Feed system model building method and device

Abstract

The present application discloses a model building method and device for a feed system. The method includes: determining a first model of the feed system based on the dynamic characteristics of a ball screw stage, wherein the first input parameter of the first model is the output torque of the servo drive, the first output parameter of the first model is the output displacement of the ball screw stage, and the servo drive is used to drive the ball screw stage; obtaining the first input parameter and the first output parameter of the ball screw stage when it is running in multiple target states, respectively, to obtain a first input parameter set and a first output parameter set; performing integral processing on the first model to obtain a second model; and modifying the first model based on the first input parameter set, the first output parameter set and the second model to obtain a target model of the feed system. The present application solves the technical problems in the related art that the high-order feed system model is complex to calculate and the parameters are difficult to determine.

Description

Technical Field

[0001] The present application relates to the technical field of servo feeding systems, and in particular to a model building method and device for a feeding system. Background Art

[0002] CNC machine tools are an important modern manufacturing equipment. As the manufacturing industry has higher and higher requirements for parts processing accuracy and production efficiency, high speed and high precision have become the main development direction of CNC machine tools. Among them, the high-speed linear feed system is the feed actuator of high-speed and high-precision CNC machine tools. Its motion performance directly determines the processing accuracy of CNC machine tools. There are two main forms of linear feed systems: ball screws and linear motors. Ball screws have been widely used in CNC machine tools because of their high rigidity, low inertia sensitivity, low cutting force sensitivity, reliable performance, and low cost.

[0003] When studying the ball screw feed system, it is usually necessary to model it. However, since the performance of the ball screw is related to many factors such as the dynamic characteristics of its mechanical structure, the working environment, the performance of the control system and the performance matching, the model order will be relatively high. When directly performing dynamic modeling on the high-order system, the calculation of the model parameters will be more complicated. If some minor factors are ignored and simplified to a second-order model, although the calculation process is simple, it will affect the accuracy of the model.

[0004] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0005] The embodiments of the present application provide a method and device for constructing a model of a feed system, so as to at least solve the technical problems in the related art that the high-order feed system model is complex to calculate and the parameters are difficult to determine.

[0006] According to one aspect of an embodiment of the present application, a model building method for a feed system is provided, including: determining a first model of the feed system based on the dynamic characteristics of a ball screw stage, wherein a first input parameter of the first model is the output torque of a servo drive, and a first output parameter of the first model is the output displacement of the ball screw stage, and the servo drive is used to drive the ball screw stage; acquiring the first input parameter and the first output parameter when the ball screw stage operates in multiple target states respectively, to obtain a first input parameter set and a first output parameter set; integrating the first model to obtain a second model; and modifying the first model based on the first input parameter set, the first output parameter set and the second model to obtain a target model of the feed system.

[0007] Optionally, according to the dynamic characteristics of the ball screw stage, a torque balance equation and a viscous friction force balance equation are established respectively; and the first model is established according to the torque balance equation and the viscous friction force balance equation.

[0008] Optionally, the servo drive is controlled to operate at multiple different target speeds, so that the ball screw stage operates under multiple target states, wherein the target speed of the servo drive corresponds one-to-one to the target state of the ball screw stage; the output torque of the servo drive at multiple different target speeds is obtained respectively to obtain the first input parameter set; the output displacement of the ball screw stage under the multiple target states is obtained respectively to obtain the first output parameter set, wherein the input parameters in the first input parameter set correspond one-to-one to the output parameters in the first output parameter set.

[0009] Optionally, current curves of the servo driver at multiple different target speeds are collected respectively; based on the current curves at the multiple target speeds, output torque curves of the servo driver at the multiple target speeds are determined respectively to obtain the output torque.

[0010] Optionally, the order of the first model is N, and the first models are subjected to 1 to M times integration processing to obtain an integral model respectively, and the second model includes M integral models, wherein 1≤M≤N.

[0011] Optionally, each of the first input parameters in the first input parameter set is integrated to obtain a second input parameter set; the second model is trained based on the second input parameter set and the first output parameter set to obtain target coefficients; the coefficients of the first model are corrected based on the target coefficients to obtain the target model.

[0012] Optionally, the order of the first model is N, the second input parameter set includes multiple second input parameter subsets, and each first input parameter in the first input parameter set is subjected to 1 to M-fold integration processing to obtain a second input parameter subset corresponding to each first input parameter, and the second input parameter subset includes M+1 second input parameters, where 1≤M≤N.

