Laser defilming control method and system based on dynamic environment compensation

Abstract

The present application relates to the technical field of automobile glass film removal, in particular to a laser film removal control method and system based on dynamic environment compensation. The steps of the control method include: constructing a multi-source vibration monitoring network at the key nodes of the galvanometer system base, linear guide driving unit and workpiece clamp of the laser film removal equipment, collecting real-time original vibration signal set containing acceleration, angular velocity and displacement through a three-axis acceleration sensor and a gyroscope group, and the sampling frequency is not less than 10 kHz. The present application realizes a breakthrough improvement in curved surface laser processing through a full-link compensation mechanism. Specifically, a three-axis sensor network is synchronously deployed at the motor flange, screw bearing seat and clamp interface, and combined with a differential anti-interference circuit, the coupling vibration characteristics of the mechanical transmission link are captured at the same time for the first time, laying a data foundation for accurate compensation.

Description

Technical Field

[0001] This invention relates to the field of automotive glass film removal technology, specifically to a laser film removal control method and system based on dynamic environmental compensation. Background Technology

[0002] In laser film removal of curved workpieces such as automotive windshields, equipment vibration and thermal deformation can cause the laser focus position to shift, affecting the film removal accuracy. While existing technologies employ vibration compensation methods, they suffer from fundamental flaws:

[0003] The sensor is only installed in a single position on the equipment base, which cannot accurately capture the coordinated vibration effect of motor drive, guide rail movement and workpiece clamping. For example, the vibration of the lead screw bearing and the loosening of the clamp will be superimposed.

[0004] Conventional filtering algorithms, such as Kalman filtering, have difficulty distinguishing between real equipment vibration and environmental noise, especially when high-frequency mechanical resonance is coupled with optical system errors.

[0005] The surface processing path planning and vibration compensation system operate independently, which can easily lead to a time lag between the processing trajectory and the focus compensation in areas where the glass curvature changes abruptly.

[0006] Based on the above-mentioned shortcomings, the technical problem proposed by this invention is: how to solve the integrated problem of accurate separation of multi-source vibration components, real-time compensation for focus drift and coordinated control of curved surface processing path under high dynamic working conditions. Summary of the Invention

[0007] This disclosure proposes a laser film removal control method and system based on dynamic environmental compensation, aiming to overcome at least one of the defects existing in the prior art.

[0008] To achieve the above objectives, the technical solution disclosed in this invention is as follows:

[0009] According to one aspect of this disclosure, a laser film removal control method based on dynamic environmental compensation is provided, the steps of which include:

[0010] A multi-source vibration monitoring network is constructed at key nodes of the galvanometer system base, linear guide drive unit, and workpiece fixture of the laser film removal equipment. The original vibration signal set containing acceleration, angular velocity, and displacement is collected in real time through a triaxial accelerometer and gyroscope group, with a sampling frequency of not less than 10kHz.

[0011] The original vibration signal set is input into a Bayesian filtering framework. Combined with the equipment dynamics transfer function and the environmental noise spectrum model, the effective vibration components are separated and a three-dimensional compensation vector is generated. The compensation vector includes vibration displacement, velocity and acceleration components.

[0012] Based on a three-dimensional curved surface model of an automobile windshield, an initial processing path is generated through a curvature adaptive algorithm. The path includes a theoretical focal coordinate sequence, feed rate, and laser energy parameters.

[0013] A spatial mapping relationship between the vibration compensation vector and the laser focus position is established. For each theoretical focus coordinate, the corrected coordinates are output in real time through a composite correction model. The model integrates the galvanometer response delay, optical refraction error and thermal deformation factors.

[0014] The corrected coordinates and processing parameters are sent to the laser controller, which drives the galvanometer and the feed system to achieve dynamic calibration of the focal position.

[0015] Furthermore, the key nodes include the motor stator flange face, the normal force-bearing surface of the ball screw bearing housing, and the junction of the hydraulic locking mechanism and the base plate. The multi-source vibration monitoring network includes multiple sets of sensors. The sensors of the galvanometer system base are installed on the motor stator flange face, the sensors of the linear guide are placed on the normal force-bearing surface of the ball screw bearing housing, and the sensors of the workpiece fixture are arranged at the junction of the hydraulic locking mechanism and the base plate. The signal transmission of the multiple sets of sensors adopts a differential amplifier circuit and is supplemented with an electromagnetic shielding layer.

[0016] Furthermore, the device dynamics transfer function is obtained through modal testing, and a second-order differential equation model containing the mass matrix M, damping matrix C, and stiffness matrix K is constructed in the 5-2000Hz frequency band. The environmental noise spectrum model is fitted with Gaussian distribution characteristics by continuously collecting data for 30 seconds under no-load conditions.

[0017] Furthermore, the execution process of the Bayesian filtering framework includes:

[0018] Define state vector ;

[0019] Constructing nonlinear observation equations:

[0020] Where H is the sensor configuration matrix, Φ(θ) is the equipment installation error transformation operator, and ▽P t ε represents the gradient transferred by the optical-mechanical coupling. t For adaptive noise terms;

[0021] The posterior probability density is obtained using the particle filter algorithm:

[0022] , where w t i This is a weight update function based on Mahalanobis distance;

[0023] We use Clamer-Rao lower bound to optimize state estimation of covariance.

[0024] Furthermore, the curvature adaptive algorithm includes: parameterizing the surface model using non-uniform rational B-spline (NURBS), calculating the curvature tensor eigenvalues ​​along the parameter direction, and automatically inserting transition path points when the rate of change of principal curvature exceeds a set threshold.

[0025] Furthermore, the process of establishing the spatial mapping relationship includes: calibrating the galvanometer deflection angle-focal displacement transformation matrix using a laser interferometer, obtaining the base vibration-focal offset transfer function through finite element analysis, and constructing a thermal deformation compensation curve through a temperature control experiment.

[0026] Furthermore, the composite correction model introduces a lead compensation stage, whose phase compensation amount ϕ(ω) satisfies:

[0027] , where γ lead γ represents the lead time constant obtained through galvanometer step response testing. lag ω represents the lag time constant obtained through optical path delay testing, and ω represents the angular frequency.

