High-precision laser welding device for massage chair iron stand

Abstract

The invention belongs to the technical field of laser welding automation, and discloses a high-precision laser welding device for a massage chair iron stand. The device comprises a laser welding execution unit, a vision measurement unit, a self-adaptive robust control unit and a man-machine interaction unit, aiming at the problems that the traditional welding quality is unstable and the anti-interference capability is poor, the core of the device is that the self-adaptive robust control unit carries out high-precision prediction through a welding process dynamic model module, and the self-adaptive robust control unit carries out real-time real-time monitoring. The robust tracking control module generates an anti-interference cooperative control instruction in real time and drives the execution unit to complete welding, precise closed-loop adjustment of the laser power, the welding speed and the track of the robot is achieved, the fluctuation range of the weld penetration depth and the weld width is reduced, the stability and consistency of the welding quality are remarkably improved, and the welding quality is improved. And the method has strong robustness for interference such as clamping errors and thermal deformation, and integrates an online quality evaluation early warning function based on deep learning.

Description

Technical Field

[0001] This invention belongs to the field of automated laser welding technology, and in particular relates to a high-precision laser welding device for massage chair frames. Background Technology

[0002] As a high-value-added home appliance, the welding quality of the internal load-bearing iron frame of a massage chair directly affects the product's safety, service life, and operating noise. Traditional welding methods (such as CO2 gas shielded welding) have problems such as high heat input, severe deformation, and unsightly welds. Laser welding, with its advantages of high energy density, small heat-affected zone, small deformation, and ease of automation, has become the preferred process for welding massage chair iron frames.

[0003] However, the following problems exist in achieving high-quality laser welding: (1) The welding process is highly nonlinear and coupled with multiple variables. Small fluctuations in parameters such as laser power, welding speed, and defocusing amount will significantly affect the weld penetration, weld width and forming quality. Traditional PID control is difficult to handle such complex coupling relationships and parameter tuning is difficult. (2) There are many uncertainties and external interferences. The massage chair frame is usually made of various specifications of pipes and stamped parts. There are errors in the workpiece clamping. The welding thermal cycle will cause structural thermal deformation. The surface condition of the material (oil stains, rust) will also change. These factors constitute continuous interference to the welding process, resulting in unstable welding quality. (3) The weld tracking accuracy is high. The iron frame structure is complex. The weld is mostly a spatial curve and there are changes in joint gap and misalignment. The welding system is required to have real-time, accurate trajectory tracking and adaptive capabilities. Existing laser welding systems mostly adopt the preset program "teach-reproduce" mode, which lacks the ability to actively suppress process dynamics and interference, making it difficult to meet the high consistency requirements in the mass production of massage chair iron frames. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of existing laser welding devices for massage chair frames, such as poor adaptability, weak anti-interference ability, and reliance on human experience. This invention provides a high-precision laser welding device with integrated adaptive robust control. By constructing a dynamic model of the welding process and designing a robust controller based on an interference observer, the system achieves adaptive optimization of welding process parameters and precise tracking of weld seams under uncertainty and interference, ensuring stable and consistent welding quality.

[0005] The technical solution adopted in this invention is as follows: a high-precision laser welding device for massage chair frames, comprising a laser welding execution unit, a vision measurement unit, an adaptive robust control unit, and a human-computer interaction unit; The adaptive robust control unit is the core of the system, including a welding process dynamic model module and a robust tracking control module. The welding process dynamic model module innovatively adopts a hybrid modeling method that integrates mechanism and data-driven approaches to accurately characterize the multivariable coupled nonlinear dynamics of laser welding. The robust tracking control module adopts an adaptive robust control algorithm based on a nonlinear disturbance observer, which can estimate and compensate for various disturbances in real time. Combined with an adaptive feedback law, it ensures the stability and tracking accuracy of the system under parameter perturbations and external disturbances. The vision measurement unit adopts active and passive vision fusion technology to provide high-precision feedback on weld position and molten pool status for the control system; The human-computer interaction unit integrates a deep learning-based online welding quality assessment module to achieve real-time monitoring and early warning of process quality.

[0006] As a preferred technical solution of this scheme, the hybrid modeling process of the welding process dynamic model module is as follows: the mechanism model skeleton is constructed with the energy balance equation and the molten pool dynamic equation, and the prediction residual of the mechanism model is rapidly learned and compensated by the online extreme learning machine network to form a high-precision prediction model that can adapt to changes in material properties.