[0013] Optionally, each second input parameter in the second input parameter set is input into the second model to obtain a predicted second output parameter set; a loss function is determined based on the first output parameter set and the second output parameter set; and the model coefficients in the second model are adjusted based on the loss function to obtain the target coefficients.

[0014] Optionally, a simulation model of a servo system is determined based on a target model of the feed system, wherein the simulation model of the servo system is used to analyze operating results of the ball screw stage when driving a load.

[0015] According to another aspect of an embodiment of the present application, a model building device for a feed system is also provided, including: a determination module, used to determine a first model of the feed system according to the dynamic characteristics of a ball screw stage, wherein a first input parameter of the first model is the output torque of a servo drive, and a first output parameter of the first model is the output displacement of the ball screw stage, and the servo drive is used to drive the ball screw stage; an acquisition module, used to acquire the first input parameter and the first output parameter when the ball screw stage operates in multiple target states respectively, to obtain a first input parameter set and a first output parameter set; a processing module, used to perform integration processing on the first model to obtain a second model; and a correction module, used to correct the first model according to the first input parameter set, the first output parameter set and the second model to obtain a target model of the feed system.

[0016] According to another aspect of an embodiment of the present application, a non-volatile storage medium is also provided, wherein the non-volatile storage medium includes a stored program, wherein when the program is running, the device where the non-volatile storage medium is located is controlled to execute the above-mentioned model building method of the feeding system.

[0017] According to another aspect of an embodiment of the present application, a processor is further provided, wherein the processor is used to run a program, wherein the program executes the above-mentioned model building method of the feeding system when running.

[0018] In the embodiment of the present application, the first model of the feed system is first determined according to the dynamic characteristics of the ball screw stage, the first input parameter of the first model is the output torque of the servo drive, the first output parameter of the first model is the output displacement of the ball screw stage, and the servo drive is used to drive the ball screw stage; the first input parameter and the first output parameter of the ball screw stage when operating in multiple target states are obtained to obtain the first input parameter set and the first output parameter set; the first model is integrated to obtain the second model; the first model is corrected according to the first input parameter set, the first output parameter set and the second model to obtain the target model of the feed system. Among them, the present application adopts the method of neural network training, and uses experimental measured data to train the established model, which greatly simplifies the process of obtaining the final high-order model, and considering the lack of forward propagation information of the BP (Error Back Propagation) algorithm during the model training process, the model and the input parameters are integrated before training, which can more accurately determine the relevant parameters of the model, thereby solving the technical problems of complex calculation of high-order feed system models and difficult parameter determination in related technologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0020] Figure 1 is a flow chart of a model building method of a feeding system according to an embodiment of the present application;

[0021] Figure 2 is a schematic diagram of a ball screw stage mechanical structure according to an embodiment of the present application;

[0022] Figure 3 is a structural schematic diagram of a ball screw stage dynamics model according to an embodiment of the present application;

[0023] Figure 4 is a schematic diagram of an interface for collecting a current curve of a servo driver according to an embodiment of the present application;

[0024] Figure 5 It is a schematic diagram of an interface for collecting an output displacement curve of a ball screw stage according to an embodiment of the present application;

[0025] Figure 6 is a schematic diagram of a feeding system model training result according to an embodiment of the present application;

[0026] Figure 7 is a schematic diagram of another feeding system model training result according to an embodiment of the present application;

[0027] Figure 8 is a schematic diagram of a feeding system model verification result according to an embodiment of the present application;

[0028] Fig. 9 It is a structural schematic diagram of a model building device of a feeding system according to an embodiment of the present application. DETAILED DESCRIPTION

[0029] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0030] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0031] Example 1

[0032] According to an embodiment of the present application, an embodiment of a model building method for a feed system is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0033] Figure 1 is a flow chart of an optional model building method of a feeding system according to an embodiment of the present application, such as Figure 1 As shown, the method includes steps S102-S108, wherein:

[0034] Step S102, determining a first model of the feed system according to the dynamic characteristics of the ball screw stage, wherein a first input parameter of the first model is an output torque of the servo drive, a first output parameter of the first model is an output displacement of the ball screw stage, and the servo drive is used to drive the ball screw stage.