[0028] Furthermore, the driving galvanometer and the feed system achieve dynamic calibration of the focal position using a motion feedforward control strategy, generating a third-order Bezier transition curve between adjacent corrected coordinate points, and the position of the curve control point is dynamically adjusted according to the focal acceleration threshold.

[0029] Furthermore, the control method also includes collecting the background values ​​of environmental vibration in each degree of freedom direction when the equipment is unloaded, establishing a correlation function with temperature and humidity, and storing it as a compensation benchmark.

[0030] According to another aspect of this disclosure, a laser film removal control system based on dynamic environmental compensation is provided for executing the laser film removal control method based on dynamic environmental compensation as described above, the control system comprising:

[0031] The multi-source vibration monitoring unit includes a distributed array of inertial sensors and signal conditioning circuitry, used to capture equipment vibration and workpiece pose disturbance signals in real time in three-dimensional space.

[0032] The real-time compensation processor, with a built-in Bayesian filtering framework and composite correction model, is used to perform dynamic environmental error modeling and spatial mapping calculations.

[0033] The path planning module is configured to execute a curvature adaptive algorithm to generate laser processing trajectories resistant to vibration interference.

[0034] The laser controller receives the corrected focal coordinates and generates a galvanometer drive signal, which is used to convert the position compensation amount into an execution command.

[0035] The synchronization control bus is used to synchronize the clocks of each unit via the Time Sensitive Network (TSN) protocol.

[0036] The beneficial effects of this invention are:

[0037] This invention achieves a breakthrough improvement in curved surface laser processing through a full-link compensation mechanism. Specifically, a three-axis sensor network is simultaneously deployed at the motor flange (drive source), lead screw bearing housing (transmission core), and fixture interface (load end). Combined with a differential anti-interference circuit, it simultaneously captures the coupled vibration characteristics of each link in the mechanical transmission chain for the first time, laying a data foundation for accurate compensation. Furthermore, based on the equipment dynamics model and Bayesian filtering algorithm, the collected vibration signal is decomposed into effective equipment vibration components and environmental noise components, significantly improving the reliability of the compensation vector, such as accurately identifying micron-level displacement deviations caused by lead screw preload fluctuations. Further, for areas with abrupt changes in surface curvature, an adaptive algorithm generates a smooth transition path in real time, and a Time-Sensitive Network (TSN) ensures that the focus position correction is strictly synchronized with the processing path point, avoiding compensation lag caused by signal transmission delays in traditional systems. Furthermore, a composite correction model unifies the calculation of vibration compensation data and inherent equipment errors, such as galvanometer response delay, optical refraction distortion, and thermal deformation effects, generating three-dimensional focus correction coordinates that can directly drive the laser controller, eliminating the superimposed errors caused by multiple factors at the source.

[0038] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, the following describes the preferred embodiments of the present invention in detail with reference to the accompanying drawings. Attached Figure Description

[0039] Figure 1 This is a flowchart of a laser film removal control method based on dynamic environmental compensation in one embodiment of the present invention;

[0040] Figure 2 This is a schematic diagram of the vibration spectrum distribution of the device in one embodiment of the present invention;

[0041] Figure 3 This is a schematic diagram of the vibration acceleration response in one embodiment of the present invention;

[0042] Figure 4 This is a schematic diagram illustrating the separation effect of the original vibration signal and Bayesian filtering in one embodiment of the present invention;

[0043] Figure 5 This is a schematic diagram comparing the dynamic calibration effect of the focal position in one embodiment of the present invention;

[0044] Figure 6 This is a schematic diagram of the spatial mapping between vibration compensation and focus offset in one embodiment of the present invention;

[0045] Figure 7 This is a schematic diagram illustrating the phase lead correction effect of the galvanometer system in one embodiment of the present invention;

[0046] Figure 8 This is an architecture diagram of a laser film removal control system based on dynamic environmental compensation in one embodiment of the present invention. Detailed Implementation

[0047] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0048] In embodiments of the present invention, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" or "for example" in embodiments of the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Rather, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0049] The present invention provides the following preferred embodiments:

[0050] Example 1

[0051] To address the technical problem of focal point shift caused by environmental vibration during laser film removal, this embodiment provides a laser film removal control method based on dynamic environmental compensation, refining the multi-source vibration monitoring and dynamic compensation control process, such as... Figure 1 As shown, the control flow is as follows:

[0052] S100: A multi-source vibration monitoring network is constructed at key nodes of the galvanometer system base, linear guide drive unit, and workpiece fixture of the laser film removal equipment. The network collects raw vibration signal sets containing acceleration, angular velocity, and displacement in real time through a triaxial accelerometer and gyroscope group, with a sampling frequency of not less than 10kHz.

[0053] S200: Input the original vibration signal set into the Bayesian filtering framework, combine the equipment dynamics transfer function and the environmental noise spectrum model, separate the effective vibration components and generate a three-dimensional compensation vector. The compensation vector includes vibration displacement, velocity and acceleration components.

[0054] S300: Based on a three-dimensional curved surface model of an automotive windshield, an initial machining path is generated through a curvature adaptive algorithm. The path includes a theoretical focal coordinate sequence, feed rate, and laser energy parameters.

[0055] S400: Establishes a spatial mapping relationship between the vibration compensation vector and the laser focus position. For each theoretical focus coordinate, the corrected coordinates are output in real time through a composite correction model. The model integrates galvanometer response delay, optical refraction error and thermal deformation factors.

[0056] S500: Sends the corrected coordinates and processing parameters to the laser controller, driving the galvanometer and feed system to achieve dynamic calibration of the focal position.

[0057] Specifically, a multi-source vibration monitoring network is constructed on the galvanometer base, linear guide unit, and workpiece fixture, deploying a triaxial accelerometer and gyroscope array to acquire raw vibration signal sets at a sampling frequency of 10kHz. This signal set includes triaxial acceleration, angular velocity, and displacement components. For example... Figure 2 As shown, key frequency bands such as 80Hz rigid resonance and 220Hz motor vibration were identified in the vibration spectrum distribution of the equipment, which need to be compensated in the control system.