[0007] As a preferred technical solution of this scheme, the working process of the robust tracking control module is as follows: design a nonlinear disturbance observer to estimate the lumped disturbance in real time, and use the estimated value as feedforward compensation; design a robust feedback controller and parameter adaptive law based on the Lyapunov method, adjust the control gain online, and together form a composite controller with strong robustness.

[0008] The beneficial effects of the present invention after adopting the above structure are as follows: (1) Significantly improved and stable welding quality: The dynamics of the welding process are accurately described by the mechanism and data fusion model, and the adaptive robust controller actively suppresses interference such as clamping error and thermal deformation, realizing precise closed-loop adjustment of key process parameters such as laser power and speed. Compared with traditional open-loop or PID control, the fluctuation range of weld penetration and weld width is reduced, effectively avoiding defects such as incomplete penetration and burn-through, and significantly improving the consistency and reliability of welded joints.

[0009] (2) It has strong anti-interference and adaptive capabilities: the feedforward compensation based on the nonlinear interference observer can quickly offset the influence of modelable interference, while the robust feedback and adaptive law can handle unmodeled dynamics and parameter changes; the system shows excellent adaptability to common working conditions such as slight changes in material thickness, uneven joint gaps, and slight surface contamination of massage chair iron frames, reducing the stringent requirements on workpiece prefabrication accuracy and environment in production.

[0010] (3) Achieving intelligent process monitoring and quality assurance: The integrated machine vision system not only provides high-precision weld tracking, but also performs online intelligent evaluation of molten pool images and weld formation through deep learning models. This system transforms "post-inspection" into "process prevention," and can issue early warnings in the early stages of defects, and even automatically fine-tune process parameters for compensation. It has constructed a complete quality closed-loop control system, laying a solid foundation for realizing intelligent and unmanned welding production of massage chair iron frames. Attached Figure Description

[0011] The accompanying drawings are provided to further understand the present invention and form part of the specification. They are used together with the embodiments of the present invention to explain the invention and do not constitute a limitation thereof.

[0012] Figure 1 This is a block diagram illustrating the core algorithm principle of the adaptive robust control unit of this invention. Detailed Implementation

[0013] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0014] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0015] Example 1: Combination Figure 1 The present invention provides a high-precision laser welding device for massage chair frames. The system includes a laser welding execution unit, a vision measurement unit, an adaptive robust control unit, and a human-computer interaction unit.

[0016] The laser welding execution unit includes a 2000W fiber laser, a six-axis high-precision industrial robot carrying the welding head, a water cooling system, and a shielding gas supply system. The welding head integrates a coaxial CCD camera, a cross-type laser structured light projector, a contact range sensor, and a shielding gas nozzle.

[0017] The vision measurement unit includes an active vision module (structured light projection + high-speed camera) positioned in front of the welding head and a coaxial vision module inside the welding head. The active vision module is used for three-dimensional trajectory tracking of the weld and measurement of joint morphology (gap, misalignment) before and during welding. The coaxial vision module is used to acquire images of the molten pool area in real time and extract the length, width, and tail flame characteristics of the molten pool.

[0018] The adaptive robust control unit, the intelligent core of the system, includes: (1) Welding process dynamic model module: Receives the molten pool characteristics, joint morphology and current process parameters (laser power P, welding speed V, defocusing amount D) measured by the vision unit. This module has a built-in hybrid model: the mechanism part is based on the three-dimensional heat conduction equation and the molten pool fluid model; the data-driven part is an online extreme learning machine network, which is used to compensate for the dynamics not covered by the mechanism model. The model output is the predicted melt depth Hp, melt width Wp and forming quality index Qp.

[0019] (2) Robust tracking control module: This module takes the deviation e between the model prediction output and the process setpoint, as well as the actual measured value from visual feedback, as inputs. First, the nonlinear disturbance observer estimates the lumped disturbance dhat caused by thermal deformation, material inhomogeneity, etc., in real time based on the system input and output. Then, the adaptive robust controller calculates the control law: u = u ff +u fb , where u ff u is the feedforward compensation amount based on disturbance estimation. fb The robust feedback control quantity is designed based on Lyapunov stability theory. Its parameters are updated online through an adaptive law, and the final output is the laser power adjustment quantity ΔP, the welding speed adjustment quantity ΔV, and the robot trajectory correction quantity ΔX.

