An automatic control method and control system for laser cutting

Abstract

This invention relates to the field of laser cutting technology, and in particular to an automatic control method and system for laser cutting. The automatic control method and system for laser cutting achieve fully automated control from data acquisition to final cutting effect evaluation through key technologies such as multi-source data fusion, intelligent path planning, vibration compensation and parameter optimization, which significantly improves the accuracy and efficiency of laser cutting.

Description

Technical Field

[0001] This invention relates to the field of laser cutting technology, and in particular to an automatic control method and control system for laser cutting. Background Technology

[0002] Metal laser cutting technology, as one of the core processes in modern precision machining, plays a crucial role in high-end manufacturing industries such as aerospace, automotive manufacturing, and electronic components due to its non-contact processing, high energy density, and excellent processing accuracy. However, as industrial applications demand increasingly higher cutting quality and efficiency, the limitations of existing technologies in practical applications are becoming increasingly apparent. Current laser cutting control suffers from the following shortcomings:

[0003] 1) Insufficient data acquisition accuracy and synchronization:

[0004] In traditional laser cutting systems, there may be problems with inaccurate data acquisition due to a single sensor or a lack of effective synchronization mechanisms between sensors.

[0005] 2) Lack of efficient multi-source data preprocessing and fusion technologies:

[0006] For data acquired from multiple sensors, existing systems may lack effective means for filtering and denoising, outlier removal, and spatiotemporal alignment of multi-source data. This may result in an inaccurate 3D model that fails to accurately reflect the geometric information, temperature distribution, and changes in cutting quality during the cutting process.

[0007] 3) The level of intelligence in path planning is not high:

[0008] Traditional laser cutting systems often rely on simple algorithms or manual settings for path planning, making it difficult to dynamically adjust the cutting path based on real-time geometric information, temperature field data, and cutting quality parameters to avoid surface defects and high-temperature zones, thus limiting cutting efficiency and quality.

[0009] 4) The vibration compensation mechanism is inadequate:

[0010] Laser cutting heads are subject to mechanical vibrations during operation. If the path deviation caused by these vibrations cannot be compensated for in a timely and effective manner, it will seriously affect cutting accuracy. Existing compensation schemes may be insufficient to cope with complex vibration modes, especially in high-speed cutting scenarios.

[0011] 5) Insufficient flexibility in parameter optimization and control:

[0012] The setting of key parameters such as cutting speed and laser power is usually based on experience or fixed rules, lacking a dynamic adjustment mechanism based on real-time feedback. Furthermore, methods for establishing a regression relationship between cutting speed and kerf width through experimental data analysis and for finding the optimal parameter combination using optimization algorithms are still underdeveloped. Summary of the Invention

[0013] This invention provides an automatic control method and control system for laser cutting. By integrating high-precision sensors, advanced data processing algorithms, intelligent path planning, vibration compensation mechanisms, and parameter optimization strategies, it provides a more efficient, accurate, and adaptable automatic control solution for laser cutting. This method not only improves the quality and efficiency of cutting but also enhances the automation and flexibility of the entire processing.

[0014] To achieve the objective of this invention, the technical solution adopted is: an automatic control method for laser cutting, comprising:

[0015] The S1 and 3D laser scanners generate three-dimensional point cloud data of the metal surface, the infrared thermal imager monitors the temperature field to obtain temperature field data, the vision camera captures the cutting scene to obtain visual image data of the cutting quality, and the encoder and inertial measurement unit monitor the position and vibration frequency of the cutting head to obtain multi-source data.

[0016] S2. Preprocess the multi-source data, map the preprocessed data to three-dimensional space, and generate a dynamic three-dimensional model containing geometric information, temperature distribution and cutting quality.