[0058] Furthermore, the collected raw vibration signals are input into a Bayesian filtering framework for signal separation. This is achieved through the following functional relationships: a dynamic transfer function, characterizing the frequency response of the equipment's mechanical structure, with the raw vibration spectrum as input and the system's natural vibration modes as output; an environmental noise spectrum model, quantifying the statistical characteristics of broadband noise in the workshop environment, with the vibration energy spectral density as input; and filtered outputs separated effective vibration components (main equipment vibration) and environmental noise components. For example... Figure 4 As shown, the original vibration signal is separated into the equipment vibration component (represented by the red curve) and the environmental noise component (represented by the blue dashed line) by Bayesian filtering. The final generated three-dimensional compensation vector is in the form of (Δx, Δy, Δt), where Δx and Δy are the displacement compensation amounts in the XY plane, and Δt is the time-series compensation delay.

[0059] Furthermore, based on the curved surface model of an automotive windshield, an initial processing path is generated using a curvature adaptive algorithm. This path consists of a discretized focus coordinate sequence {P}. i =(x i ,y i ,z i The structure is composed of z, where z i Here are the coordinates normal to the surface. The feed rate v at each point... i =κ·E max / ρ i Where κ is the material removal coefficient, E max ρ is the peak energy of the laser. i Let be the local radius of curvature.

[0060] Furthermore, a spatial mapping relationship is established between the vibration compensation vector and the focal position offset. The mapping function is defined as: Focal offset function: f(Δx,Δy)=α·(Δx) 2 +β·sin(Δy), where α is the optical distortion coefficient of the galvanometer and β is the thermal deformation sensitivity factor. For example... Figure 6As shown, the blue arrows represent the equipment vibration vector field, and the red scatter dots represent the focal offset positions generated by the mapping. The two are linked by a green dashed line, forming a nonlinear spatial relationship. The composite correction model calculates the theoretical focal point P in real time. i Corrected coordinates P i '=P i +f(Δx,Δy,Δt) synchronously compensates for mirror response delay and optical refraction error.

[0061] Furthermore, the corrected coordinate sequence {P} will be... i The energy parameters are sent to the laser controller. For example... Figure 5 As shown, the red dashed line represents the uncompensated vibration path, and the green solid line represents the corrected path. It is evident that the focal position deviation has significantly converged. Figure 7 As shown, the galvanometer system improves the control bandwidth through phase lead correction, enhances phase response characteristics in the 30-200Hz frequency band, and ensures real-time dynamic calibration.

[0062] The advantage of this embodiment is that it establishes a complete vibration sensing-separation-mapping-compensation chain, accurately distinguishes equipment vibration from environmental noise through a Bayesian framework, and achieves sub-millimeter-level focal stability by combining a spatial mapping model, thus meeting the high-precision film removal requirements of automotive curved glass.

[0063] Example 2

[0064] To address the issue of vibration signal coupling interference in equipment, this embodiment refines the deployment scheme of the multi-source vibration monitoring network at key nodes. The sensors on the motor stator flange are directly mounted on the outer edge of the stator winding support structure to capture lateral vibrations caused by electromagnetic torque pulsations. A triaxial accelerometer is arranged along the guide rail axis on the normal force-bearing surface of the ball screw bearing housing to simultaneously acquire the radial vibration component caused by bearing preload fluctuations. Sensors at the junction of the hydraulic locking mechanism and the base plate are arranged in a ring array, covering a 120° phase angle around the locking cylinder, used to calculate torsional vibrations caused by non-uniform clamping force distribution. Each sensor group achieves common-mode noise suppression via a differential amplifier circuit. The transmission cable uses a double-layer copper mesh shielding structure, with the inner shield grounded to the sensor body and the outer shield grounded at a single point to the signal acquisition end, forming a Faraday cage-like electromagnetic isolation.

[0065] Furthermore, a vibration signal fusion analytical model is constructed to separate environmental noise from equipment vibration components:

[0066] Where Λ(ω) is the nodal vibration spectrum matrix; Φ is the sensor position topology matrix; H(ω) is the structural transfer function tensor; and Υ(ω) is the environmental coupling factor matrix. It is important to understand that the position topology matrix is ​​obtained by calibrating the mechanical impedance characteristics of each mounting surface using a laser interferometer, ensuring that the sensitivity coefficient is controlled within the range of [0.85, 1.15] to guarantee sensitivity consistency.

[0067] The signal transmission process employs adaptive impedance matching technology, and an adjustable capacitor network is designed at the output of the differential amplifier circuit.

[0068] , where C adj For the dynamic matching capacitance value (unit: F), f c Z0 is the current signal center frequency (Hz) and Z0 is the characteristic impedance of the transmission line (Ω). This structure can suppress high-frequency signal attenuation and improve the frequency response flatness to ±1.5dB (10Hz-5kHz).

[0069] The advantages of this embodiment are: the differential amplification and double-layer shielding architecture can optimize the electromagnetic interference signal-to-noise ratio to below -110dB; the ring array arrangement of the hydraulic locking surface can resolve micro-torque fluctuations on the order of 0.05N·m; and the adaptive impedance matching technology ensures that the signal distortion rate is below 0.8% within a 200m transmission distance.

[0070] Example 3

[0071] To establish an accurate vibration compensation benchmark, this embodiment optimizes the method for constructing the equipment dynamic transfer function and the environmental noise spectrum. Within the 5-2000Hz frequency band, the structural frequency response function is obtained through impact modal testing, and a state-space model is constructed using polynomial matrix fitting technology.

[0072] In this model, X(t) is the system state vector, containing third-order components of displacement, velocity, and acceleration; Γ is the state transition matrix, whose singular value distribution characterizes the system's damping properties; Ω is the input coupling matrix; Ξ is the output observation matrix; and U(t) is the excitation vector. It's important to understand that the dimension of this state-space model is determined to be 12th order using the AIC criterion (Akaike Information Criterion), which effectively characterizes key resonance points such as 80Hz and 220Hz. (See figure...) Figure 2 The spectrum data shown.

