CN-121972806-A - Laser welding parameter self-adaptive regulation and control system and method combined with OCT on-line monitoring

CN121972806ACN 121972806 ACN121972806 ACN 121972806ACN-121972806-A

Abstract

The invention relates to the technical field of laser welding and discloses a laser welding parameter self-adaptive regulation and control system and method combining OCT on-line monitoring, wherein the system consists of a high-power laser welding head, an OCT monitoring module which is strictly and coaxially integrated with an optical path of the high-power laser welding head, and a high-speed signal processing and controller, interference signals of a welding lock hole are acquired in real time through the OCT module, a longitudinal section image of the welding lock hole is reconstructed, the depth and the penetration of the lock hole are rapidly extracted as core feedback quantities, the controller utilizes a built-in dynamic response model and model predictive control algorithm based on the feedback, real-time rolling optimization and output of laser power and welding speed regulation instructions are realized, the closed-loop self-adaptive regulation and control of the penetration in the welding process can be effectively conducted on processing disturbance, the stability of the welding process and the consistency of the welding quality are improved, and intelligent precision welding is realized.

Inventors

NI WENLI
PENG BIAO
WU DI

Assignees

泰尔智慧(上海)激光科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260327

Claims (10)

1. Laser welding parameter self-adaptive regulation and control system combining OCT on-line monitoring, which is characterized by comprising: The laser welding head is arranged on the six-axis industrial robot, a scanning vibrating mirror and a focusing mirror are arranged in the laser welding head, a vibrating mirror scanning driver and a defocusing servo driver are respectively arranged on the scanning vibrating mirror and the focusing mirror, and a motion controller is arranged on the six-axis industrial robot and used for driving the laser welding head to move along X, Y, Z axes according to a welding path; The coaxial integrated OCT monitoring module is used for transmitting detection light into a welding lock hole formed by the processing laser beam and receiving back scattering light signals from a self-locking hole wall and a molten pool interface in real time, wherein a detection light path is strictly and coaxially integrated with the processing laser beam in the laser processing head through an internal common path light splitting element and is fixedly connected with the laser welding head into a whole; The input end of the high-speed signal processing and controller is connected with the data output end of the coaxial integrated OCT monitoring module, and the control output end of the high-speed signal processing and controller is connected to the power controller, the motion controller, the galvanometer scanning driver and the defocusing servo driver of the high-power laser welding head; wherein the high-speed signal processing and controller is configured to perform the following real-time closed-loop control loop: (a) Synchronously acquiring and processing the original interference spectrum signals acquired by the coaxial integrated OCT monitoring module at a sampling rate not lower than 1MS/s, and reconstructing the axial depth reflectivity distribution of the welding area; (b) According to preset scanning logic, assembling the continuously acquired axial depth reflectivity distribution into a two-dimensional section image which is perpendicular to the welding direction and comprises a keyhole depth structure; (c) Extracting a depth value of the bottom of the lock hole at the current moment from the two-dimensional section image in real time through an edge detection algorithm And a bath bottom depth value As a core feedback quantity; (d) The core feedback quantity and a preset depth target value are combined Compare and based on pre-established laser power And welding speed As input, a dynamic response model with predicted depth as output is adopted, and a model predictive control algorithm is adopted to perform rolling optimization to calculate the optimal laser power adjustment And a welding speed adjustment amount ; (E) The calculated optimal laser power adjustment amount And a welding speed adjustment amount And the adjustment instruction is issued to the high-power laser welding head in real time, and the continuous fiber laser and the six-axis industrial robot are driven to execute so as to realize dynamic correction of weld penetration.
2. The system for adaptively adjusting and controlling laser welding parameters in combination with OCT on-line monitoring according to claim 1, wherein the internal common-path beam splitting element is a dichroic mirror coated with an optical film that satisfies a transmittance of >95% for 1060 nm wavelengths and a reflectance of >99% for 1070nm wavelengths.
3. The adaptive control system for laser welding parameters in combination with online OCT monitoring according to claim 1, wherein the scanning logic used by the high-speed signal processing and control unit in performing the real-time closed-loop control cycle step (b) is to control the scanning galvanometer to laterally scan the OCT probe at a fixed offset position behind the processing laser beam to obtain the solidification interface image at the tail of the molten pool.
