CN-121635006-B - Composite swing welding control method with self-adaptive variable parameters and related equipment

CN121635006BCN 121635006 BCN121635006 BCN 121635006BCN-121635006-B

Abstract

The application provides a composite swing welding control method with self-adaptive variable parameters and related equipment, and relates to the technical field of welding; the method comprises the steps of inputting real-time multi-physical-field sensing data into a simulation model for predicting molten pool behaviors, outputting predicted values of molten pool state parameters in a preset time period, matching composite swing track parameters corresponding to the predicted values of the molten pool state parameters from a preset database, and controlling the tip of a welding gun to execute asymmetric spiral oscillation movement in a preset resident area according to a microscopic modulation track until the tip of the welding gun leaves the preset resident area when the welding gun enters the resident area of the swing edge of a welding seam according to a macroscopic reference track. By implementing the method, the problems of overheating, undercut or insufficient fusion of the edge caused by fixed parameters can be alleviated, and the forming quality of the welding seam and the stability of the welding process are improved.

Inventors

SUN QIUYANG
FANG XIAONAN
LIU JINPING
TANG HAO
FENG YINGCHAO
WANG YING
SUN LIUQING
LU KUN
REN JINGXIN
JIANG YONG
YAN CANCAN
Liu Kangtai

Assignees

中国核工业二三建设有限公司
核工业工程研究设计有限公司

Dates

Publication Date: 20260508
Application Date: 20251203

Claims (9)