[0017] S3. Cutting path planning and obstacle avoidance based on dynamic 3D model; Combine dynamic 3D model to generate the optimal cutting path in real time and avoid surface defects and high temperature areas;

[0018] S4. Establish the regression relationship between cutting speed and kerf width, fit the coefficients through experimental data, define the objective function to evaluate the cutting effect, and use optimization algorithms to find the optimal cutting speed and laser power.

[0019] As an optimized solution of the present invention, in step S1, the 3D laser scanner is used to acquire three-dimensional point cloud data of the metal surface to be cut in real time, providing metal surface features, including undulations, holes and edge features, to help identify and locate key features on the cutting path; the 3D laser scanner is linked with the encoder to ensure that the scanning data is strictly synchronized with the position of the cutting head, and a high-speed processor module is added to achieve hardware-level trigger synchronization, unifying the clock signals of the 3D laser scanner, encoder and inertial measurement unit to μs-level accuracy; the vision camera is used to capture real-time cutting images and analyze the kerf quality, which includes the cutting width and slag residue, thereby evaluating the cutting effect; an encoder and an inertial measurement unit are also set to track the three-dimensional motion trajectory of the laser cutting head, thereby compensating for mechanical vibration errors.

[0020] As an optimized solution of the present invention, in step S2, the preprocessing includes filtering and denoising of the three-dimensional point cloud data, using Kalman filtering to eliminate measurement noise from the 3D laser scanner; mapping the preprocessed data to three-dimensional space to generate a dynamic three-dimensional model containing geometric information, temperature distribution, and cutting quality, specifically including:

[0021] 1) Spatiotemporal alignment of multi-source data; unify the coordinate system and time reference of 3D point cloud data, temperature field data, and cutting quality visual image data, and align multi-source data through calibration parameters to ensure spatial consistency;

[0022] 2) 3D geometric model reconstruction; In order to generate a continuous surface model from the filtered point cloud data, the Poisson surface reconstruction method combined with a dynamic update strategy is adopted; In the real-time point cloud data stream scenario, the surface model is dynamically updated to reflect the latest geometric changes. Incremental Delaunay triangulation is adopted to gradually update the mesh vertices V(t) and faces F(t).

[0023] 3) Temperature distribution mapping: Associate temperature field data with vertices / faces of the 3D geometric model, and use inverse distance weighted interpolation for areas that are not directly measured.

[0024] 4) Integration of cutting quality parameters; quantification of visual image data on cutting quality and mapping to a three-dimensional geometric model;

[0025] 5) Dynamic 3D model update, realizing the dynamic effect of model evolution over time, and updating the point cloud position x. t and temperature T t Perform a status update:

[0026] The state prediction equation is:

[0027] The observation equation is:

[0028] in: This is the state estimate at time t. A is the state transition matrix, describing how the system state transitions from the previous time t-1 to the current time t. B is the control input matrix, describing the control input u. t Impact on system state; w t Process noise represents the random error present during the system's state transition process; v t Observation noise represents the random error present in the sensor measurement process; H is the observation matrix, which maps the state variables to the observation space; z t This is the observed value at the current time t.

[0029] As an optimized solution of the present invention, the integration of cutting quality parameters and the quantification of cutting quality visual image data, and mapping to a three-dimensional geometric model, specifically include:

[0030] (1) Calculate the surface roughness C(p) using the point cloud curvature formula, and measure the surface smoothness by the difference in normal vectors;

[0031] Where: n(p) is the normal vector of point p, p i Its neighboring points;

[0032] (2) Establish the regression relationship between cutting speed v and kerf width W to predict and optimize cutting process parameters;

[0033] (3) Display cutting quality parameters on the geometric model by means of color coding or label overlay.