4. The adaptive control system for laser welding parameters in combination with OCT on-line monitoring according to claim 1, wherein the high-speed signal processing and controller, while executing the real-time closed-loop control cycle step (c), calculates the standard deviation of the keyhole bottom depth value within a preset time window And takes the stability as auxiliary feedback quantity for representing the stability of the welding process when When the preset first threshold value is exceeded, triggering a control strategy, and fine-adjusting the defocus amount through a defocus servo driver to stabilize the lock hole.
5. The adaptive regulation and control system for laser welding parameters in combination with OCT on-line monitoring according to claim 1, wherein the dynamic response model is a black box model based on data-driven training or a simplified physical model established based on energy balance and fluid dynamics principles and subjected to parameter identification.
6. The adaptive regulation and control system for laser welding parameters in combination with OCT on-line monitoring according to claim 1, wherein the model predicts the objective function of the control algorithm For minimizing the deviation of predicted penetration from target penetration and the variation of control quantity, the expression is: Wherein, the In order to predict the time domain of the signal, In order to control the time domain of the signal, And As a matrix of weights, the weight matrix, To include laser power adjustment And the welding speed adjustment amount The objective function is used for realizing multi-objective optimization of penetration tracking and control stability.
7. The laser welding parameter self-adaptive regulation and control method combined with OCT on-line monitoring is characterized in that the laser welding parameter self-adaptive regulation and control system combined with OCT on-line monitoring as claimed in any one of claims 1-6 is adopted, and comprises the following steps: S1, process initialization and model loading, namely setting a target penetration value Controlling algorithm parameters and loading a dynamic response model; S2, synchronously starting welding and monitoring, namely simultaneously starting the processing laser beam output of the high-power laser welding head and the sweep frequency light source of the coaxial integrated OCT monitoring module, and starting a real-time control thread of the high-speed signal processing and controller; S3, OCT signals are synchronously collected and depth reconstructed, namely, OCT interference spectrum signals are synchronously triggered and collected in each control period, and the axial depth reflectivity distribution of a welding area is reconstructed in real time; S4, welding state real-time sensing, namely constructing a two-dimensional cross-section image based on a plurality of continuous axial depth reflectivity distributions, and extracting the actual depth of the bottom of the lock hole at the current moment by utilizing an image processing algorithm And depth of bottom of molten pool ; S5, control decision generation, namely extracting the bottom depth of the molten pool And setting the target penetration value Deviation of (2) and depth of bottom of lock hole Is input to a model predictive controller, and an optimal laser power adjustment quantity sequence in a plurality of control cycles in the future is solved based on the dynamic response model in a rolling way And an optimal welding speed adjustment sequence ; S6, controlling instruction execution and feedback according to the laser power at the current moment And welding speed Superimposing the instantaneous adjustment in the optimal adjustment sequence And Calculating a control amount to be executed And (3) with And sending the data to a continuous fiber laser and a six-axis industrial robot for execution in real time; and S7, closed loop control, namely returning to the step S3 in the next control period until the welding stroke is finished, and forming closed loop control.
8. The method according to claim 7, wherein in step S4, the specific image processing algorithm is that the two-dimensional cross-sectional image is subjected to median filtering and morphological operation to suppress noise, then the adaptive threshold segmentation algorithm is adopted to distinguish the highly reflective metal vapor region, keyhole wall interface and molten pool interface, and finally the furthest point of the reflective interface in the depth direction is searched to determine And 。
9. The method according to claim 7, wherein in step S5, when the system detects that the workpiece joint gap or the surface is contaminated When mutation occurs, the model predictive controller calls a preset expert rule base, and superimposes a feedforward compensation amount on the basis of model calculation To speed up the response of the system.
10. The method for adaptively adjusting and controlling parameters of laser welding combined with OCT on-line monitoring according to claim 7, wherein the method further comprises the step S8 of self-learning, wherein after single-pass welding is completed, the system performs incremental update training on the dynamic response model to realize continuous performance optimization of a specific material and equipment combination by using time series data recorded in the welding process, including OCT depth data, control instruction data and macroscopic metallographic penetration data detected after welding.

