Multi-physics field collaborative method for dynamic accuracy assurance of the bottom dead point of the slider of stamping equipment

Abstract

The present invention discloses a method for ensuring the dynamic accuracy of the bottom dead point of the slider of a stamping equipment in a coordinated manner with multiple physical fields. The method collects real-time monitoring data of physical field factors such as the temperature of the main bearing of the stamping equipment, the punching force, the number of strokes, and the displacement of the bottom dead point of the slider to construct a training and test data set; an approximate model between the physical field factors and the bottom dead point error of the slider is established based on a radial basis function neural network, and the model parameters are solved by the Bayesian method; with the number of strokes and the temperature of the main bearing as control variables and the minimization of the bottom dead point error of the slider as the goal, a bottom dead point accuracy compensator of the slider is constructed, and the bottom dead point accuracy compensator of the slider is executed to obtain the optimal control amount; the optimal control amount is transmitted to the control system of the stamping equipment to compensate the bottom dead point accuracy of the slider in real time. The present invention takes into account the coordinated influence mechanism of the multi-physical fields of the stamping equipment on the dynamic accuracy of the bottom dead point of the slider, and effectively ensures the dynamic accuracy of the bottom dead point of the slider through real-time compensation of the dynamic accuracy of the bottom dead point of the slider of the stamping equipment.

Description

Technical Field

[0001] The present invention belongs to the field of bottom dead point accuracy assurance of a stamping equipment slider, and specifically relates to a method for ensuring the dynamic accuracy of the bottom dead point of a stamping equipment slider in a multi-physical field coordinated manner. Background Art

[0002] Stamping equipment is the basic manufacturing equipment in important industrial fields such as automobile manufacturing, military industry, aerospace and rail transportation. Its technical level has become one of the important indicators to measure the comprehensive strength of a country's manufacturing industry. As a key performance indicator of high-speed stamping equipment, the accuracy of the slider bottom dead point directly affects the accuracy of stamping parts of stamping equipment, the service life of the mold and the qualified rate of stamping parts. Therefore, it is urgent to deeply analyze the influence mechanism of multi-physical field factors on the accuracy of the slider bottom dead point, explore the distribution law of the slider bottom dead point motion error, and develop a multi-physical field coordinated stamping equipment slider bottom dead point dynamic accuracy guarantee method.

[0003] At present, the research on dynamic compensation of slider bottom dead point accuracy does not consider the influence of multi-physical field coupling, and the existing technology still has deficiencies in real-time performance, compensation range and adaptability under complex working conditions. Especially in high-frequency stamping operations, the problems of compensation delay and poor dynamic accuracy stability need to be solved urgently. Summary of the invention

[0004] In order to solve the shortcomings of the prior art and achieve the purpose of ensuring the dynamic accuracy of the bottom dead point of the slider of the stamping equipment with multi-physical field collaboration, the present invention adopts the following technical solutions:

[0005] The method for ensuring the dynamic accuracy of the bottom dead point of the slider of a stamping equipment in cooperation with multiple physical fields includes the following steps:

[0006] Step 1: Collect real-time monitoring data of physical field factors of stamping equipment to construct a data set containing multi-physical field data and corresponding slider bottom dead point error data, and construct a training data set and a test data set respectively;

[0007] Step 2: construct a radial basis function neural network approximate model between multi-physics field factors and the slider bottom dead point error to reflect the nonlinear mapping relationship between multi-physics field data and the slider bottom dead point error data, substitute the training data set into the radial basis function neural network approximate model, train the radial basis function neural network, and solve the model weights; consider the uncertainty of the neural network model, and solve the shape parameters of the Gaussian radial basis function based on the test data set to improve the accuracy of the slider bottom dead point error approximate model; obtain the optimal approximate model based on the solved weights and shape parameters;

[0008] Step 3: Real-time monitoring of the slider bottom dead point error. If it exceeds the accuracy tolerance, the slider bottom dead point accuracy compensation is triggered. The optimal control amount of the multi-physical field control factor is found through the optimal approximate model to minimize the slider bottom dead point error.

[0009] Step 4: Transmit the optimal control variable to the control system of the stamping equipment.

[0010] Furthermore, the radial basis function neural network approximation model in step 2 is specifically:

[0011]

[0012] in, Representing Multiphysics Factors The corresponding slider bottom dead point error, represents multi-physics factors, represents the weight from the i-th node in the hidden layer of the model to the output layer, p represents the number of nodes in the hidden layer, represents the training sample center of each hidden layer node, represents the output of the i-th hidden layer node, represents the Euclidean distance between two points, represents the Gaussian radial basis function, and a represents the shape parameter of the Gaussian radial basis function.