1. The method for controlling the self-adaptive variable-parameter composite swing welding is applied to a composite swing welding control system and is characterized by comprising the following steps of: Acquiring real-time multi-physical field sensing data, wherein the real-time multi-physical field sensing data comprises at least two groups of heterogeneous sensor information for representing the geometric shape, temperature field distribution and arc stability of a molten pool; inputting the real-time multi-physical-field sensing data into a simulation model for predicting the molten pool behavior, and outputting a predicted value of a molten pool state parameter in a future preset time period; Matching composite swing track parameters corresponding to the predicted value of the molten pool state parameter from a preset database, wherein the composite swing track parameters comprise macroscopic reference track parameters and microscopic modulation track parameters for defining a composite motion track of a welding gun tip; When a welding gun enters a preset residence area of the welding seam swinging edge according to a macroscopic reference track defined by the macroscopic reference track parameters, controlling a welding gun tip to execute asymmetric spiral vibration movement in the preset residence area according to a microscopic modulation track defined by the microscopic modulation track parameters until the welding gun tip leaves the preset residence area, wherein the track shape of the asymmetric spiral vibration movement is asymmetrically distributed about the swinging direction; wherein after the step of controlling the welding gun tip to execute asymmetric spiral oscillation motion in the preset residence zone according to the micro-modulation track defined by the micro-modulation track parameters, the method further comprises the following steps: Monitoring a molten pool width change value in the preset residence zone in real time, wherein the molten pool width change value reflects the heat accumulation degree of the edge zone; When the molten pool width change value exceeds a preset threshold value, adjusting an amplitude attenuation coefficient of the spiral oscillation motion according to a preset proportion, wherein the amplitude attenuation coefficient is used for controlling the decreasing rate of the spiral motion radius; and adjusting the residence time of the welding gun tip in the preset residence zone according to the adjusted amplitude attenuation coefficient.
2. The method according to claim 1, wherein the step of acquiring real-time multi-physical field sensing data comprises: Acquiring molten pool surface image data by a visual sensor, wherein the molten pool surface image data comprises molten pool contours, surface textures and liquid metal flow characteristics; Acquiring welding area temperature field data by an infrared thermal imaging sensor, wherein the welding area temperature field data comprises molten pool center temperature, temperature gradient distribution and a heat affected zone range; Monitoring welding electrical parameter data by a current-voltage sensor, wherein the welding electrical parameter data comprises an instantaneous current value, a voltage fluctuation amplitude and an arc length change rate; And carrying out time synchronization processing on the molten pool surface image data, the welding area temperature field data and the welding electric parameter data to generate time-aligned multi-physical field sensing data.
3. The method according to claim 1, wherein the step of inputting the real-time multi-physical field sensing data into a simulation model for predicting the behavior of the molten bath and outputting predicted values of the state parameters of the molten bath within a predetermined time period in the future specifically comprises: carrying out noise reduction pretreatment on the real-time multi-physical-field sensing data to generate standardized sensing data, wherein the standardized sensing data is normalized data after abnormal values and noise interference are removed; extracting a time sequence feature vector of the standardized sensing data, wherein the time sequence feature vector comprises trend information and a periodic mode of dynamic change of a molten pool; inputting the time sequence feature vector into a molten pool behavior prediction model based on deep learning, wherein the molten pool behavior prediction model is a neural network model trained by historical welding data; And calculating the geometric parameters, the temperature distribution parameters and the flow state parameters of the molten pool in a preset time period in the future through the molten pool behavior prediction model, and generating a predicted value of the state parameters of the molten pool containing a confidence score.
4. The method of claim 1, further comprising, prior to the step of matching the composite oscillation track parameter corresponding to the predicted bath state parameter value from a predetermined database: Constructing a mapping relation library of welding process parameters and swing track parameters, wherein the mapping relation library comprises optimal track parameter combinations under different materials, plate thicknesses and welding positions; determining a track parameter search space according to a mapping relation library corresponding to the material attribute and the structural feature of the current welding task, wherein the track parameter search space is a parameter value range meeting the technological constraint condition; searching a history case closest to the predicted value of the molten pool state parameter in the track parameter search space by adopting a fuzzy matching algorithm, wherein the history case comprises a verified successful welding parameter combination; And generating a candidate composite swing track parameter set based on the historical case, and determining the composite swing track parameter from the candidate composite swing track parameter set according to a preset selection rule.
5. The method of claim 1, further comprising, after the step of controlling the gun tip to perform an asymmetric helical oscillating motion in the predetermined dwell region according to the microscopic modulation trajectory defined by the microscopic modulation trajectory parameters: Establishing a trajectory optimization feedback mechanism based on reinforcement learning, wherein the trajectory optimization feedback mechanism comprises a reward function design and strategy network updating; Acquiring weld joint forming quality indexes in the execution process of asymmetric spiral oscillation movement, wherein the weld joint forming quality indexes comprise weld joint residual height uniformity, undercut depth and surface ripple coefficient; Calculating a comprehensive rewarding value of the current swing period, wherein the comprehensive rewarding value is formed by weighting a weld forming quality index, a heat input uniformity index and an edge fusion state index; When the comprehensive rewarding value is lower than the historical average value, updating a control strategy of the microscopic modulation track through a strategy gradient algorithm to generate an optimized spiral oscillation parameter; And applying the optimized spiral oscillation parameters to the next swing period in real time.
6. The method of claim 1, further comprising, prior to the step of the welding gun entering a predetermined dwell region of the weld joint wobble edge according to a macro reference trajectory defined by the macro reference trajectory parameters: Determining a space boundary of the preset residence zone based on weld geometry and heat accumulation distribution characteristics, wherein the space boundary defines a key zone for executing spiral oscillation control; identifying high-risk position points in the swing track by analyzing defect distribution rules in historical welding data, wherein the high-risk position points are high-incidence areas with undercut and unfused defects; Calculating a critical heat accumulation radius of the swinging edge according to the current welding speed, the swinging amplitude and the thermophysical property of the material, wherein the critical heat accumulation radius is a distance threshold value for starting to generate overheat symptoms at the edge of the molten pool; Establishing a dynamic residence zone boundary function by taking the high-risk position point as a center and the critical heat accumulation radius as a reference, wherein the dynamic residence zone boundary function outputs a residence zone range which is adjusted in real time along with the welding process; the dwell region is divided into a core dwell region that performs a complete helical oscillating motion and a transition buffer that performs a smooth transition motion of gradual amplitude change.
7. A compound swing welding control system is characterized in that the system comprises one or more processors and a memory; the memory is coupled to the one or more processors, the memory for storing computer program code comprising computer instructions that the one or more processors invoke to cause the system to perform the method of any of claims 1-6.
8. A computer readable storage medium comprising instructions that, when run on a compound swing welding control system, cause the system to perform the method of any of claims 1-6.
9. A computer program product, characterized in that the computer program product, when run on a compound swing welding control system, causes the system to perform the method according to any of claims 1-6.