[0073] Furthermore, the environmental noise spectrum model is established through the following steps:

[0074] 1. Under no-load conditions, continuously collect vibration data for 30 seconds at a sampling rate of 32kHz.

[0075] 2. Calculate the acceleration power spectral density function:

[0076] Where T is the sampling duration (30 seconds) and a(t) is the time-domain acceleration signal;

[0077] 3. The power spectrum was fitted using a Gaussian mixture model:

[0078] , where α k μ represents the weighting coefficient of the k-th Gaussian component. k For the characteristic frequency, σ k This is the bandwidth factor.

[0079] Furthermore, to separate the base vibration from the equipment dynamic response, a coupling factor is defined:

[0080] , where Γ full Let Γ be the overall system state matrix. isolated Let κ represent the device state matrix after decoupling of the vibration isolation system, and Ⅱ˙ⅡF denote the Frobenius norm. Active vibration isolation compensation is initiated when κ > 0.15, and this threshold corresponds to... Figure 5 The critical value for the center focus offset exceeding the tolerance.

[0081] In this embodiment: the state-space model has a prediction error of ≤1.2Hz for the 80Hz and 220Hz resonant frequencies; the Gaussian mixture model can accurately characterize the typical environmental noise spectrum of 20-150Hz; such as Figure 6 The spatial mapping logic shown uses a coupling factor κ to quantify the intensity of dynamic disturbances at the base, providing an analytical basis for mapping vibration compensation quantities.

[0082] Example 4

[0083] To address the limitations in laser focus state estimation accuracy and the coupling effects of installation errors in complex dynamic environments, this embodiment optimizes the state vector definition and nonlinear observation equation construction process based on the Bayesian filtering framework, and provides implementation details of the particle filter algorithm.

[0084] Specifically, the execution process of the Bayesian filtering framework includes:

[0085] Define state vector ;

[0086] Constructing nonlinear observation equations:

[0087] Where H is the sensor configuration matrix, Φ(θ) is the equipment installation error transformation operator, and ▽P t ε represents the gradient transferred by the optical-mechanical coupling. t For adaptive noise terms;

[0088] The posterior probability density is obtained using the particle filter algorithm:

[0089] , where w t i This is a weight update function based on Mahalanobis distance;

[0090] We use Clamer-Rao lower bound to optimize state estimation of covariance.

[0091] This embodiment first defines a state vector containing key dynamic parameters. This state vector encompasses the position coordinate components of the laser focus in three-dimensional space, its velocity component represented by its first derivative, and its acceleration component represented by its second derivative. It also includes two components describing the focus's attitude angle. This state vector provides a basic framework for comprehensively characterizing the dynamic behavior of the focus in space.

[0092] Furthermore, a nonlinear observation equation integrating multi-source information was constructed. The core of this equation lies in explicitly expressing the mathematical relationship between actual sensor measurements, inherent equipment installation errors on the observation results, the gradient effect caused by optical-mechanical system coupling, and the system's inherent adaptive noise term. Specifically, the observation equation integrates sensor configuration information through a sensor configuration matrix; errors such as angular offsets generated during equipment installation are modeled as equipment installation error transformation operators to correct measurement deviations introduced by improper installation; the physical coupling relationship between the optical path transmission path and mechanical structure deformation is quantified as the optical-mechanical coupling transfer gradient; in addition, an adaptive noise term is used to capture random disturbances that are difficult to model, such as environmental fluctuations.

[0093] Understandably, to achieve high-precision estimation of the aforementioned state vector under conditions of strong nonlinearity and non-Gaussian noise, this embodiment employs a particle filtering algorithm based on random sampling approximation to solve for the posterior probability density. The algorithm's implementation involves generating a large number of sample points, or particles, extracted according to their importance distribution. Each particle represents a possible state in the system's state space. After propagating and predicting each particle using the observation equation, the importance weights of all particles are updated and evaluated based on the latest sensor measurements and a weight update function calculated using Mahalanobis distance. The Mahalanobis distance here measures the difference between the predicted and actual observations, while also considering the structure of the observation noise covariance, making the weight update more statistically reasonable. The normalized weights represent the contribution of the corresponding particle to the true posterior probability density. Finally, the posterior probability density is represented as a weighted sum of all particle states, discretely approximating the distribution of the system state given all historical observation data.

[0094] Furthermore, to optimize the statistical efficiency of state estimation and approach the theoretical optimal estimation bound, this embodiment introduces the theoretical guidance of the Cramero lower bound, applied to the optimization process of the state estimation covariance matrix. This step aims to evaluate the statistical effectiveness of the current particle filter design and guide the direction of model parameter adjustment or algorithm improvement through theoretical analysis, so as to get as close as possible to the theoretical optimal estimation accuracy.

[0095] The benefits of this embodiment are as follows: it comprehensively characterizes the dynamics of the focal point through a precisely defined state vector; it effectively separates the sources of system error and improves the model fidelity by constructing a nonlinear observation equation that includes an installation error transformation operator and an optical-mechanical coupling gradient; it enhances the state estimation capability under complex disturbances and nonlinear systems by employing a particle filtering algorithm based on Mahalanobis distance; and it provides a basis for improving the theoretical performance boundary of the entire state estimation system by using the Cramerlow lower bound for covariance optimization.

[0096] Example 5

[0097] To address the lack of smoothness and adaptability in path planning when laser processing complex curved surfaces, and to avoid problems such as decreased processing quality or impact on focal point movement caused by abrupt curvature changes, this embodiment refines the parameterization process of non-uniform rational B-splines for the curved surface model and the dynamic path point insertion strategy based on curvature tensor analysis.

[0098] Specifically, a non-uniform rational B-spline model is used for parametric modeling of the surface to be processed. This method can flexibly and accurately represent the shape of freeform surfaces and transform the three-dimensional surface into a continuous mapping relationship dependent on two parameter directions. This parametric model provides the basic data structure for subsequent calculations.