[0035] In some optional embodiments of the present application, the dynamic characteristics of the ball screw stage mainly consider the torque balance relationship and the viscous friction balance relationship during its operation. Therefore, the torque balance equation and the viscous friction balance equation can be established respectively according to the dynamic characteristics of the ball screw stage; the differential equation of the ball screw stage is determined according to the torque balance equation and the viscous friction balance equation, and then the first model of the feed system is obtained. Among them, considering the accuracy of the model, the order of the first model should be at least higher than the second order, and can be the third order, fourth order, fifth order, etc. The fourth order model is taken as an example for explanation here, but it does not constitute a specific limitation.

[0036] Specifically, Figure 2 A schematic diagram of the mechanical structure of a ball screw table is shown, which includes a servo motor, a coupling, a left rolling bearing group, a linear guide, a ball screw, a guide slider, a nut seat, a screw nut, a right rolling bearing group and a workbench. The servo motor is used to output torque to drive the associated components to ultimately achieve the displacement of the workbench.

[0037] Based on the above mechanical structure, the viscous friction is mainly considered, and the dynamic model structure of the ball screw table is obtained as follows: Figure 3 As shown, where T is the output torque of the servo motor, J m is the total moment of inertia of the rotating parts, θ m is the angular displacement of the servo motor, c b is the total viscous friction coefficient of the rotating parts, l is the lead of the ball screw, K is the total stiffness of the feed system, c i is the total viscous friction coefficient in the axial direction, c t is the total viscous friction coefficient of the guide rail, M t is the total mass of the workbench, x t is the output displacement of the screw stage. Based on the dynamic relationship, the torque balance equation is established as follows:

[0038]

[0039] The viscous friction force balance equation is established as follows:

[0040]

[0041] Then, by eliminating the variable θ from the two equilibrium equations m , the differential equation of the ball screw table is as follows:

[0042]

[0043] Among them, the coefficients (a, b, c, d) related to the output displacement of the ball screw stage in the differential equation are all unknown parameters, so the equation is a fourth-order differential equation, that is, the first model of the feed system obtained is a fourth-order model, in which the input of the first model is the output torque T of the servo motor (servo drive), and the output of the first model is the output displacement x of the ball screw stage. t .

[0044] It can be understood that the dynamic characteristics of the ball screw stage described in the above embodiment are T, J m ,θ m 、c b , l, K, c i For, c t For M t and x t The dynamic relationship satisfied between them is the above torque balance equation and viscous friction balance equation.

[0045] Step S104, obtaining first input parameters and first output parameters of the ball screw stage when it operates in multiple target states, and obtaining a first input parameter set and a first output parameter set.

[0046] If the values ​​of the coefficients in the above differential equations are to be directly calculated, the calculation process is very complicated, and the results obtained may not be satisfactory. Therefore, the present application proposes to train the model using experimental measured data through neural network training to determine the correlation coefficients in the model. In order to ensure the accuracy of the training results, the input and output parameters of the ball screw stage when running under multiple target states can be obtained to obtain the corresponding first input parameter set and first output parameter set, which are used as training data.

[0047] In some optional embodiments of the present application, the running state of the ball screw stage is controlled by the speed of the servo drive, so the servo drive can be controlled to run at multiple target speeds to drive the ball screw stage to run at multiple target states, wherein the target speed of the servo drive corresponds to the target state of the ball screw stage one by one, and it can be understood that the target state of the ball screw stage is the running state of the ball screw stage when the servo drive drives the ball screw stage at the corresponding target speed; then the output torque of the servo drive at multiple target speeds is obtained to obtain a first input parameter set; then the output displacement of the ball screw stage at multiple target states is obtained to obtain a first output parameter set, wherein the input parameters in the first input parameter set correspond to the output parameters in the first output parameter set one by one. It should be noted that since the output torque of the servo drive and the output displacement of the ball screw stage both change over time, the premise for confirming that a set of input parameters corresponds to an output parameter is that the time for obtaining the set of input parameters and output parameters is the same.

[0048] Specifically, when determining the first input parameter set, since the servo drive cannot directly collect its output torque, it needs to be calculated indirectly, considering that there is the following relationship between the output torque and current of the servo drive:

[0049]

[0050] Among them, f is the maximum flux linkage between the permanent magnet pole and the stator winding, L d and L q The direct axis and quadrature axis main inductances, i d and i q They are the direct-axis and quadrature-axis components of the stator three-phase current respectively.

