CN-122007735-A - Welding path optimization method and system for welding robot based on data driving

CN122007735ACN 122007735 ACN122007735 ACN 122007735ACN-122007735-A

Abstract

The application relates to the technical field of automatic operation, in particular to a welding path optimization method and system of a welding robot based on data driving. The method comprises the steps of presetting a thermal-mechanical coupling model, setting an initial sensitivity coefficient, collecting welding current and arc voltage in real time in a welding process, calculating instantaneous thermal power, performing time integration to obtain an accumulated enthalpy value, inputting the accumulated enthalpy value into the thermal-mechanical coupling model to obtain a theoretical deformation predicted value, calculating an actual deformation estimated value based on a difference value between the arc voltage and a reference voltage, constructing a sliding time window to calculate a residual sequence, integrating to obtain a drift characteristic value, updating the sensitivity coefficient based on the drift characteristic value, and correcting a reference position coordinate of a welding robot by using a deformation compensation quantity to generate a welding gun position compensation instruction. The application realizes the sub-millimeter dynamic compensation of the thermal deformation of the workpiece through a sensitivity self-adaptive correction mechanism of residual integration, and obviously improves the tracking precision and consistency of a welding path.

Inventors

LIN ZHILIANG
LIANG TIANSHENG
FENG JIANCHUN
WU BINGHENG

Assignees

广东新威博电器有限公司

Dates

Publication Date: 20260512
Application Date: 20260310

Claims (10)

1. The welding robot welding path optimization method based on data driving is characterized by comprising the following steps of: Presetting a thermo-mechanical coupling model, wherein the thermo-mechanical coupling model defines a nonlinear mapping relation between unit accumulated heat input quantity and surface variables of a workpiece, and sets an initial sensitivity coefficient; collecting welding current and arc voltage in real time in a welding process, calculating the product of the welding current and the arc voltage to obtain instantaneous heat power, and performing time integration on the instantaneous heat power to obtain an accumulated enthalpy value; The accumulated enthalpy value is used as input of a thermo-mechanical coupling model to obtain a theoretical deformation predicted value, and meanwhile, an actual deformation estimated value is obtained based on the difference value between the arc voltage and the reference voltage; constructing a sliding time window, calculating the difference value between the actual deformation estimated value and the corrected theoretical deformation predicted value in the sliding time window to obtain a residual sequence, and integrating the residual sequence to obtain a drift characteristic value; And multiplying the updated sensitivity coefficient with the theoretical deformation predicted value to obtain a deformation compensation quantity, correcting the reference position coordinate of the welding robot by using the deformation compensation quantity, generating a welding gun position compensation instruction, and driving the welding robot to execute path correction.
2. The method for optimizing a welding path of a welding robot based on data driving of claim 1, wherein the process of creating the thermo-mechanical coupling model comprises: acquiring thermal physical property parameters and mechanical property parameters of a workpiece material; Establishing an initial mapping function of the heat input quantity and the deformation quantity based on a heat conduction equation and a thermoelastic constitutive relation; and acquiring actual measurement deformation data under different heat input conditions through an off-line calibration experiment, and performing fitting correction on parameters of the initial mapping function to generate the heat-machine coupling model.
3. The method for optimizing a welding path of a data-driven welding robot of claim 1, the method is characterized in that the calculation process of the accumulated enthalpy value comprises the following steps: Reading output values of a welding current sensor and an arc voltage sensor in a preset sampling period; calculating the product of the welding current and the arc voltage to obtain instantaneous thermal power; And carrying out local time integration on the instantaneous thermal power in a preset sliding time window to obtain the accumulated enthalpy value.
4. The method for optimizing a welding path of a welding robot based on data driving according to claim 1, wherein the obtaining of the theoretical deformation prediction value comprises: inputting the accumulated enthalpy value to an input of the thermo-mechanical coupling model; And the thermal-mechanical coupling model outputs a theoretical deformation predicted value corresponding to the accumulated enthalpy value through interpolation or table lookup operation based on a pre-stored nonlinear mapping curve.
5. The method for optimizing a welding path of a welding robot based on data driving of claim 1, wherein the obtaining of the measured deformation estimate comprises: acquiring an arc voltage value at the current moment and a preset reference voltage value; Calculating the difference value between the arc voltage value and the reference voltage value to obtain a voltage deviation value; Converting the voltage deviation amount into arc length variation from the end of the welding gun to the surface of the workpiece based on the corresponding relation between the arc length and the voltage in the static arc characteristic; And taking the absolute value of the arc length variation as an actual deformation estimated value of the current workpiece surface.
6. The method for optimizing a welding path of a welding robot based on data driving of claim 1, wherein the correction of the theoretical deformation prediction value comprises: acquiring a sensitivity coefficient at the current moment; and multiplying the theoretical deformation predicted value output by the thermo-mechanical coupling model by the sensitivity coefficient to obtain a corrected theoretical deformation predicted value.
7. The data-driven welding robot welding path optimization method of claim 1, wherein the update process of the sensitivity coefficient comprises: multiplying the drift characteristic value by a preset self-adaptive learning rate to obtain the adjustment increment; Adding the current sensitivity coefficient and the adjustment increment to obtain an updated sensitivity coefficient; And the self-adaptive learning rate is dynamically adjusted according to the absolute value of the drift characteristic value, the self-adaptive learning rate is increased when the absolute value exceeds a preset threshold value, and the self-adaptive learning rate is decreased when the absolute value is lower than the preset threshold value.
8. The data-driven welding robot welding path optimization method of claim 1, wherein the generation of the gun position compensation command comprises: Reading the current reference position coordinates of the welding robot; multiplying the updated sensitivity coefficient by the theoretical deformation predicted value to obtain the deformation compensation quantity; and adding the height direction component of the reference position coordinate and the deformation compensation quantity to generate a target position coordinate in the height direction as the welding gun position compensation instruction.
9. The data-driven welding robot welding path optimization method of claim 1, further comprising a safety protection step of: Setting an upper limit value and a lower limit value of a safety interval of the sensitivity coefficient; when the updated sensitivity coefficient exceeds the safety interval, judging that the sensor is in an abnormal physical deformation state; And in the abnormal physical deformation state, outputting an abnormal alarm signal and switching the welding robot to a degraded operation mode or triggering emergency shutdown.
10. Welding robot welding path optimizing system based on data drive, characterized by comprising: A processor; A memory in which a computer program is stored; Wherein the processor is configured to implement the data driven based welding robot welding path optimization method of any one of claims 1 to 9 when executing the computer program.

