CN-121979106-A - Error compensation method for numerical control machining of automobile stamping die

CN121979106ACN 121979106 ACN121979106 ACN 121979106ACN-121979106-A

Abstract

The invention relates to the technical field of industrial control, in particular to an error compensation method for numerical control machining of an automobile stamping die, which comprises the following steps of collecting temperature and thermal displacement data of a plurality of heat sensitive points of a numerical control machine tool; the method comprises the steps of calculating effective thermal driving potential by considering time lag effects of temperature level and historical temperature change, calculating thermal phase direction gradient by combining a temperature difference value and nonlinear gain, combining the effective thermal driving potential and the thermal phase direction gradient to form a feature vector training support vector regression model, inputting the feature vector in real time to obtain a compensation value, and superposing the compensation value to a coordinate instruction by using dynamic origin offset. The method can accurately reflect thermal inertia, capture dynamic evolution of the thermal state, comprehensively characterize the adaptability of the thermal state lifting model, correct thermal errors in real time, and therefore improve the machining precision and the surface quality of the automobile stamping die.

Inventors

ZHANG XIANGJUN
SU XIAOFEI

Assignees

泊头市金键模具有限责任公司

Dates

Publication Date: 20260505
Application Date: 20260408

Claims (10)

1. The error compensation method for numerical control machining of the automobile stamping die is characterized by comprising the following steps of: acquiring temperature data of a plurality of heat sensitive points of the numerical control machine tool and knife tip heat displacement data at corresponding moments; Calculating an effective thermal driving potential based on the temperature data, wherein the effective thermal driving potential is positively correlated with the temperature data in the set window length and is negatively correlated with the historical temperature data and the current temperature data in the set window length based on the square of the time difference of the Gaussian kernel; Calculating a thermal phase direction gradient based on the temperature data and the effective thermal driving potential, wherein the thermal phase direction gradient comprises a difference value of temperature values at the current moment and the previous moment and a nonlinear gain term of the effective thermal driving potential, and the thermal phase direction gradient is positively correlated with the difference value and the nonlinear gain term; combining the effective thermal driving potential and the thermal phase direction gradient to form a feature vector, and training a support vector regression model based on the feature vector and the thermal displacement data of the tool nose; In the numerical control machining process of the automobile stamping die, calculating the characteristic vector in real time and inputting a trained support vector regression model to obtain a compensation value; And superposing the compensation value to a coordinate motion instruction of the numerical control machine tool in a dynamic origin offset mode to realize error compensation of numerical control machining of the automobile stamping die.
2. The error compensation method for numerical control machining of an automobile stamping die according to claim 1, wherein the effective thermal driving potential is obtained by the following steps: and summing products of temperature values at all moments within the set window length and Gaussian kernel functions to obtain effective thermal driving potential, wherein the Gaussian kernel functions take time differences between the current moment and the historical moment as independent variables.
3. The error compensation method for numerical control machining of an automobile stamping die according to claim 2, wherein the value range of the set window length covers an effective heat influence range of a numerical control machine tool heat time constant.
4. The error compensation method for numerical control machining of an automobile stamping die according to claim 1, wherein the thermal phase direction gradient satisfies: ; In the formula, For the moment of time Is provided with a thermal phase direction gradient of (a), For the moment of time Is used for the temperature value of (a), For the moment of time Is a temperature value of a previous time instant of (a), In order to sample the time interval of the time, Is a nonlinear correction coefficient of the thermal rigidity of the structure, For the moment of time Is used for the effective thermal driving potential of the (c), And the temperature rise reference value is set as the limit temperature rise reference value.
5. The error compensation method for numerical control machining of an automobile stamping die according to claim 4, wherein the nonlinear structural thermal stiffness correction coefficient is used for amplifying the weight of the gradient in a high-temperature high-thermal potential state, so that an error evolution track in a heating process and a cooling process is separated in a characteristic space.
6. The error compensation method for numerical control machining of an automobile stamping die according to claim 1, wherein the support vector regression model adopts a radial basis function.
7. The error compensation method for numerical control machining of an automobile stamping die according to claim 6, wherein the kernel width parameters of the radial basis function are optimized through 5-fold cross validation.
8. The error compensation method for numerical control machining of an automobile stamping die according to claim 1, further comprising, before calculating the feature vector in real time and inputting the feature vector to the trained support vector regression model: And performing Z-score standardization treatment on the feature vectors, wherein the mean value and standard deviation of the Z-score standardization treatment are calculated based on all feature vectors in a training stage.
9. The error compensation method for numerical control machining of an automobile stamping die according to claim 1, wherein the plurality of heat sensitive points comprise a main shaft front bearing seat, a column guide rail side surface, a lathe bed interior and a machining environment.
10. The error compensation method for numerical control machining of an automobile stamping die according to claim 1, wherein the method for acquiring the thermal displacement data of the tool nose comprises the following steps: the coordinate deviation of the standard ball or the reference block is periodically measured in the machining gap using a laser displacement sensor or a contact probe.