[0034] As an optimized solution of the present invention, step S3 specifically includes:

[0035] S31. The optimal cutting path is generated using the A* algorithm, with the following formula:

[0036]

[0037] Where: c(si, si+1): from node s i to s i+1 The cost function, where Path is the optimal path;

[0038] S32. Considering both cutting efficiency and cutting quality, the cutting efficiency is to minimize the total path length L. path Cutting quality is achieved by avoiding high-temperature and high-roughness areas; specifically, this is implemented as follows:

[0039] 31) Initialization; Define start and end points: Determine the start and end points based on the cutting task. Set the initial path: Use a straight line connecting the start and end points as the initial path;

[0040] 32) Path evaluation; cost function calculation: based on geometric distance, temperature constraints, and surface feature factors,

[0041] Calculate the total cost of the current path;

[0042] J = w1·L path +w2·J T +w3·J C

[0043] Where: w1, w2, w3: weighting coefficients, representing the degree of importance attached to path length, temperature constraints, and surface features, respectively;

[0044] 33) Use the A* algorithm to optimize the path step by step; in each iteration, select the path expansion direction with the minimum cost until the termination condition is met;

[0045] S33. Real-time correction of path deviation caused by mechanical vibration using encoder and inertial measurement unit measurement data: Vibration compensation model:

[0046]

[0047] a i : Vibration amplitude; ω i : Angular frequency of vibration; φ i : Phase angle; ΔP(t) represents the vibration offset of the laser cutting head in three-dimensional space.

[0048] As an optimized solution of the present invention, step S4 specifically includes:

[0049] S41. Establish the regression relationship between cutting speed v and seam width W: W=k1×v+k2×P+b;

[0050] Where: W is the kerf width; v is the cutting speed; P is the laser power; k1, k2, b are the fitting coefficients;

[0051] S42. Define an objective function to evaluate the cutting effect; J(v,P)=λ1·W+λ2·C(p)+λ3·T; where: J(v,P) is the objective function value, W is the kerf width, C(p) is the surface roughness, and T is the temperature distribution. λ1, λ2, and λ3 are weighting coefficients;

[0052] S43. Use particle swarm optimization algorithm to find the optimal cutting speed v and laser power P. During the cutting process, continuously monitor the changes in geometric information, temperature distribution and cutting quality parameters, and adjust the cutting speed v and laser power P in a timely manner.

[0053] [v * ,P * ] = arg min J(v,P); used to find the optimal cutting speed v and laser power P.

[0054] Finally, a PID controller is used to adjust the laser power and cutting speed in real time.

[0055] As an optimized solution of the present invention, in step S43, the particle swarm optimization algorithm is used to simulate the foraging behavior of bird flocks, and the global optimal solution is found through group cooperation. The steps are as follows:

[0056] 1) Initialize a group of particles, each particle representing a candidate solution (v, P);

[0057] 2) Calculate the objective function value J(v,P) for each particle;

[0058] 3) Update the velocity and position of each particle to move it toward a better solution;

[0059] 4) Repeat steps 2 and 3 until the termination condition is met. The termination condition includes reaching the maximum number of iterations or the objective function value no longer changing significantly.

[0060] To achieve the objectives of this invention, the technical solution adopted is: a control system for an automatic laser cutting control method, characterized by comprising a data acquisition module, a data preprocessing and modeling module, a path planning and obstacle avoidance module, a vibration compensation module, and a parameter optimization and control module. The data acquisition module generates three-dimensional point cloud data of the metal surface using a 3D laser scanner, obtains temperature field data by monitoring the temperature field with an infrared thermal imager, obtains visual image data of the cutting quality by capturing the cutting scene with a vision camera, and obtains multi-source data by monitoring the position and vibration frequency of the cutting head with an encoder and an inertial measurement unit. The data preprocessing and modeling module processes the multi-source data... The system performs preprocessing, mapping the preprocessed data to a three-dimensional space to generate a dynamic three-dimensional model containing geometric information, temperature distribution, and cutting quality. The path planning and obstacle avoidance module plans and avoids obstacles based on the dynamic three-dimensional model. Combining the dynamic three-dimensional model, it generates the optimal cutting path in real time, avoiding surface defects and high-temperature areas. The vibration compensation module uses encoder and IMU measurement data to correct path deviations caused by mechanical vibration in real time. The parameter optimization and control module establishes a regression relationship between cutting speed and kerf width, fits coefficients through experimental data, defines an objective function to evaluate the cutting effect, and uses optimization algorithms to find the optimal cutting speed and laser power.