[0051] It can be found that in the control strategy i d =0, there is a linear relationship between the output torque and the current:

[0052]

[0053] Therefore, the current curves of the servo driver at multiple target speeds can be collected first, and then the output torque curves at multiple target speeds can be determined based on the current curves at the multiple target speeds to obtain the output torque, thereby obtaining the corresponding first input parameter set. Figure 4 A schematic diagram of an interface for collecting a servo drive current curve is shown, wherein the current changes over time. Of course, the output torque of the servo drive can also be obtained by other methods, which are not limited here and are not listed one by one.

[0054] When the first output parameter set is determined, the output displacement curve of the ball screw stage under multiple target states can be directly collected through a displacement sensor, wherein a commonly used displacement sensor is a laser sensor with higher precision. Figure 5 The figure shows a schematic diagram of the interface of a sensor collecting the output displacement curve of a ball screw stage, in which the output displacement changes with time. Due to the problem of sensor setting, Figure 5 The data in needs to be negated and converted into positive numbers.

[0055] It should be noted that in the above process, when setting multiple target speeds of the servo driver, it is necessary to control each target speed within a reasonable range. Since the range of the distance sensor is limited (such as 20mm), if the servo driver speed is set too large, the sensor will collect fewer points, affecting the acquisition of training data. Therefore, a speed threshold can be set so that multiple target speeds do not exceed the speed threshold.

[0056] Step S106, performing integration processing on the first model to obtain a second model.

[0057] Step S108, modifying the first model according to the first input parameter set, the first output parameter set and the second model to obtain a target model of the feeding system.

[0058] Specifically, each first input parameter in the first input parameter set can be integrated to obtain a second input parameter set; the second model can be trained based on the second input parameter set and the first output parameter set to obtain target coefficients; the coefficients of the first model can be corrected based on the target coefficients to obtain a target model.

[0059] If the first model is trained directly using the first input parameter set and the first output parameter set, that is, directly using T as input, x t As output to train the above four-order model, Figure 6 A schematic diagram of a training result is shown, and it can be seen that there is a large error between the predicted result of the model output obtained through training and the actual result. That is to say, it is impossible to obtain a high-precision feed system model by directly training the first model using the first input parameter set and the first output parameter set.

[0060] By analyzing the above problems, we find that it is mainly due to the multiple solutions of high-order differential equations. For the same input T, there are multiple output solutions x t ; Since the basic basis of neural network training is BP algorithm, that is, the connection weights of each neuron are adjusted inversely according to the error, when the output solution x cannot be determined according to the input T t The lack of information will lead to insufficient forward propagation information under the BP algorithm, which will ultimately lead to low accuracy of the training results.

[0061] Therefore, in some optional embodiments of the present application, the first model may be appropriately integrated according to its order to obtain a second model. Specifically, when the order of the first model is N, the first model may be integrated from 1 to M times to obtain an integral model, and then the second model is obtained, and the second model includes the obtained M integral models, where 1≤M≤N.

[0062] Afterwards, each first input parameter in the first input parameter set may be integrated according to the order of the first model to obtain a second input parameter set, and the second input parameter set includes multiple second input parameter subsets. Specifically, when the order of the first model is N, each first input parameter in the first input parameter set may be integrated from 1 to M times to obtain a second input parameter subset corresponding to each first input parameter, and each second input parameter subset includes M+1 second input parameters, where 1≤M≤N.

[0063] Still taking the fourth-order model as an example, for example, the fourth-order differential equation can be processed by single integration, double integration, triple integration and quadruple integration in sequence, and the obtained second model includes the following four equations:

[0064]

[0065] In this way, we can get the output solution x t Directly related equations.

[0066] Specifically, for each first input parameter T in the first input parameter set, single integration, double integration, triple integration and quadruple integration can be performed respectively to obtain the corresponding ∫T, ∫∫T, ∫∫∫T and ∫∫∫∫T, and then the five parameters including T are taken together as the second input parameter subset corresponding to the first input parameter T, so as to obtain the input variables corresponding to the above-mentioned second model.