Description

Error compensation method for numerical control machining of automobile stamping die Technical Field The invention relates to the technical field of industrial control. More specifically, the invention relates to an error compensation method for numerical control machining of an automobile stamping die. Background The automobile stamping die is core technological equipment for manufacturing automobile bodies, wherein the large covering part dies such as side walls, wings and the like directly determine the appearance forming precision and assembly matching property of the automobile bodies, have strict industry requirements on the indexes such as the dimensional precision of machining molded surfaces, the smoothness of curved surfaces and the like, usually rely on a large gantry type numerical control machining center to finish the machining of the dies, and are influenced by factors such as huge size of the die bodies, complex topological structure of the curved surfaces and the like, and the single finish machining period is often as long as tens of hours. In a long-time machining process, the cutting heat generated by high-speed rotation of a main shaft and day and night changes of ambient temperature can cause the numerical control machine tool structure to generate complex thermal deformation, the profile precision of a die is affected, and the existing thermal error compensation technology mostly adopts support vector regression (Support Vector Regression, SVR) to establish a mapping model of temperature and error. However, most of the existing modeling methods are static mapping, neglecting thermal inertia of a numerically-controlled machine tool structure, in actual machining, obvious time lag exists in heat conduction inside the numerically-controlled machine tool structure, so that temperature and displacement curves show hysteresis ring characteristics, such as frequent fluctuation of spindle rotation speed in die corner cleaning machining, or in a cooling stage of rough and finish machining switching, even if the surface temperatures displayed by sensors are the same, due to different internal thermal states of the machine tool, such as different thermal errors generated in a heating or cooling stage, the traditional static model cannot sense the historical thermal state, and prediction failure is caused under an unsteady working condition, so that tool contact marks or ripple degree exceeding is generated on the surface of an automobile stamping die, and quality of the automobile stamping die is further affected. Disclosure of Invention In order to solve the problems that in the prior art, thermal error prediction distortion and insufficient compensation precision are caused under an unsteady working condition due to neglecting of thermal inertia of a numerical control machine tool, and cutter marks or out-of-tolerance waviness are generated on the surface of an automobile stamping die, so that the quality of the automobile stamping die is affected, the invention provides an error compensation method for numerical control machining of the automobile stamping die, which comprises the following steps: The method comprises the steps of collecting temperature data of a plurality of heat sensitive points of a numerical control machine tool and knife tip heat displacement data at corresponding moments, calculating effective heat driving potential based on the temperature data, wherein the effective heat driving potential is positively correlated with the temperature data in a set window length, is negatively correlated with historical temperature data and current temperature data in the set window length based on the square of a time difference of a Gaussian kernel, calculates heat phase direction gradient based on the temperature data and the effective heat driving potential, the heat phase direction gradient comprises a difference value of temperature values at the current moment and the previous moment and a nonlinear gain term of the effective heat driving potential, the heat phase direction gradient is positively correlated with the difference value and the nonlinear gain term, combines the effective heat driving potential and the heat phase direction gradient into a feature vector, trains a support vector regression model based on the feature vector and the knife tip heat displacement data, calculates the feature vector in real time and inputs the trained support vector regression model to obtain a compensation value in a numerical control machining process of an automobile stamping die, and the compensation value is superimposed to a numerical control die in a dynamic offset mode to realize the numerical control machine tool coordinate motion error compensation form. The method realizes time sequence modeling of the temperature driving effect by calculating the effective thermal driving potential, considers the time delay effect of the current temperature level and the historical temperature c