[0061] As an optimized solution of the present invention, a 3D laser scanner acquires high-precision point cloud data of the metal surface to be cut in real time; provides metal surface features to help identify and locate key features on the cutting path, thereby improving the accuracy of cutting path planning; an infrared thermal imager monitors the temperature distribution in the cutting area, and maps the temperature field data onto the geometric model to provide temperature constraint information for path planning, avoiding high-temperature areas; a vision camera captures real-time cutting images and analyzes the cut quality; an encoder and an inertial measurement unit track the three-dimensional motion trajectory of the laser cutting head, compensate for mechanical vibration errors, and reduce dynamic position errors caused by mechanical vibration.

[0062] The present invention has the following positive effects: 1) High-precision data acquisition: The cutting process is fully monitored through multiple sensors, providing high-quality raw data.

[0063] 2) Precise data preprocessing and modeling: Through preprocessing steps such as filtering, noise reduction, and outlier removal, as well as dynamic 3D model generation technology, the quality of the data and the accuracy of the model are ensured.

[0064] 3) Intelligent path planning and obstacle avoidance: Combining the A* algorithm and multi-source data, intelligent path planning is realized, which effectively avoids high-temperature areas and surface defects, and improves cutting efficiency and quality.

[0065] 4) Effective vibration compensation: The vibration compensation model corrects the path deviation caused by mechanical vibration in real time, improving the cutting accuracy.

[0066] 5) Optimized parameter control: By establishing regression relationships, defining objective functions, and applying optimization algorithms, the optimal configuration of cutting speed and laser power was achieved, further improving cutting quality and efficiency. Attached Figure Description

[0067] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0068] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0069] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” or “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0070] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 One embodiment of the automatic control method for laser cutting according to the present invention includes:

[0071] The S1 and 3D laser scanners generate three-dimensional point cloud data of the metal surface, the infrared thermal imager monitors the temperature field to obtain temperature field data, the vision camera captures the cutting scene to obtain visual image data of the cutting quality, and the encoder and inertial measurement unit monitor the position and vibration frequency of the cutting head to obtain multi-source data.

[0072] Specifically, the 3D laser scanner is used to acquire high-precision point cloud data of the metal surface to be cut in real time, providing surface features, including undulations, holes, and edge features, to help identify and locate key features on the cutting path. The 3D laser scanner acquires the three-dimensional point cloud of the metal surface using triangulation, with the coverage area customized according to the cutting area. The 3D laser scanner is linked with an encoder to ensure strict synchronization between the scanning data and the cutting head position. A high-speed processor module is added to achieve hardware-level trigger synchronization, unifying the clock signals of the 3D laser scanner, encoder, and inertial measurement unit to μs-level accuracy. An infrared thermal imager uses short-wave infrared (SWIR, 1-3μm) to monitor the temperature distribution in the cutting area, preventing localized overheating that could lead to material deformation. A vision camera is used to capture real-time images of the cutting process and analyze the kerf quality, including the cutting width and slag residue, to evaluate the cutting effect. An encoder and inertial measurement unit are also included to track the three-dimensional motion trajectory (X / Y / Z axes) of the laser cutting head, thereby compensating for mechanical vibration errors and improving cutting accuracy.

[0073] S2. Preprocess the multi-source data, map the preprocessed data to three-dimensional space, and generate a dynamic three-dimensional model containing geometric information, temperature distribution and cutting quality.

[0074] Preprocessing of multi-source data includes filtering and denoising of 3D point cloud data, using Kalman filtering to eliminate measurement noise from 3D laser scanners, and improving the stability and accuracy of point clouds.