[0067] After the above-mentioned integration processing, each second input parameter in the second input parameter set can be input into the second model to obtain a predicted second output parameter set; the loss function is determined based on the first output parameter set and the second output parameter set; the model coefficients in the second model are adjusted based on the loss function to obtain the target coefficients.

[0068] For example, T, ∫T, ∫∫T, ∫∫∫T and ∫∫∫∫T in the above second input parameter set are used as input variables of the second model, and the model predicts the corresponding second output parameter x t ', and determine the first output parameter x corresponding to the second input parameter T t , according to the first output parameter x tand the second output parameter x t 'Determine the mean square error loss function and iteratively train it using the parameters in the second input parameter set.

[0069] Among them, due to m 1 With m 11 、m 21 、m 31 、m 41 Between 2 With m 12 、m 22 、m 32 、m 42 Between a and a 1 、a 2 、a 3 、a 4 Between b and b 1 、b 2 、b 3 、b 4 Between c and c 1 、c 2 、c 3 、c 4 Between d and d 1 ,d 2 ,d 3 ,d 4 The data are interrelated rather than independent, and the coefficients of each model in the second model (m 11 、m 21 、m 31 、m 41 、m 12 、m 22 、m 32 、m 42 、a 1 、a 2 、a 3 、a 4 、b 1 、b 2 、b 3 、b 4 、c 1 、c 2 、c 3 、c 4 ,d 1 ,d 2 ,d 3 ,d 4 ), so that the error of the second model is less than the preset error, and finally the target coefficient (m) for the first model can be obtained 1 、m 2, a, b, c, d), and then the first model is modified according to the target coefficient to obtain the final target model of the feeding system:

[0070]

[0071] It should be noted that when modifying the first model, only the coefficients that are different from the original coefficients and the target coefficients need to be modified. If the original coefficients and the target coefficients are the same, there is no need to modify the coefficients. 1 、m 2 When a, b, c, and d have all changed, all of them will be corrected; when the target coefficient m 1 、m 2 , a, d, c, d, a does not change, so only m needs to be corrected 1 、m 2 , b, c, d.

[0072] Of course, in other embodiments, the order N of the first model is continued to be 4, and the fourth-order first model can be integrated once to obtain the second model, and then the first input parameter T can be integrated once to obtain ∫T, and then T and ∫T can be used as input variables to train the second model in the neural network to obtain the target model; or the first model can be integrated once and twice to obtain the second model, and then the first input parameter T can be integrated once and twice to obtain ∫T, ∫ ∫T, and then use T, ∫T and ∫∫T as input variables to train the second model in the neural network to obtain the target model; or the first model can be integrated once, twice and tripled to obtain the second model, and then the first input parameter T can be integrated once, twice and tripled to obtain ∫T, ∫∫T and ∫∫∫T respectively, and then T, ∫T, ∫∫T and ∫∫∫T can be used as input variables to train the second model in the neural network to obtain the target model. The details are not elaborated here.

[0073] Among them, the number of times the first model needs to be integrated can be determined based on the error of the first model, that is, the value of M can be determined based on the error of the first model. For example, the average error of the first model is calculated, the difference between the average error and the error threshold is calculated, and the M value is determined based on the interval in which the difference falls; when the difference integral is within the first interval, M=1, and a single integration is performed; when the difference is within the second interval, M=2, and a single and double integration are performed respectively; when the difference is within the third interval, M=3, and a single integration, a double integration, and a triple integration are performed respectively; they are not described one by one here. The M value can also be determined directly based on the interval in which the average error is located. In this way, the amount of calculation can be reduced and the efficiency of model construction can be improved.

[0074] Figure 7A schematic diagram of a training result is shown, and it can be seen that when the target speed of the servo drive is determined, the error between the predicted result output by the target model of the feed system and the actual result is basically zero.

[0075] In some optional embodiments of the present application, the results of the servo drive driving the ball screw stage at other speeds other than the above-mentioned multiple target speeds are predicted by the feed system target model. The results are as follows: Figure 8 As shown in the figure, the maximum error between the predicted result and the actual experimental result is 0.057mm. Part of the error comes from the randomness of the initial weights of the neural network model, and the other part should be that the initial points of the current curve and the displacement curve are not completely corresponding in the actual acquisition, because the acquisition platforms of the two curves are not together, resulting in a deviation in the acquisition time. That is, through verification, the feed system target model trained by this application can accurately reflect the characteristics of the actual ball screw stage.