[0099] Furthermore, based on the parametric properties of the NURBS model, the curvature tensor of points on the surface is systematically calculated along the parameter directions on the predetermined machining trajectory path. This calculation process involves solving for the fundamental form coefficients and second fundamental form coefficients of the surface at that point, thereby determining the components of the curvature tensor. Eigenvalue decomposition is then performed on the obtained curvature tensor to extract its principal curvature eigenvalues. These principal curvatures objectively describe the degree of curvature of the surface along the principal directions at that point.

[0100] It is important to understand that curvature itself is a core geometric quantity characterizing the local shape of a surface. The rate of change of principal curvature reflects the drastic change or abrupt change in the trend of the surface curvature along a predetermined path. This embodiment sets a threshold to determine whether such a change exceeds the acceptable range for smooth processing. When the rate of change of principal curvature at a point along the calculation path is detected to be greater than the set threshold, this indicates that the surface shape has undergone a relatively abrupt change at that point. To ensure that the focal trajectory can still smoothly transition in this region and adapt to the abrupt change in surface shape, the system automatically triggers a transition path point insertion mechanism. Specifically, near the abrupt change region, one or more additional transition path points are dynamically inserted between the originally planned discrete path points, based on the specific shape and degree of curvature change. The coordinate positions of these new points are generated by analyzing the continuous change law of the surface shape in the abrupt change region and combining it with focal dynamic constraints.

[0101] Furthermore, the specific location and number of inserted transition path points can be configured and adjusted according to the surface geometry and process requirements. Ultimately, all path points constitute a spatial trajectory sequence with smoother curvature changes and more continuous focal acceleration, such as... Figure 3 As shown, safe operating areas and recommendations for avoiding areas of peak vibration are provided.

[0102] This embodiment achieves adaptive and high-precision laser processing trajectory generation for highly complex curved surfaces. Its advantages are mainly reflected in: the use of the NURBS model ensures the accuracy and flexibility of surface representation; by calculating the curvature tensor eigenvalues ​​along the parameter direction and monitoring their rate of change, it accurately captures abrupt changes in surface shape; and the automated insertion of transition path points based on a set threshold significantly improves the smoothness and geometric adaptability of the focal trajectory on complex curved surfaces, effectively avoiding processing defects and equipment impact caused by drastic curvature changes.

[0103] Example 6

[0104] To address the issue of laser focus position shift caused by the coupling effect of multiple physical factors in actual operating environments, such as galvanometer deflection dynamics, base vibration transmission, and thermal deformation, this embodiment refines the specific calibration method for establishing the spatial mapping relationship between galvanometer deflection-displacement conversion, base vibration-focus shift transmission, and thermal deformation compensation.

[0105] Furthermore, this embodiment mainly includes three independent calibration steps, which together construct a comprehensive mathematical model of focal position disturbance and environmental impact.

[0106] The first step focuses on the precise mapping between the galvanometer's angle deflection command and its actual resulting focal position movement. A high-precision laser interferometer is used as the spatial reference measurement tool. By precisely controlling the galvanometer system to generate a series of discrete and uniform known angle deflection command sequences along the two scanning axes, the laser interferometer simultaneously measures the actual three-dimensional position offset of the corresponding laser focal point in the workpiece surface coordinate system under each deflection command. After recording and organizing all the angle command and corresponding measured displacement data points, a matrix fitting algorithm based on the least squares principle is applied for processing and analysis. Finally, a transformation matrix expressing the linear mapping relationship between the galvanometer deflection angle and the focal position coordinate offset is established, namely, the galvanometer deflection angle-focal displacement transformation matrix.

[0107] The second stage focuses on quantifying the dynamic process of how vibrations generated by external or internal excitations on the mechanical base are transmitted through the equipment structure and ultimately cause a shift in the laser focus position. This stage employs the finite element method to construct a physical model. First, a three-dimensional finite element mesh discretization model is created for the complete optomechanical structure, including the laser processing head, galvanometer system, and support structure, accurately assigning material physical properties to each component. Dynamic response calculations are then performed through modal analysis and / or by applying measured or simulated environmental vibration spectrum boundary conditions. The focus is on analyzing the frequency response characteristics of the displacement in each degree of freedom direction at the focus during the transmission of the vibration excitation signal input from the connection point between the base and the equipment to the laser focus positioning module within the target vibration frequency range. These frequency response functions are extracted and integrated into a focus shift transfer function model describing the vibration transmission effect.

[0108] The third step specifically addresses the systematic drift of the laser focus position caused by thermal deformation of the equipment due to temperature changes. In a specific thermal control experimental environment, such as a temperature-controlled chamber, a series of stable and uniform temperature gradient change points are set. Under each set stable temperature field condition, ensuring no actual processing load on the equipment, the laser outputs a reference optical path. A high-precision sensor measures and records the constant offset of the laser focus relative to the standard temperature reference position caused by the thermal expansion of the equipment at that temperature. A large number of temperature-focus offset data points covering the expected operating temperature range are collected. Subsequently, a curve fitting algorithm is applied to construct a thermal deformation compensation curve function library describing the relationship between temperature values ​​and the offset required for focus position compensation. Each curve typically corresponds to the thermal expansion characteristics of a specific material component or direction.

[0109] The benefits of this embodiment are as follows: through systematic calibration experiments, such as laser interferometer calibration, finite element dynamic simulation, and temperature-controlled environment measurement, a clear spatial mapping function relationship between three main environmental disturbance factors and the laser focus position offset was accurately established, laying a key data foundation and methodological support for the subsequent integration of these environmental disturbance models for real-time dynamic compensation.

[0110] Example 7

[0111] To improve the real-time control bandwidth and phase tracking accuracy of the laser focus position when dealing with complex dynamic disturbances such as vibration and thermal deformation, and to solve the problem of lag in conventional feedback control response, this embodiment refines the specific implementation method of introducing a lead correction element into the composite correction model and the method of obtaining the relevant time constant.