[0075] An improved threshold segmentation algorithm is used to remove outliers (such as oxide layer reflection interference) and extract valid metal surface information. Specifically:

[0076] Dynamic threshold calculation: m=u1×(L max -L min )+u2×σ;

[0077] Where: m is the dynamic threshold, used to determine which points in the 3D point cloud data belong to normal regions and which belong to abnormal regions; u1 is used to balance the influence of local range on the dynamic threshold; u2 is used to balance the influence of standard deviation on the dynamic threshold; L max L represents the maximum local point cloud height. max σ represents the minimum local point cloud height. σ is the standard deviation of the local point cloud height. Points with heights exceeding [μ-m, μ+m] are marked as noise (μ is the local mean), and the RANSAC algorithm is used to fit a plane and remove outliers. A local point cloud is a set of points within a defined range in point cloud data.

[0078] The preprocessed data is mapped to three-dimensional space to generate a dynamic three-dimensional model containing geometric information, temperature distribution, and cutting quality. Specifically, this includes:

[0079] 1) Spatiotemporal alignment of multi-source data;

[0080] A unified coordinate system and time reference are established for 3D point cloud data, temperature field data, and cutting quality visual image data. Multi-source data are aligned using calibration parameters (rotation matrix R and translation vector t) to ensure spatial consistency. Data streams are aligned based on timestamp interpolation, with interpolation strategies designed for different sampling rates to ensure correspondence between geometric, temperature, and cutting quality data at the same time point.

[0081] 2) Reconstruction of three-dimensional geometric models;

[0082] To generate a continuous surface model from filtered point cloud data, a Poisson surface reconstruction method combined with a dynamic update strategy is employed. Poisson surface reconstruction is an implicit function-based method that fits the point cloud surface using an implicit function χ, solves the Poisson equation, and then generates a continuous surface model from the point cloud data. By fitting the point cloud surface through solving the Poisson equation, it can handle noisy point clouds and generate smooth, continuous surfaces. It also exhibits strong robustness to sparse and irregularly distributed point clouds.

[0083] In real-time point cloud data stream scenarios, the surface model is dynamically updated to reflect the latest geometric changes. Incremental Delaunay triangulation is employed to progressively update mesh vertices V(t) and faces F(t). Real-time updates are supported, making it suitable for dynamic point cloud data streams. Incremental updates reduce the overhead of recalculating the entire mesh, improving efficiency. It can quickly adapt to changes in the point cloud, generating high-quality, continuous surface models.

[0084] 3) Temperature distribution mapping;

[0085] The temperature field data is associated with the vertices / faces of the 3D geometric model, and inverse distance weighted (IDW) interpolation is used for areas that are not directly measured.

[0086]

[0087] Where: T(x,y,z) is the estimated temperature value of the point (x,y,z) to be interpolated, T i Let d be the temperature value at the i-th known temperature point. i From the point to be interpolated (x,y,z) to the i-th known temperature point (x i ,y i ,z i The Euclidean distance is given by the coordinate system, where n is the number of known temperature points. During texture mapping, the temperature field is converted into a UV texture map, which is then rendered using vertex colors or heatmap shaders to visually display the temperature distribution. Through attribute interpolation and texture mapping, temperature data can be effectively associated with the vertices or faces of the geometric model, and the temperature distribution can be visually displayed on the 3D model. This method is suitable for monitoring the surface temperature of metals during laser cutting. UV coordinates are used to map the geometric model onto the texture image, and the shader dynamically adjusts the color according to the temperature value to visualize the temperature distribution.

[0088] 4) Integration of cutting quality parameters; quantification of visual image data of cutting quality and mapping to a three-dimensional geometric model.

[0089] (1) The surface roughness C(p) is calculated using the point cloud curvature formula, and the surface smoothness is measured by the difference in normal vectors.

[0090]

[0091] Where: n(p) is the normal vector of point p, p i It is a neighboring point.