[0076] It should be noted that in the embodiments of the present application, the order of the first model of the feed system is only 4th order for exemplary explanation. It can be understood that the above construction method is not limited to the model of the 4th order feed system, but can also be 3rd order, 5th order, 6th order, 7th order, 8th order, etc., which are not listed here one by one. The different orders of the model will result in different numbers of integrations for the above first model. For example, if the target order is n, the first model can be integrated n times, that is, the first model is respectively integrated once, double integrated, ..., n times, to obtain the corresponding equations, and then the above step S108 is executed to obtain the corresponding n-order target model.

[0077] In some optional embodiments of the present application, a servo system simulation model can also be determined based on the target model of the feed system, wherein the servo system simulation model is used to analyze the operating results of the ball screw stage when driving the load. Specifically, the target model obtained by the above training can be embedded in the servo system model, and the optimal solution of the servo system model can be sought by simulating and analyzing the operating rules of the system under various load devices. Since the servo system model is based on an accurate feed system target model and an application load model built in advance, the servo system is directly analyzed and studied from the application side, and the operating results of the driver are predicted. The results are more in line with reality, which improves the practicality of the servo system model for the environment, and can also be used to solve the problem of remote debugging when the debugger is not on site.

[0078] In the embodiment of the present application, the first model of the feed system is first determined according to the dynamic characteristics of the ball screw stage, the first input parameter of the first model is the output torque of the servo drive, the first output parameter of the first model is the output displacement of the ball screw stage, and the servo drive is used to drive the ball screw stage; the first input parameter and the first output parameter of the ball screw stage when operating in multiple target states are obtained to obtain the first input parameter set and the first output parameter set; the first model is integrated to obtain the second model; the first model is corrected according to the first input parameter set, the first output parameter set and the second model to obtain the target model of the feed system. Among them, the present application adopts the method of neural network training, and uses experimental measured data to train the established model, which greatly simplifies the process of obtaining the final high-order model, and considering the lack of forward propagation information of the BP (Error Back Propagation) algorithm during the model training process, the model and the input parameters are integrated before training, which can more accurately determine the relevant parameters of the model, thereby solving the technical problems of complex calculation of high-order feed system models and difficult parameter determination in related technologies.

[0079] Example 2

[0080] According to an embodiment of the present application, a model building device for a feeding system for implementing the above-mentioned model building method for a feeding system is also provided. Figure 4 As shown, the device includes a determination module 90, an acquisition module 92, a processing module 94 and a correction module 96, wherein:

[0081] Determination module 90 is used to determine the first model of the feed system according to the dynamic characteristics of the ball screw stage, wherein the first input parameter of the first model is the output torque of the servo drive, the first output parameter of the first model is the output displacement of the ball screw stage, and the servo drive is used to drive the ball screw stage.

[0082] In some optional embodiments of the present application, the dynamic characteristics of the ball screw stage mainly consider the torque balance relationship and the viscous friction balance relationship during its operation. Therefore, the torque balance equation and the viscous friction balance equation can be established respectively according to the dynamic characteristics of the ball screw stage; the differential equation of the ball screw stage is determined according to the torque balance equation and the viscous friction balance equation, and then the first model of the feed system is obtained. Among them, considering the accuracy of the model, the order of the first model should be at least higher than the second order, and can be the third order, fourth order, fifth order, etc., which is not specifically limited here.

[0083] The acquisition module 92 is used to acquire the first input parameter and the first output parameter of the ball screw stage when the ball screw stage is running in multiple target states, and obtain a first input parameter set and a first output parameter set.

[0084] In some optional embodiments of the present application, the running state of the ball screw stage is controlled by the speed of the servo drive, so the servo drive can be controlled to run at multiple target speeds to drive the ball screw stage to run at multiple target states, wherein the target speed of the servo drive corresponds to the target state of the ball screw stage one by one, and it can be understood that the target state of the ball screw stage is the running state of the ball screw stage when the servo drive drives the ball screw stage at the corresponding target speed; then the output torque of the servo drive at multiple target speeds is obtained to obtain a first input parameter set; then the output displacement of the ball screw stage at multiple target states is obtained to obtain a first output parameter set, wherein the input parameters in the first input parameter set correspond to the output parameters in the first output parameter set one by one. It should be noted that since the output torque of the servo drive and the output displacement of the ball screw stage both change over time, the premise for confirming that a set of input parameters corresponds to an output parameter is that the time for obtaining the set of input parameters and output parameters is the same.