[0013] Furthermore, in step 2, the multi-physics field training data set And the corresponding slider bottom dead point error training data set Substitute the radial basis function neural network approximation model, train the radial basis function neural network, and solve the model weights , n represents the number of multi-physical field factors, and m represents the number of monitoring data samples.

[0014] Furthermore, in step 2, based on the multi-physics field test data set And the corresponding slider bottom dead point error test data set , n represents the number of multi-physics factors, m represents the number of monitoring data samples, and the shape parameter a of the Gaussian radial basis function is solved by the Bayesian update method. The posterior distribution of the shape parameter a is:

[0015]

[0016] in, Indicates the proportional sign; Represents the prior distribution of shape parameter a; represents the likelihood function.

[0017] Furthermore, the likelihood function in step 2 is as follows:

[0018]

[0019] in:

[0020]

[0021]

[0022] in, represents the standard deviation of the observation error, represents the m×m unit matrix; based on the above formula, the transition Markov Monte Carlo method is used to obtain the posterior distribution of the shape parameter a.

[0023] Furthermore, after obtaining the posterior distribution of the shape parameter a, the specific value of the shape parameter is obtained by the following formula:

[0024]

[0025] in, Indicates the expectation symbol, Represents the probability space of shape parameters.

[0026] Furthermore, in step 3, accuracy compensation is performed by executing the following formula:

[0027]

[0028] in, represents the optimal control quantity of the multi-physics field control factor, represents the optimal approximate model, represents the monitoring data of the multi-physical field at the kth operating moment, e represents the absolute value of the slider bottom dead point error obtained by the optimal approximate model, represents the control quantity at the kth running time, and Represent the lower limit vector and upper limit vector of the control variable, X k Indicates uncontrolled physical factors.

[0029] Furthermore, in step 1, the real-time monitoring data of the physical field factors include the main bearing temperature, punching force, number of strokes, bottom dead point displacement of the slider and the running time of the stamping equipment, and the collected data are cleaned and denoised; the optimal control quantity of the multi-physical field control factors in step 3 includes the control quantity of the number of strokes and the main bearing temperature.

[0030] Furthermore, in step 4, the optimal control amount of the number of strokes and the main bearing temperature is transmitted to the drive control system and the temperature control system of the stamping equipment, as follows:

[0031] The optimal control value of the main bearing temperature is transmitted to the temperature controller, and the controller outputs instructions to control the actuator to increase or decrease the lubricating oil temperature, so as to reduce the influence of the thermal deformation of the slider and improve the bottom dead point accuracy of the slider;

[0032] The optimal control amount of stroke frequency is transmitted to the drive control system, and the stroke frequency adjustment amount is input to the driver through the controller to adjust the stroke frequency of the slider in real time to ensure the accuracy of the slider bottom dead point position.

[0033] The advantages and beneficial effects of the present invention are:

[0034] (1) The present invention establishes a radial basis function neural network model between the multi-physical field factors of stamping equipment and the bottom dead point error of the slider, and realizes the mathematical representation of the nonlinear mapping relationship between the multi-physical field factors and the bottom dead point error of the slider;

[0035] (2) The present invention proposes a method for solving the parameters of the radial basis function neural network model based on Bayesian updating, taking into account the uncertainty of the model. Through Bayesian inference, it can provide an uncertainty measure for each prediction, achieve full capture of monitoring data information, and improve the accuracy of the prediction of the bottom dead point error of the slider;

[0036] (3) The present invention integrates the radial basis function neural network model to establish a slider bottom dead point accuracy compensator that takes into account the influence of multi-physical field coupling. When the dynamic accuracy of the bottom dead point exceeds the given accuracy tolerance, the accuracy compensator is triggered to obtain the optimal control amount, dynamically compensate for the slider bottom dead point accuracy, and effectively ensure the dynamic accuracy of the slider bottom dead point of the stamping equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a flow chart of a method for ensuring the dynamic accuracy of the bottom dead point of a slider of a stamping equipment in collaboration with multiple physical fields provided in an embodiment of the present invention.

[0038] Figure 2 A flow chart of dynamic precision control of the bottom dead point of a slider provided in an embodiment of the present invention.

[0039] Figure 3 This is a diagram of the dynamic accuracy compensation results of the bottom dead point of the slider provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0040] The specific implementation of the present invention is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described here is only used to illustrate and explain the present invention, and is not used to limit the present invention.