[0112] Specifically, the composite compensation model introduces a lead compensation element, whose phase compensation amount ϕ(ω) satisfies:

[0113] , where γ lead γ represents the lead time constant obtained through galvanometer step response testing. lag ω represents the lag time constant obtained through optical path delay testing, and ω represents the angular frequency.

[0114] The enhanced feature of this embodiment is to add a compensation stage with phase lead function to the existing compensation control law based on the aforementioned spatial mapping relationship. The ultimate goal of this lead correction stage is to improve the phase margin of the entire focus control loop, thereby offsetting the inherent hysteresis effect in the system and significantly improving the system's tracking capability and response speed to high-frequency disturbances or rapid command changes.

[0115] Understandably, the core of phase lead compensation lies in its ability to provide a positive phase increment within a specific frequency range. This increment is mathematically precisely characterized by the formula used. A key parameter here represents the angular frequency corresponding to the maximum value of the phase lead amplitude; it is essentially a dynamic property description of the lead compensation element.

[0116] Furthermore, this embodiment specifically clarifies the exact sources of the two time constants constituting the phase compensation function, ensuring a strict correlation between the correction model and the physical characteristics of the device. The parameters in the time constants used to describe the leading dynamic component are not arbitrarily set, but determined through a specific step response test experiment for the laser galvanometer deflection actuator. In this test, a step-shaped deflection command signal is applied to the galvanometer, while the dynamic response process of the galvanometer reaching the commanded position is recorded at high speed, such as the angle change curve over time. By performing feature analysis on this step response data, such as observing the rise time, overshoot, or performing system identification, physical parameters reflecting the inherent dynamic response speed of the galvanometer actuator can be extracted; this is the basis for the source of the time constants.

[0117] Furthermore, another time constant is related to the inherent physical delay of the optical path system. This delay originates from the combined effects of factors such as the optical path propagation time of the laser beam from generation to deflection by the galvanometer and transmission to the workpiece surface, the detector signal processing delay, and the control loop calculation delay. This time constant is obtained through specialized optical path delay testing experiments. For example, an ultrafast optical pulse emission and high-precision timing system can be used to measure the round-trip time of the signal from the emission of the laser command to the reception of the focal position change signal on the target surface; or the equivalent delay time constant can be identified by analyzing the phase lag frequency response curves of the system under sinusoidal command inputs of different frequencies.

[0118] Furthermore, the two time constant values ​​obtained according to the above testing method are substituted into the phase compensation function to construct a lead correction link that perfectly matches the physical dynamic characteristics of the specific laser processing equipment. The output signal of this link is superimposed or fused with the real-time predictive compensation signal based on spatial mapping relationship to jointly generate the final feedforward / feedback integrated control command that drives the galvanometer actuator and the feed system.

[0119] This embodiment integrates a physically measurable lead element into the composite correction model, effectively offsetting the limitations of device optical-mechanical delay and galvanometer dynamic response on system performance. It enhances the response bandwidth and phase margin of the focus position control system, helping to maintain higher focus positioning accuracy and trajectory tracking stability in complex dynamic environments containing rapid disturbances.

[0120] Example 8

[0121] To address the challenges of maintaining the continuity and stability of laser focus motion between adjacent correction coordinate points during dynamic focus position correction, and to prevent impacts or velocity abrupt changes due to trajectory transitions, this embodiment refines the dynamic trajectory generation method and its control point adaptive mechanism that employs a motion feedforward control strategy combined with a third-order Bezier transition curve.

[0122] Specifically, during the execution of focus dynamic calibration control, after the control algorithm generates a series of discrete but time-ordered corrected position coordinates of the focus target based on the aforementioned state estimation and environmental compensation results, the focus needs to continuously move from the current point to these target points. To connect any two adjacent three-dimensional space corrected coordinates, the system does not simply use linear interpolation, but automatically generates a third-order Bézier space curve as a transition trajectory.

[0123] Furthermore, the mathematical properties of a third-order Bézier curve determine that it possesses C² continuity, meaning that position, velocity, and acceleration are continuous, meeting the requirements of high-precision and smooth motion. A third-order Bézier curve is defined by four control points: a start point, an end point, and two internal control points that determine the curve's shape. In this embodiment, the defined sequence of focus correction target points will serve as the starting and ending points of these Bézier curves.

[0124] It is important to understand that the key factor determining the smoothness and stability of the curve lies in the location of the two internal control points. To ensure that the motion of the focal point never exceeds the physical limits of its dynamic system under any circumstances, this embodiment sets the locations of these internal control points to be dynamically adjustable according to the system state. Based on the real-time calculated maximum allowable acceleration threshold of the focal point, the system performs constraint optimization calculations in the region near the straight line segment or spatial path formed by adjacent target points to determine the optimal locations of the two internal control points. Specifically, the objective of dynamic constraint optimization calculations is usually to make the acceleration / jerk of the generated curve as uniformly distributed as possible throughout the entire interval, ensuring that its maximum value never exceeds the set focal acceleration threshold, while simultaneously approximating the straight line path in space as closely as possible to shorten the trajectory length. This can be transformed into an objective function optimization problem that minimizes the peak value or integral value of a certain derivative of the curve.

[0125] Furthermore, for each pair of adjacent target points to be connected, the system calculates and updates the four control points of the Bézier curve in real time or in advance. Finally, the calculated smooth 3D Bézier curve is resolved into a high-frequency, continuous sequence of focal position setpoints, which is then input to the underlying motion servo control system for execution. In this way, even if the distance between correction points is short or the direction changes drastically, a physically achievable motion trajectory with smoothly varying acceleration can be generated.

[0126] The benefits of the dynamic trajectory smoothing method provided in this embodiment are as follows: by utilizing the inherent high continuity of the third-order Bézier curve, the natural smoothness of the transition between points in terms of focus position, velocity, and acceleration is ensured; the position of the internal control point is dynamically optimized based on the real-time physical constraints of the focus, effectively preventing overload impact or vibration of the focus motion caused by improper trajectory planning; and the motion stability and control accuracy of the focus under frequent corrections are improved.