[0085] When determining the first input parameter set, the current curves of the servo driver at multiple target speeds can be first collected, and then the output torque curves at multiple target speeds can be determined based on the current curves at the multiple target speeds to obtain the output torque; when determining the first output parameter set, the output displacement curves of the ball screw table under multiple target states can be directly collected through the displacement sensor to obtain the output displacement.

[0086] The processing module 94 is used to perform integration processing on the first model to obtain a second model.

[0087] The correction module 96 is used to correct the first model according to the first input parameter set, the first output parameter set and the second model to obtain a target model of the feeding system.

[0088] In some optional embodiments of the present application, the first model may be appropriately integrated according to its order to obtain a second model. Specifically, when the order of the first model is N, the first model may be integrated from 1 to M times to obtain an integral model, and the second model includes the obtained M integral models, where 1≤M≤N.

[0089] Afterwards, each first input parameter in the first input parameter set may be integrated according to the order of the first model to obtain a second input parameter set, and the second input parameter set includes multiple second input parameter subsets. Specifically, when the order of the first model is N, each first input parameter in the first input parameter set may be integrated from 1 to M times to obtain a second input parameter subset corresponding to each first input parameter, and each second input parameter subset includes M+1 second input parameters, where 1≤M≤N.

[0090] After the above-mentioned integral processing, each second input parameter in the second input parameter set can be input into the second model to obtain a predicted second output parameter set; a loss function is determined based on the first output parameter set and the second output parameter set; and the model coefficients in the second model are adjusted based on the loss function to obtain target coefficients. Among them, the commonly used loss function is the mean square error loss function.

[0091] It should be noted that each module in the model building device of the feeding system in the embodiment of the present application corresponds one by one to the implementation steps of the model building method of the feeding system in Example 1. Since a detailed description has been given in Example 1, some details not reflected in this embodiment can be referred to Example 1 and will not be elaborated here.

[0092] Example 3

[0093] According to an embodiment of the present application, a non-volatile storage medium is also provided, which includes a stored program, wherein when the program is running, the device where the non-volatile storage medium is located is controlled to execute the above-mentioned model building method of the feeding system.

[0094] According to an embodiment of the present application, a processor is also provided, which is used to run a program, wherein the above-mentioned model building method of the feeding system is executed when the program is running.

[0095] Specifically, the following steps are executed when the program is running: S102: Determine the first model of the feed system according to the dynamic characteristics of the ball screw stage, wherein the first input parameter of the first model is the output torque of the servo drive, the first output parameter of the first model is the output displacement of the ball screw stage, and the servo drive is used to drive the ball screw stage; S104: Obtain the first input parameter and the first output parameter when the ball screw stage is running in multiple target states, and obtain a first input parameter set and a first output parameter set; S106: Integrate the first model to obtain a second model; S108: According to the first input parameter set, the first output parameter set and the second model, modify the first model to obtain the target model of the feed system. Optionally, when the program is running, the sub-steps of step S102, step S104, step S106 and step S108 can also be executed to implement the sub-steps of the above step S102, step S104, step S106 and step S108.

[0096] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0097] In the above embodiments of the present application, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0098] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0099] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed over multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0100] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0101] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or all or part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of each embodiment method of the present application. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk, etc. Various media that can store program codes.

[0102] The above are only preferred implementations of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A model construction method for a feed system, applied to a ball screw stage, It is characterized in that The method comprises: Determine a first model of the feed system according to the dynamic characteristics of the ball screw stage, wherein a first input parameter of the first model is an output torque of a servo drive, a first output parameter of the first model is an output displacement of the ball screw stage, and the servo drive is used to drive the ball screw stage; Acquire the first input parameter and the first output parameter when the ball screw stage is running in a plurality of target states respectively, to obtain a first input parameter set and a first output parameter set; Performing integration processing on the first model to obtain a second model; According to the first input parameter set, the first output parameter set and the second model, the first model is modified to obtain a target model of the feeding system; The step of performing integration processing on the first model to obtain the second model comprises: performing 1 to M times integration processing on the first model to obtain an integral model respectively, the order of the first model is N, and the second model includes M integral models, wherein 1≤M≤N; According to the first input parameter set, the first output parameter set and the second model, the first model is modified to obtain the target model of the feed system, including: integrating each of the first input parameters in the first input parameter set to obtain the second input parameter set; training the second model according to the second input parameter set and the first output parameter set to obtain target coefficients; and modifying the coefficients of the first model according to the target coefficients to obtain the target model.