[0041] like Figure 1 As shown, the method for ensuring the dynamic accuracy of the bottom dead point of the slider of a stamping equipment in cooperation with multiple physical fields includes the following steps:

[0042] Step 1: Collect real-time monitoring data of physical field factors such as main bearing temperature, punching force, punching frequency, slider bottom dead point displacement and running time of stamping equipment, clean and reduce noise of the collected data, and build a training data set, including multi-physical field data And the slider bottom dead point error data ; Build test data sets, including multi-physics data And the slider bottom dead point error data , where n=5 represents the number of multi-physics field factors, and m=200 represents the number of monitoring data samples.

[0043] Step 2: In order to reflect the nonlinear mapping relationship between the multi-physical field factors and the slider bottom dead point error, a radial basis function neural network approximate model between the multi-physical field factors and the slider bottom dead point error is constructed, specifically:

[0044] (1)

[0045] in, represents multi-physics factors, represents the weight from the i-th (i=1, 2, …, p; p=12) node in the hidden layer to the output layer, represents the training sample center of each hidden layer node, represents the output of the i-th hidden layer node, represents the Euclidean distance between two points, represents the Gaussian radial basis function, and a represents the shape parameter of the Gaussian radial basis function.

[0046] The Bayesian update method is used to solve the approximate model parameters of the slider bottom dead point error, specifically:

[0047] Step 2.1: Multiphysics training dataset And the corresponding slider bottom dead point error data Substitute into formula (1), train the radial basis function neural network, and solve

[0048]

[0049] Step 2.2: Considering the uncertainty of the neural network model, in order to improve the accuracy of the slider bottom dead point error approximation model, based on the test data set , the shape parameter a of the Gaussian radial basis function is solved by the Bayesian update method, and the posterior distribution of the shape parameter a is:

[0050] (2)

[0051] in, Indicates the proportional sign; Represents the prior distribution of shape parameter a; Represent the likelihood function as follows:

[0052] (3)

[0053] in:

[0054] (4)

[0055] (5)

[0056] in, represents the standard deviation of the observation error, represents the m×m unit matrix; based on formula (2) to formula (5), the transition Markov Monte Carlo method is used to obtain the posterior distribution of the shape parameter a; after obtaining the posterior distribution of the parameter, the specific value of the parameter can be obtained by the following formula:

[0057] (6)

[0058] in, Indicates the expectation symbol, Represents the probability space of shape parameters.

[0059] Step 3: Real-time monitoring of the slider bottom dead point error, if it exceeds the accuracy tolerance =0.005μm, the slider bottom dead center accuracy compensator is triggered and formula (7) is executed, as follows:

[0060] (7)

[0061] in, represents the optimal control quantity of the stroke frequency and the main bearing temperature, It represents the approximate model of the slider bottom dead point error, represents the k=10th running time t 10 = 2h monitoring data of multi-physics field, e represents the absolute value of the slider bottom dead point error obtained by the approximate model, represents the control quantity at the kth running time, and Represent the lower limit vector and upper limit vector of the control variable, X k Indicates uncontrolled physical factors.

[0062] Step 4: Transmit the optimal control quantity of the stroke frequency and the main bearing temperature to the drive control system and temperature control system of the stamping equipment, such as Figure 2 As shown, the details are as follows:

[0063] Step 4.1: The optimal control value of the main bearing temperature is transmitted to the temperature controller, and the controller outputs instructions to control the actuator to increase or decrease the lubricating oil temperature, so as to reduce the influence of the thermal deformation of the slider and improve the bottom dead point accuracy of the slider;

[0064] Step 4.2: The optimal control amount of the stroke frequency is transmitted to the drive control system. The controller inputs the stroke frequency adjustment amount to the driver, and the stroke frequency of the slider is adjusted in real time to ensure the accuracy of the slider bottom dead point position.

[0065] Figure 3 The compensation effect of the dynamic accuracy of the bottom dead point of the slider after the implementation of the method proposed in the present invention is demonstrated. It can be seen that the method proposed in the present invention effectively guarantees the dynamic accuracy of the bottom dead point of the slider, and the calculation results verify the effectiveness of the method for guaranteeing the dynamic accuracy of the bottom dead point of the slider of the stamping equipment in the multi-physical field collaboration of the present invention.