Description

Welding path optimization method and system for welding robot based on data driving Technical Field The application relates to the technical field of automatic operation. More particularly, the application relates to a data-driven welding robot-based welding path optimization method and system. Background The water heater inner container is used as a core pressure-bearing member of the water heater, and the welding quality of the water heater inner container is directly related to the safety and the service life of a product. The welding of the inner container of the water heater usually involves long-distance end socket circular seams and barrel longitudinal seams, and most of materials used are enamel steel or stainless steel sheets with the wall thickness ranging from 1.0 millimeter to 3.0 millimeters. In actual production, welding robots generally perform welding operations according to a fixed trajectory taught in advance. However, due to the significant heat accumulation effect of the thin-walled material under continuous welding heat input, the workpiece can undergo nonlinear radial expansion and axial expansion, resulting in real-time shifting of the actual weld trajectory relative to the taught trajectory. Existing welding path optimization techniques are largely divided into two categories. The first type is an off-line planning method based on an intelligent algorithm, for example, a path between weld joint endpoints is planned by using a deep reinforcement learning algorithm, and global point sequence optimization is performed by using an ant colony algorithm. The method reduces the dependence on an accurate mathematical model, but the core is a deep learning black box model, the method lacks of interpretability in an industrial field, and the method lacks of dynamic self-adaptive adjustment capability for real-time path deviation caused by thermal deformation in the welding process. The second type is a real-time tracking method based on laser or vision sensor, and the welding gun is provided with a laser sensor at the front end to scan the shape of the welding seam in real time and feed back the deviation value for correction. However, such methods are lag-reactive adjustments, are greatly affected by arc and spatter disturbances, and cannot utilize historical welding data to achieve global predictive optimization of the path. In summary, the prior art has the defects that the sensing and the optimization are mutually disjointed, the existing sensing device can only passively correct the current deviation and cannot actively predict the path change trend, massive historical data such as current and voltage generated in the welding process are not converted into knowledge for improving the path precision, so that the device cannot self-evolve along with the increase of production batches, and nonlinear deformation caused by heat accumulation in the welding of a thin-wall container is difficult to realize accurate sub-millimeter compensation through a traditional geometric modeling method. Disclosure of Invention The application aims to provide a welding path optimization method and a system for a welding robot based on data driving, which are used for solving the problem of insufficient tracking precision of a welding path in the prior art. In a first aspect, the welding path optimization method of the welding robot based on data driving comprises the steps of presetting a heat-machine coupling model, defining a nonlinear mapping relation between unit accumulated heat input quantity and surface variables of a workpiece, setting an initial sensitivity coefficient, collecting welding current and arc voltage in real time in a welding process, calculating the product of the welding current and the arc voltage to obtain instantaneous heat power, carrying out time integration on the instantaneous heat power to obtain an accumulated heat enthalpy value, taking the accumulated heat enthalpy value as input of the heat-machine coupling model to obtain a theoretical deformation predicted value, simultaneously calculating based on the difference value between the arc voltage and a reference voltage to obtain an actual deformation estimated value, constructing a sliding time window, calculating the difference value between the actual deformation estimated value and the corrected theoretical deformation predicted value in the sliding time window to obtain a residual sequence, integrating the residual sequence to obtain a drift characteristic value, updating the sensitivity coefficient based on the drift characteristic value, multiplying the updated sensitivity coefficient and the deformation predicted value to obtain a deformation compensation quantity, correcting the reference position coordinate of the welding robot by using the deformation compensation quantity, and generating a position compensation command of the welding robot to drive the welding gun to correct the welding path. According to the applic