2. The method according to claim 1, It is characterized in that Determining a first model of the feed system according to the dynamic characteristics of the ball screw stage includes: According to the dynamic characteristics of the ball screw stage, a torque balance equation and a viscous friction balance equation are established respectively; The first model is established according to the torque balance equation and the viscous friction balance equation.

3. The method according to claim 1, It is characterized in that Acquiring the first input parameter and the first output parameter when the ball screw stage is running in a plurality of target states respectively, and obtaining a first input parameter set and a first output parameter set, including: Controlling the servo drive to operate at a plurality of different target speeds, respectively, so that the ball screw stage operates at a plurality of target states, wherein the target speeds of the servo drive correspond one to one to the target states of the ball screw stage; Respectively acquiring the output torque of the servo driver at a plurality of different target speeds to obtain the first input parameter set; The output displacements of the ball screw stage under the plurality of target states are respectively acquired to obtain the first output parameter set, wherein the input parameters in the first input parameter set correspond one-to-one to the output parameters in the first output parameter set.

4. The method according to claim 3, It is characterized in that Respectively obtaining the output torque of the servo drive at a plurality of different target speeds, including: respectively collecting current curves of the servo driver at a plurality of different target speeds; According to the current curves at the target rotational speeds, the output torque curves of the servo driver at the target rotational speeds are determined respectively to obtain the output torque.

5. The method according to claim 1, It is characterized in that The order of the first model is N, the second input parameter set includes a plurality of second input parameter subsets, and each of the first input parameters in the first input parameter set is integrated to obtain a second input parameter set, including: Each first input parameter in the first input parameter set is processed by 1 to M times integration to obtain a second input parameter subset corresponding to each first input parameter, wherein the second input parameter subset includes M+1 second input parameters, where 1≤M≤N.

6. The method according to claim 1, It is characterized in that Training the second model according to the second input parameter set and the first output parameter set to obtain a target coefficient includes: Inputting each second input parameter in the second input parameter set into the second model to obtain a predicted second output parameter set; Determining a loss function according to the first output parameter set and the second output parameter set; The model coefficients in the second model are adjusted according to the loss function to obtain the target coefficients.

7. The method according to claim 1, It is characterized in that The method further comprises: A simulation model of a servo system is determined based on the target model of the feed system, wherein the simulation model of the servo system is used to analyze the operation result of the ball screw stage when driving a load.

8. A model building device for a feeding system, It is characterized in that include: A determination module, used for determining a first model of the feed system according to the dynamic characteristics of the ball screw stage, wherein a first input parameter of the first model is an output torque of a servo drive, a first output parameter of the first model is an output displacement of the ball screw stage, and the servo drive is used for driving the ball screw stage; An acquisition module, used for acquiring the first input parameter and the first output parameter when the ball screw stage is running in a plurality of target states respectively, to obtain a first input parameter set and a first output parameter set; A processing module, configured to perform integration processing on the first model to obtain a second model, comprising: performing one to M times integration processing on the first model to obtain an integral model respectively, the order of the first model is N, and the second model includes M integral models, wherein 1≤M≤N; A correction module is used to correct the first model according to the first input parameter set, the first output parameter set and the second model to obtain a target model of the feed system, including: integrating each of the first input parameters in the first input parameter set to obtain a second input parameter set; training the second model according to the second input parameter set and the first output parameter set to obtain a target coefficient; and correcting the coefficients of the first model according to the target coefficients to obtain the target model.

9. A non-volatile storage medium, It is characterized in that The non-volatile storage medium includes a stored program, wherein when the program is executed, the device where the non-volatile storage medium is located is controlled to execute the model building method of the feeding system according to any one of claims 1 to 7.

10. A processor, It is characterized in that The processor is used to run a program, wherein the program, when running, executes the model building method of the feeding system according to any one of claims 1 to 7.

Images