[0066] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some or all of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for ensuring the dynamic accuracy of the bottom dead point of the slider of a stamping equipment in collaboration with multiple physical fields, characterized in that The steps include: Step 1: Collect real-time monitoring data of physical field factors of stamping equipment to construct a data set containing multi-physical field data and corresponding slider bottom dead point error data, and construct a training data set and a test data set respectively; the real-time monitoring data of physical field factors include the main bearing temperature, punching force, punching times, slider bottom dead point displacement and running time of stamping equipment; Step 2: construct a radial basis function neural network approximate model between multi-physical field factors and the bottom dead point error of the slider to reflect the nonlinear mapping relationship between the multi-physical field factors and the bottom dead point error of the slider, substitute the training data set into the radial basis function neural network approximate model, train the radial basis function neural network, and solve the model weights; based on the test data set, solve the shape parameters of the Gaussian radial basis function; based on the solved weights and shape parameters, obtain the optimal approximate model; Step 3: monitor the slider bottom dead point error in real time. If it exceeds the accuracy tolerance, trigger the slider bottom dead point accuracy compensation. Through the optimal approximate model, find the optimal control amount of the multi-physical field control factor to minimize the slider bottom dead point error; the optimal control amount of the multi-physical field control factor includes the control amount of the stroke frequency and the main bearing temperature; Step 4: transmitting the optimal control amount to the control system of the stamping equipment; The optimal control quantity of the stroke frequency and the main bearing temperature is transmitted to the drive control system and temperature control system of the stamping equipment, as follows: The optimal control value of the main bearing temperature is transmitted to the temperature controller, and the lubricating oil temperature is increased or decreased by controlling the actuator to reduce the influence of thermal deformation of the slider; The optimal control amount of stroke frequency is transmitted to the drive control system, and the stroke frequency adjustment amount is input to the driver through the controller to adjust the stroke frequency of the slider in real time.

2. The method for ensuring the dynamic accuracy of the bottom dead point of the slider of a stamping equipment in a multi-physical field coordinated manner according to claim 1 is characterized in that: The radial basis function neural network approximation model in step 2 is specifically: in, Indicates the bottom dead point error of the slider corresponding to the multi-physics factor S, S represents the multi-physics factor, ω i represents the weight from the i-th node in the hidden layer of the model to the output layer, p represents the number of nodes in the hidden layer, Ψ i represents the training sample center of each hidden layer node, f(||S-Ψ i ||,a) represents the output of the i-th hidden layer node, ||·|| represents the Euclidean distance between two points, f represents the Gaussian radial basis function, and a represents the shape parameter of the Gaussian radial basis function.

3. The method for ensuring the dynamic accuracy of the bottom dead point of the slider of a stamping equipment in a multi-physical field coordinated manner according to claim 1 is characterized in that: In step 2, the multi-physics field training data set And the corresponding slider bottom dead point error training data set Substitute the radial basis function neural network approximation model, train the radial basis function neural network, and solve the model weight ω=[ω1,ω2,...,ω p ] T , n represents the number of multi-physical field factors, and m represents the number of monitoring data samples.

4. The method for ensuring the dynamic accuracy of the bottom dead point of the slider of a stamping equipment in a multi-physical field coordinated manner according to claim 2 is characterized in that: In step 2, based on the multi-physics field test data set And the corresponding slider bottom dead point error test data set n represents the number of multi-physics factors, m represents the number of monitoring data samples, and the shape parameter a of the Gaussian radial basis function is solved by the Bayesian update method. The posterior distribution of the shape parameter a is: p(a|S test ,AND test )∝p(Y test |S test ,a)p(a) Among them, ∝ represents the proportional symbol; p(a) represents the prior distribution of shape parameter a; p(Y test |S test ,a) represents the likelihood function.

5. The method for ensuring the dynamic accuracy of the bottom dead point of the slider of a stamping equipment in a multi-physical field coordinated manner according to claim 4 is characterized in that: The likelihood function in step 2 is as follows: in: Among them, σ p represents the standard deviation of the observation error, and I represents the m×m unit matrix. Based on the above formula, the transition Markov Monte Carlo method is used to obtain the posterior distribution of the shape parameter a.

6. The method for ensuring the dynamic accuracy of the bottom dead point of the slider of a stamping equipment in a multi-physical field coordinated manner according to claim 4 is characterized in that: After obtaining the posterior distribution of the shape parameter a, the specific value of the shape parameter is obtained by the following formula: Where E[·] represents the expected sign, Ω a Represents the probability space of shape parameters.

7. The method for ensuring the dynamic accuracy of the bottom dead point of the slider of a stamping equipment in a multi-physical field coordinated manner according to claim 1 is characterized in that: In step 3, accuracy compensation is performed by executing the following formula: Among them, d optimal represents the optimal control quantity of the multi-physics field control factor, represents the optimal approximate model, represents the monitoring data of the multi-physics field at the kth operating moment, e represents the absolute value of the slider bottom dead point error obtained by the optimal approximate model, and d k represents the control quantity at the kth running time, and Respectively represent the lower and upper limits of the control quantity, X k Indicates uncontrolled physical factors.

Images