradient and the third parameter update amount; and construct a learning rate positive definite matrix of the gradient optimization method based on the first learning rate, the second learning rate, and the third learning rate.

[0117] In some embodiments, the first learning rate is determined based on an exponentially decaying average of the square of the first gradient and an exponentially decaying average of the square of the first parameter update amount; the second learning rate is determined based on an exponentially decaying average of the square of the second gradient and an exponentially decaying average of the square of the second parameter update amount; and the third learning rate is determined based on an exponentially decaying average of the square of the third gradient and an exponentially decaying average of the square of the third parameter update amount.

[0118] In some embodiments, a construction module 810 is used to establish a dynamic equation of a direct-drive turntable; the dynamic equation of a direct-drive turntable is a mathematical model determined based on the dynamic parameters of the direct-drive turntable; the dynamic equation of the direct-drive turntable is discretized to obtain a discrete dynamic equation; and an adjustable model is constructed based on the discrete dynamic equation.

[0119] The present invention also provides an electronic device. Figure 9Schematic diagram of the structure of the electronic device provided by the present invention, such as Figure 9 As shown, the electronic device may include: a processor 910, a communications interface 920, a memory 930, and a communication bus 940, wherein the processor 910, the communications interface 920, and the memory 930 communicate with each other via the communication bus 940. The processor 910 may call logic instructions in the memory 930 to execute the dynamic parameter identification method.

[0120] Furthermore, the logic instructions in the aforementioned memory 930 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product, stored in a storage medium, includes instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0121] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the kinetic parameter identification methods provided by the above methods.

[0122] The present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the dynamic parameter identification methods provided by the above methods.

[0123] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0124] Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in each embodiment or certain portions of the embodiments.

[0125] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for identifying dynamic parameters, characterized in that: include: Building adjustable models; The adjustable model is a mathematical model constructed based on the parameters to be identified of the direct-drive turntable, and the parameters to be identified are determined based on the dynamic parameters of the direct-drive turntable; constructing an objective function based on the adjustable model; Based on the gradient optimization method, the parameters to be identified in the adjustable model are updated until the objective function reaches a minimum value, thereby obtaining the converged parameters to be identified; Based on the converged parameters to be identified, the dynamic parameters of the direct-drive turntable are identified.

2. The method for identifying kinetic parameters according to claim 1, characterized in that: The dynamic parameters include the turntable moment of inertia, the viscous friction damping coefficient and the disturbance torque, and the parameters to be identified include a first parameter to be identified, a second parameter to be identified and a third parameter to be identified; Wherein, the first parameter to be identified is determined based on the signal sampling period and the moment of inertia of the turntable; The second parameter to be identified is determined based on the signal sampling period, the turntable moment of inertia and the disturbance torque; The third parameter to be identified is determined based on the signal sampling period, the viscous friction damping coefficient and the turntable moment of inertia.

3. The method for identifying dynamic parameters according to claim 2, characterized in that: The method of updating the parameters to be identified in the adjustable model based on the gradient optimization method until the objective function reaches a minimum value to obtain the converged parameters to be identified includes: Determine a first gradient, a second gradient, and a third gradient; the first gradient is the gradient of the objective function with respect to the first parameter to be identified, the second gradient is the gradient of the objective function with respect to the second parameter to be identified, and the third gradient is the gradient of the objective function with respect to the third parameter to be identified; Determining a first parameter update amount of the first parameter to be identified, a second parameter update amount of the second parameter to be identified, and a third parameter update amount of the third parameter to be identified; Determining a learning rate positive definite matrix of the gradient optimization method based on the first gradient, the second gradient, the third gradient, the first parameter update amount, the second parameter update amount, and the third parameter update amount; Based on the learning rate positive definite matrix, the parameters to be identified in the adjustable model are updated according to the gradient direction until the objective function reaches a minimum value, thereby obtaining the converged parameters to be identified.

4. The method for identifying dynamic parameters according to claim 3, characterized in that: The determining, based on the first gradient, the second gradient, the third gradient, the first parameter update amount, the second parameter update amount, and the third parameter update amount, of a learning rate positive definite matrix of the gradient optimization method includes: Determining a first learning rate for the first parameter to be identified based on the first gradient and the first parameter update amount; Determining a second learning rate of the second parameter to be identified based on the second gradient and the second parameter update amount; Determining a third learning rate of the third parameter to be identified based on the third gradient and the third parameter update amount; The learning rate positive definite matrix of the gradient optimization method is constructed based on the first learning rate, the second learning rate, and the third learning rate.

5. The method for identifying dynamic parameters according to claim 4, characterized in that: The first learning rate is determined based on an exponentially decaying average of the square of the first gradient and an exponentially decaying average of the square of the first parameter update amount; The second learning rate is determined based on an exponentially decaying average of the square of the second gradient and an exponentially decaying average of the square of the second parameter update amount; The third learning rate is determined based on an exponentially decaying average of the square of the third gradient and an exponentially decaying average of the square of the third parameter update amount.

6. The method for identifying kinetic parameters according to claim 1, characterized in that: The constructing of the adjustable model comprises: Establishing a dynamic equation of a direct-drive turntable; the dynamic equation of the direct-drive turntable is a mathematical model determined based on the dynamic parameters of the direct-drive turntable; Discretizing the dynamic equation of the direct-drive turntable to obtain a discrete dynamic equation; Based on the discrete dynamic equations, the adjustable model is constructed.

7. A dynamic parameter identification device, characterized in that: include: Building blocks for constructing adjustable models; The adjustable model is a mathematical model constructed based on the parameters to be identified of the direct-drive turntable, and the parameters to be identified are determined based on the dynamic parameters of the direct-drive turntable; An objective function determination module, configured to construct an objective function based on the adjustable model; A first identification module is configured to update the parameters to be identified in the adjustable model based on a gradient optimization method until the objective function reaches a minimum value, thereby obtaining converged parameters to be identified; The second identification module is used to identify the dynamic parameters of the direct-drive turntable based on the converged parameters to be identified.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the kinetic parameter identification method according to any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the kinetic parameter identification method according to any one of claims 1 to 6 is implemented.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the kinetic parameter identification method according to any one of claims 1 to 6 is implemented.

Images