Description

Composite swing welding control method with self-adaptive variable parameters and related equipment Technical Field The application relates to the technical field of welding, in particular to a self-adaptive variable parameter compound swing welding control method and related equipment. Background The welding technology, in particular to automatic welding, is an indispensable key process link in the modern high-end manufacturing industry and is widely applied to the fields of aerospace, rail transit, pressure vessels, ocean engineering, energy equipment and the like. In these fields, the connection of medium plates is a common working condition, and the welding quality directly determines the service performance, reliability and safety life of the whole structural member. In order to achieve effective filling and fusion of a wider weld in a single or a small number of welding processes, a swing welding technique has been developed. By controlling the welding gun to periodically swing transversely in the direction perpendicular to the welding direction, the technology can obviously increase the melting width of a single-pass welding line, improve the production efficiency and improve the welding line forming. Therefore, how to precisely control the swing welding process, ensure that a dense, uniform, defect-free weld is obtained, is always the core focus of research and practice in this field. In the related art, a control system generally adopts an implementation manner based on offline programming and fixed parameters. Before the welding task starts, the process personnel preset a set of fixed welding process parameters according to the material, plate thickness, groove form and empirical data of the workpiece, wherein the set of parameters define the whole movement mode of the welding gun, such as constant swing frequency, fixed swing amplitude and fixed residence time at the edge positions at both sides of the welding seam. During the whole welding process, the welding robot or special machine drives the welding gun strictly according to the preset program, and completely repeated and periodical transverse swinging motion is performed from the starting point to the end point of the welding seam. The logical basis of this control strategy is to complete the entire welding task through a set of "one-time" actions, assuming that the welding conditions remain ideal and uniform throughout the process. However, welding is essentially a severe, dynamic, unbalanced thermophysical process. As welding continues, a significant amount of heat is continuously accumulated and conducted within the workpiece, resulting in significant dynamic changes in the temperature field of the weld and its vicinity. This cumulative effect of heat can destabilize the size, morphology and flow behavior of the bath. When the welding gun moves to the edge area of the swinging stroke, if the preset, symmetrical and fixed residence action is continuously executed, defects such as undercut, collapse and the like can be generated due to overheating and rapid increase of fluidity of molten pool metal caused by excessive concentration of local heat. Conversely, if the initial parameter settings are conservative, then there may be problems with edge unfused during the later stages of welding due to relatively insufficient heat input. Disclosure of Invention The application provides a self-adaptive variable parameter composite swing welding control method and related equipment, which are used for solving the problems that dynamic heat effect is difficult to adapt to a welding process due to the adoption of fixed swing parameters in the related technology, and welding defects are easy to generate at the swing edge. In a first aspect, the present application provides a method for controlling composite swing welding with adaptive variable parameters, applied to a composite swing welding control system, the method comprising: Acquiring real-time multi-physical field sensing data, wherein the real-time multi-physical field sensing data comprises at least two groups of heterogeneous sensor information for representing the geometric shape, temperature field distribution and arc stability of a molten pool; inputting the real-time multi-physical-field sensing data into a simulation model for predicting the molten pool behavior, and outputting a predicted value of a molten pool state parameter in a future preset time period; Matching composite swing track parameters corresponding to the predicted value of the molten pool state parameter from a preset database, wherein the composite swing track parameters comprise macroscopic reference track parameters and microscopic modulation track parameters for defining a composite motion track of a welding gun tip; When the welding gun enters a preset residence zone of the welding seam swinging edge according to the macroscopic reference track defined by the macroscopic reference track parameters, controlling the ti