[0020] The human-machine interaction unit, including an industrial computer and a touch screen, is used to set welding process specifications, display three-dimensional weld seam tracking paths, real-time process parameter curves, visual images of the molten pool, and quality assessment results. It integrates a quality assessment module based on convolutional neural networks to perform real-time analysis of coaxial camera images, label defect risks, and assign ratings.

[0021] Example 2: This embodiment, based on Embodiment 1, details the compensation mechanism of the online limit learning machine in the dynamic model module of the welding process.

[0022] Step M1: In each control cycle, obtain the predicted value (H) of the mechanistic model. m W m ) and the actual melt pool features extracted from visual units (H v W v The residuals between (e) H e W ).

[0023] Step M2: Use the current process parameters [P, V, D], the residual from the previous time step, and several texture feature vectors extracted from the molten pool image as input to the Extreme Learning Machine network.

[0024] Step M3: The single-hidden-layer neural network calculates the compensation amount (ΔH) at the current time using the aforementioned input. c ΔW c The network is characterized by randomly generated and fixed hidden layer node parameters, requiring only online updates to the output layer weights, resulting in extremely fast learning speed and suitability for real-time dynamic compensation.

[0025] Step M4: The final prediction output of the mixture model is: (H p W p ) = (H m +ΔH c W m +ΔW c This model can quickly adapt to minor changes in energy transmission caused by changes in protective gas, lens contamination, etc.

[0026] Example 3: This embodiment, based on Embodiment 1, details the design of the nonlinear disturbance observer and controller in the robust tracking control module.

[0027] Step R1 (Disturbance Observer Design): Define the nominal dynamic model of the system as x = f(x) + g(x)u, where x is the system state (e.g., molten pool characteristics) and u is the control input. Design the observer: ˙ξ = f(x) + g(x)u + l(x)(y - ˆy), d_hat = ξ - p(x), where ξ is the internal state of the observer, l(x) is the observer gain function, y is the actual measured output, ˆy is the observer output, and p(x) is the designed nonlinear function. This observer can asymptotically estimate the lumped disturbance d.

[0028] Step R2 (Adaptive Robust Controller Design): Define the tracking error e = x - x_d. Design the control law: u = u_nom + u_rob. u_nom is the equivalent control based on the nominal model and disturbance feedforward (-d_hat / g(x)). u_rob is the robust term, designed as u_rob = -(k + δ)e, where k > 0 is the fixed gain, and δ is the adaptive gain adjusted online by the adaptive law ˙δ = γ||e||^2 (γ > 0). This design ensures that the closed-loop system remains stable even with observation errors.

[0029] Step R3: Convert the calculated control quantity u into laser power setting value, robot speed command and trajectory offset command, and send them to the execution unit.

[0030] Example 4: This embodiment is based on Embodiment 1 and demonstrates the welding application of the system to the "C-shaped" main load-bearing iron frame of the massage chair.

[0031] The workpiece to be welded is a spatial frame structure made of rectangular tubes and stamped steel plates. The welds include straight lines, spatial curves, and butt joints of different plate thicknesses.

[0032] Test conditions: Target weld depth set at 2.0 mm, weld width at 3.0 mm. During the welding process, artificial interference was introduced: ① When welding halfway through, the workpiece was lightly tapped to simulate a misalignment caused by loose clamping (approximately 0.2 mm); ② A workpiece from the same batch but with slight rust was used instead.

[0033] Application effect: Anti-offset interference: When offset occurs, the active vision system detects the trajectory deviation within 5ms, and the robust control module coordinates the robot to generate trajectory compensation instructions within 50ms. At the same time, it fine-tunes the laser power to prevent insufficient penetration due to changes in defocusing amount. The weld is continuous throughout, without undercut or lack of fusion.

[0034] Adaptive material change: For rusty workpieces, coaxial vision detects initial anomalies in the brightness and oscillation frequency of the molten pool. Based on this, the process model predicts that the molten depth may be too shallow. The controller adaptively increases the laser power by about 5% to compensate, and the final weld formation is basically the same as that of rust-free workpieces.