Description

Laser welding parameter self-adaptive regulation and control system and method combined with OCT on-line monitoring Technical Field The invention relates to the technical field of laser welding, in particular to a laser welding parameter self-adaptive regulation and control system and method combined with OCT on-line monitoring. Background Laser deep penetration welding is a key process for realizing high-efficiency and high-quality connection, and is widely applied to the fields of aerospace, new energy automobiles, precise instruments and the like. The stability of the process is highly dependent on the dynamic behavior of the keyhole during the welding process. The keyhole is a steam cavity formed by metal vaporization under the action of laser, and the depth and stability of the keyhole directly determine the penetration and forming quality of the final welding seam and whether internal defects such as air holes, unfused and the like exist. However, the dynamic process of the inside of the keyhole and the molten pool is extremely complex, and is interfered by various factors such as material characteristics, assembly precision, heat input fluctuation and the like, and conventionally, the quality consistency under long-time and variable working conditions is difficult to ensure by an open-loop welding mode depending on preset process parameters. At present, the guarantee of welding quality in industry mainly depends on carrying out a large number of process tests before welding to optimize a parameter window and carrying out nondestructive detection such as X-ray, ultrasonic and the like on finished products after welding. The former is high in cost and cannot cope with real-time disturbance in production, and the latter belongs to post inspection, cannot intervene in the welding process, and possibly leads to batch waste. To enable on-line monitoring of the welding process, various sensing techniques are introduced, such as visual sensors to monitor the surface radiation of the weld puddle, acoustic sensors to collect welding sounds, spectrometers to analyze the plasma spectrum, etc. But the visual and spectral signals mainly reflect the plasma state on the surface or above the molten pool, the internal penetration information which is most critical to the quality of the welding line cannot be directly obtained, the acoustic signals are easily interfered by environmental noise, and the corresponding relation with specific defects is complex and difficult to be used for accurate quantitative control. In recent years, optical Coherence Tomography (OCT) technology has provided direction for welding process monitoring due to its high resolution, non-contact and depth-resolved imaging capabilities. OCT techniques can acquire cross-sectional images of the internal structure of a detected object, like performing an "optical ultrasound" scan. There have been studies attempting to apply it to welding, which have been initially demonstrated to be capable of detecting depth information of a keyhole. However, converting these studies into reliable closed loop control systems that can be truly applied to industrial sites, the following problems remain: The conventional OCT system has lag in data acquisition and processing speed compared with dynamic changes of millisecond or microsecond level of laser welding, is difficult to realize 'real-time' perception, and strong light, heat and smoke interference on a welding site form a serious challenge on stable transmission and signal quality of OCT detection light, meanwhile, most of the existing researches stay at the stage of observation and post analysis by utilizing OCT, lack of high-speed and accurate correlation between depth information acquired in real time and welding core parameters (such as laser power and welding speed), and form a complete technical scheme of millisecond-level response closed-loop control strategy. This has made OCT technology truly unable to take its potential for online measurement. Therefore, a set of OCT on-line monitoring and self-adaptive regulation integrated system which can be deeply integrated with a high-power laser welding head, can resist severe welding environment interference and can drive automatic adjustment of technological parameters based on real-time penetration information is developed, and the system has urgent practical requirements and important engineering values for breaking through the bottleneck of laser welding quality control and promoting intelligent manufacturing level improvement. Therefore, the application provides a laser welding parameter self-adaptive regulation and control system and method combined with OCT on-line monitoring. Disclosure of Invention Aiming at the defects of the prior art, the invention provides a laser welding parameter self-adaptive regulation and control system and method combined with OCT on-line monitoring so as to solve the background technical problem. In order to achieve the abo