[0127] Example 9

[0128] To address the issue of environmental vibration background drift caused by environmental factors and improve the long-term stability and environmental adaptability of vibration compensation benchmarks, this embodiment details the specific operational steps for collecting environmental vibration background spectra under no-load conditions and establishing a correlation model with temperature and humidity environmental parameters.

[0129] This embodiment emphasizes the systematic nature and environmental relevance of the compensation benchmark establishment. Under no-load conditions where the laser processing equipment is not performing any workpiece processing tasks, a dedicated vibration background data acquisition process is initiated. This process covers all physical degrees of freedom involved in the laser focus positioning system during equipment operation, such as translational X / Y / Z axes and rotational axes.

[0130] Furthermore, using high-precision micro-vibration measurement devices such as accelerometers or optical interferometry systems integrated into the equipment body, micro-vibration signal data are synchronously and continuously collected in each degree of freedom direction over a pre-set, sufficiently long time period. The raw vibration data undergoes necessary preprocessing and time-frequency domain analysis to extract core quantitative indicators characterizing the equipment's background vibration level, forming a dataset of multi-degree-of-freedom direction environmental vibration background values ​​associated with the acquisition event.

[0131] Simultaneously, during the same vibration data acquisition period, temperature and humidity sensors deployed at key locations on the equipment need to continuously and synchronously record time-varying data sequences of ambient temperature and humidity values. These sequences are then preprocessed and analyzed to extract representative sets of temperature and humidity environmental parameters corresponding to the vibration acquisition state.

[0132] Understandably, it is necessary to explore and establish a quantitative functional relationship between the acquired environmental vibration background values ​​and the recorded environmental temperature and humidity parameters. This typically requires conducting multiple independent no-load vibration background acquisition experiments under different temperature and humidity conditions to form a data sample library covering the expected operating environment range of the equipment. Then, using multiple regression analysis or other nonlinear system identification algorithms, with the recorded temperature and humidity parameters as input independent variables and the vibration background quantification index in each degree of freedom as the output dependent variable, one or more explicit or implicit mathematical function models describing how the vibration background value changes with temperature and humidity are trained and fitted, i.e., correlation functions.

[0133] Finally, the successfully trained and validated correlation function model, along with its parameters, is stored in the device's non-volatile memory, forming a dynamically updated compensation benchmark database. During actual operation, the device only needs to monitor the current temperature and humidity values ​​in real time. Using this correlation function model, the system can predict and calculate the "standard" environmental vibration background values ​​for each degree of freedom under the current environmental conditions, serving as the baseline reference values ​​for the real-time vibration compensation system. The system can then compare the measured vibration signal with the predicted background values ​​and extract the "effective" vibration interference signals that exceed the background environment for subsequent compensation.

[0134] The benefits of this embodiment are as follows: by systematically collecting raw environmental vibration data under no-load conditions, the influence of processing load is separated, and a purer background vibration characteristic of the equipment environment is obtained; by associating the background value with temperature and humidity parameters and storing it as a function library, the vibration compensation benchmark can be dynamically adjusted according to changes in the actual operating environment; the adaptability of the compensation benchmark to environmental changes is improved, and a mechanism is provided to offset long-term drift caused by environmental factors.

[0135] Example 10

[0136] To address the problem of dynamic compensation failure caused by asynchronous compensation signals and execution actions due to timing deviations in various units within a multi-subsystem collaborative laser focus control architecture, this embodiment provides a laser film removal control system based on dynamic environmental compensation. It further optimizes the global clock synchronization mechanism based on the Time-Sensitive Network (TSN) protocol and the multi-level bus architecture design. Combined with… Figure 8 The system architecture shown eliminates coordination errors between heterogeneous functional units through a unified time-domain reference.

[0137] like Figure 8 As shown, this embodiment deploys an independent TSN clock source in the control system architecture. This clock source distributes nanosecond-precision synchronous clock signals to six functional units—sensor acquisition, filtering, path planning, calibration calculation, laser control, and galvanometer driving—through the physical layer link. Each unit is equipped with a timestamp and buffer management module to ensure that it can strictly align with the global timescale reference when processing or transmitting data packets.

[0138] Furthermore, a hierarchical communication bus was constructed at the hardware topology level. The bottom layer employs a time-sensitive network backbone bus based on the IEEE 802.1AS standard. A time-aware shaper schedules Ethernet data frames, ensuring that critical messages such as vibration sensing data streams, trajectory correction commands, and laser trigger signals are transmitted deterministically within predetermined time windows. Point-to-point physical links are configured between the backbone bus and each functional unit to avoid cumulative latency jitter introduced by multi-hop transmissions.

[0139] It is important to understand that the spatially distributed inertial sensor array deployed in the multi-source vibration monitoring unit requires its raw sampling clock to be phase-locked with the system's master clock via the TSN synchronization protocol. When the sensor nodes acquire three-dimensional vibration signals and workpiece pose disturbance data, the time stamp information of each sampling point is marked in real time by the local synchronization clock. These timestamped raw data streams are filtered and amplified by the signal conditioning circuit, then encapsulated into Ethernet frames conforming to the TSN standard and injected into the bus according to the preset transmission priority.

[0140] Furthermore, the real-time compensation processor, as the core computing unit, uses its built-in Bayesian filtering framework to perform dynamic environmental error modeling calculations according to a unified clock beat after receiving the time-aligned vibration data stream. The time reference of the output compensation signal maintains strict causality with the upstream sensing data. Synchronously, the path planning module generates an anti-interference machining trajectory based on a unified system clock, and the path node sequence output by its curvature adaptive algorithm carries precise timestamps.

[0141] Understandably, when the laser controller receives the focus coordinate correction command from the correction calculation unit and the trajectory data from the path planning module, it needs to perform a timescale consistency check on both. The controller only generates the corresponding galvanometer deflection drive signal when the command and trajectory timestamps match. The galvanometer drive unit triggers the power amplifier circuit based on a synchronous clock, ensuring that the focus position correction action and the vibration acquisition time maintain a fixed phase relationship.