[0035] Quality Assessment: The quality assessment module performs online analysis on all weld images, classifying them as "Level 1" welds with no defect warning. Compared with traditional PID-controlled welding samples, the standard deviation of weld penetration obtained by this invention is reduced from 0.15mm to 0.05mm.

[0036] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A high-precision laser welding device for massage chair frames, characterized in that: The system comprises a laser welding execution unit, a visual measurement unit and a human-computer interaction unit, the laser welding execution unit completes laser welding work on a massage chair iron frame; the visual measurement unit collects weld image and position information; the human-computer interaction unit is used for inputting welding process parameters and monitoring welding state; characterized in that the system further comprises a self-adaptive robust control unit; The self-adaptive robust control unit comprises a welding process dynamic model module and a robust tracking control module; The welding process dynamic model module constructs a multivariable coupling welding process model with laser power, welding speed and defocusing amount as inputs and weld penetration, weld width and forming quality as outputs according to real-time data of the visual measurement unit and the welding execution unit; The robust tracking control module generates anti-interference laser welding process parameter adjustment instructions in real time based on the output of the welding process dynamic model and actual measurement feedback by using an adaptive robust control algorithm based on an interference observer.

2. The high-precision laser welding device for the iron frame of a massage chair according to claim 1, characterized in that: The welding process dynamic model module adopts a hybrid modeling method based on physical mechanism and data driving fusion, specifically comprising: A mechanism model framework is established based on energy conservation and molten pool fluid dynamics, and historical welding data are used to identify and correct key time-varying parameters (such as material absorption rate and heat efficiency coefficient) in the model online.

3. The high-precision laser welding device for the iron frame of a massage chair according to claim 1, characterized in that: The robust tracking control module adopts an adaptive robust control algorithm based on a nonlinear interference observer, specifically comprising the following steps: Step R1: a nonlinear interference observer is designed to estimate complex interference (including clamping error, iron frame thermal deformation and material non-uniformity) in the welding process in real time; Step R2: based on Lyapunov stability theory, an adaptive robust controller is designed, the estimated value of the interference observer is used as a feedforward compensation, and an adaptive law is designed combining feedback error to adjust controller parameters online; Step R3: feedforward compensation and robust feedback are integrated to generate final control instructions that can suppress interference and ensure tracking accuracy.

4. The high-precision laser welding device for the iron frame of a massage chair according to claim 1, characterized in that: The visual measurement unit adopts a combination of active vision and passive vision, comprising: A high-resolution CCD camera is used to collect passive visible light images of the welding area; An auxiliary structured light projector is used to project laser lines or grids of specific patterns to the weld area; An image processing module fuses the two kinds of image information to calculate the three-dimensional position, gap, misalignment and molten pool geometric features of the weld in real time.

5. The high-precision laser welding device for the iron frame of a massage chair according to claim 1, characterized in that: The laser welding execution unit comprises a high-power fiber laser, a six-axis linkage industrial robot, a welding head and a follow-up device; the welding head integrates a coaxial vision sensor, a protective gas nozzle and a ranging sensor; the follow-up device can adjust the welding gun pose in real time to track the weld according to the feedback of the visual measurement unit.

6. The high-precision laser welding device for the iron frame of a massage chair according to claim 2, characterized in that: In the hybrid modeling method, the data driving part adopts an online extreme learning machine network, which takes the prediction residual of the mechanism model and the current process parameters as inputs, outputs the dynamic compensation amount for the next time weld forming, and superimposes the mechanism model output to form the final prediction output.

7. The high-precision laser welding device for the iron frame of a massage chair according to claim 3, characterized in that: The adaptive law described in step R2, whose update rate is associated with the norm of tracking error and the norm of disturbance estimation error, ensures that the closed-loop system can still achieve consistent ultimate bounded stability when there are modeling errors and unestimated disturbances.

8. The high-precision laser welding device for the iron frame of a massage chair according to claim 1, characterized in that: The human-computer interaction unit also includes a welding quality online evaluation module, which, based on the molten pool image and the weld forming image collected by the visual measurement unit, judges in real time whether the weld has defects such as pores, undercut and incomplete penetration through a trained convolutional neural network model, and performs quality grade classification and early warning.

Images