[0142] This embodiment provides nanosecond-level synchronization accuracy for heterogeneous functional units through the TSN protocol, eliminating the phase lag of compensation actions caused by the drift of the factor system clock; the data consistency verification mechanism based on timestamps ensures the temporal causality between instructions and execution; the hierarchical bus design takes into account the transmission requirements of high real-time signals and batch data, providing underlying timing guarantees for dynamic environment compensation.

[0143] Although the present invention has been specifically described above with reference to preferred embodiments, it should be understood that the present invention is not limited to the embodiments described above. Various modifications and variations can be made by those skilled in the art without departing from the spirit of the present invention, and such modifications and variations should fall within the scope defined by the appended claims and their equivalents.

Claims

1. A laser film removal control method based on dynamic environmental compensation, characterized in that, The control method includes the following steps: A multi-source vibration monitoring network is constructed at key nodes of the galvanometer system base, linear guide drive unit, and workpiece fixture of the laser film removal equipment. The original vibration signal set containing acceleration, angular velocity, and displacement is collected in real time through a triaxial accelerometer and gyroscope group, with a sampling frequency of not less than 10kHz. The original vibration signal set is input into a Bayesian filtering framework. Combined with the equipment dynamics transfer function and the environmental noise spectrum model, the effective vibration components are separated and a three-dimensional compensation vector is generated. The compensation vector includes vibration displacement, velocity and acceleration components. Based on a three-dimensional curved surface model of an automobile windshield, an initial processing path is generated through a curvature adaptive algorithm. The path includes a theoretical focal coordinate sequence, feed rate, and laser energy parameters. A spatial mapping relationship between the vibration compensation vector and the laser focus position is established. For each theoretical focus coordinate, the corrected coordinates are output in real time through a composite correction model. The model integrates the galvanometer response delay, optical refraction error and thermal deformation factors. The corrected coordinates and processing parameters are sent to the laser controller, which drives the galvanometer and the feed system to achieve dynamic calibration of the focal position.

2. The laser film removal control method based on dynamic environmental compensation as described in claim 1, characterized in that, The key nodes include the motor stator flange face, the normal force-bearing surface of the ball screw bearing housing, and the junction of the hydraulic locking mechanism and the base plate. The multi-source vibration monitoring network includes multiple sets of sensors. The sensors of the galvanometer system base are installed on the motor stator flange face, the sensors of the linear guide are placed on the normal force-bearing surface of the ball screw bearing housing, and the sensors of the workpiece fixture are arranged at the junction of the hydraulic locking mechanism and the base plate. The signal transmission of the multiple sets of sensors adopts a differential amplifier circuit and is supplemented with an electromagnetic shielding layer.

3. The laser film removal control method based on dynamic environmental compensation as described in claim 1, characterized in that, The device dynamics transfer function was obtained through modal testing. A second-order differential equation model containing a mass matrix M, a damping matrix C, and a stiffness matrix K was constructed in the 5-2000Hz frequency band. The environmental noise spectrum model was fitted with Gaussian distribution characteristics by continuously collecting data for 30 seconds under no-load conditions.

4. The laser film removal control method based on dynamic environmental compensation as described in claim 3, characterized in that, The execution process of the Bayesian filtering framework includes: Define state vector ; Constructing nonlinear observation equations: Where H is the sensor configuration matrix, Φ(θ) is the equipment installation error transformation operator, and ▽P t ε represents the gradient transferred by the optical-mechanical coupling. t For adaptive noise terms; The posterior probability density is obtained using the particle filter algorithm: , where w t i This is a weight update function based on Mahalanobis distance; We use Clamer-Rao lower bound to optimize state estimation of covariance.

5. The laser film removal control method based on dynamic environmental compensation as described in claim 1, characterized in that, The curvature adaptive algorithm includes: parameterizing the surface model with non-uniform rational B-splines, calculating the curvature tensor eigenvalues ​​along the parameter direction, and automatically inserting transition path points when the rate of change of principal curvature exceeds a set threshold.

6. The laser film removal control method based on dynamic environmental compensation as described in claim 1, characterized in that, The process of establishing the spatial mapping relationship includes: calibrating the transformation matrix of mirror deflection angle-focal displacement using a laser interferometer, obtaining the base vibration-focal offset transfer function through finite element analysis, and constructing a thermal deformation compensation curve through a temperature control experiment.

7. The laser film removal control method based on dynamic environmental compensation as described in claim 6, characterized in that, The composite correction model introduces a lead compensation stage, and its phase compensation amount ϕ(ω) satisfies: , where γ lead γ represents the lead time constant obtained through galvanometer step response testing. lag ω represents the lag time constant obtained through optical path delay testing, and ω represents the angular frequency.

8. The laser film removal control method based on dynamic environmental compensation as described in claim 1, characterized in that, The driving galvanometer and the feed system achieve dynamic calibration of the focal position using a motion feedforward control strategy, generating a third-order Bezier transition curve between adjacent corrected coordinate points. The position of the curve control point is dynamically adjusted according to the focal acceleration threshold.

9. The laser film removal control method based on dynamic environmental compensation as described in claim 1, characterized in that, The control method also includes collecting the background values ​​of environmental vibration in each degree of freedom direction when the equipment is unloaded, establishing a correlation function with temperature and humidity, and storing it as a compensation benchmark.

10. A laser film removal control system based on dynamic environmental compensation, used to execute the laser film removal control method based on dynamic environmental compensation as described in any one of claims 1-9, characterized in that, The control system includes: The multi-source vibration monitoring unit includes a distributed array of inertial sensors and signal conditioning circuitry, used to capture equipment vibration and workpiece pose disturbance signals in real time in three-dimensional space. The real-time compensation processor, with a built-in Bayesian filtering framework and composite correction model, is used to perform dynamic environmental error modeling and spatial mapping calculations. The path planning module is configured to execute a curvature adaptive algorithm to generate laser processing trajectories resistant to vibration interference. The laser controller receives the corrected focal coordinates and generates a galvanometer drive signal, which is used to convert the position compensation amount into an execution command. The synchronization control bus is used to synchronize the clocks of each unit via the Time Sensitive Network (TSN) protocol.

Images