[0092] (2) Establish the regression relationship between cutting speed v and kerf width W to predict and optimize cutting process parameters. W = k1·v + k2·P + b; where: W: kerf width; v: cutting speed; P: laser power. k1, k2, b: fitting coefficients in the regression model, obtained by fitting experimental data, used to describe the influence of cutting speed and laser power on kerf width.

[0093] (3) Display cutting quality parameters (such as roughness, kerf width, temperature, etc.) on the geometric model using color coding or label overlay. Use different colors to represent different attribute values. Color gradients can be dynamically adjusted according to actual attribute values ​​to create an intuitive visualization effect. Add numerical labels to key locations on the geometric model to display specific attribute values ​​(such as roughness values, kerf width values, etc.). Numerical labels: Directly display specific attribute values ​​(such as C(p), W, etc.), making it easy for users to quickly understand the quality information of key areas. Through color coding and label overlay, the spatial distribution of cutting quality parameters can be intuitively displayed, helping to quickly identify problem areas (such as overheating, excessive roughness, etc.) and adjust process parameters in a timely manner.

[0094] 5) Dynamic 3D model update, realizing the dynamic effect of model evolution over time, and updating the point cloud position x. t and temperature T t Perform a status update:

[0095] The state prediction equation is:

[0096] The observation equation is:

[0097] in: This is the state estimate at time t. A is the state transition matrix, describing how the system state transitions from the previous time t-1 to the current time t. B is the control input matrix, describing the control input u. t Impact on system state; w t Process noise represents the random errors present during system state transitions. t , where represents observation noise, indicating random errors present during sensor measurements. H is the observation matrix, mapping state variables to the observation space. t This is the observed value at the current time t.

[0098] S3. Cutting path planning and obstacle avoidance based on dynamic 3D model: Combine dynamic 3D model to generate the optimal cutting path in real time and avoid surface defects (holes) and high temperature areas.

[0099] The dynamic 3D model generated in step S2 contains the following information:

[0100] Geometric information: a continuous surface model reconstructed from point cloud data.

[0101] Temperature distribution: The temperature field mapped onto the geometric model through interpolation.

[0102] Cutting quality parameters: surface roughness, kerf width, etc.

[0103] 1. The optimal cutting path is generated using a heuristic algorithm (A* algorithm). The formula is:

[0104]

[0105] Where: c(si, si+1): from node s i to s i+1 The cost function is defined as follows. Path represents the optimal path. The cost function c(si, si+1) typically consists of multiple sub-terms, each reflecting a different optimization objective or constraint. The following are the components and their meanings:

[0106] 1) Geometric distance: L path =||s i+1 -s i ||, where L path : Total length of the cutting path. ||s i+1 -s i || represents the Euclidean distance between nodes.

[0107] 2) Temperature constraint: J T =T i ·Δt i , among which, T i Let Δt be the temperature of the region where the i-th path segment is located. i Let J be the dwell time on the i-th path segment. T For temperature constraints. Temperature must be satisfied. i ≤T max (Local temperature threshold).

[0108] 3) Surface features: Prioritize smooth areas to reduce cutting errors. Where: JC is the cumulative cost of surface roughness. C(pi): the surface roughness at node i.

[0109] 2. Taking multiple performance indicators into account:

[0110] Cutting efficiency: Minimize the total path length L path .

[0111] Cutting quality: Avoid high-temperature zones and high-roughness zones.

[0112] 31) The steps to implement the optimal cutting path include:

[0113] 31) Initialization; Define start and end points: Determine the start and end points based on the cutting task. Set the initial path: Use a straight line connecting the start and end points as the initial path.

[0114] 32) Path evaluation; Cost function calculation: Calculate the total cost of the current path based on factors such as geometric distance, temperature constraints, and surface features.

[0115] J = w1·L path +w2·J T +w3·J C

[0116] Where: w1, w2, w3: weighting coefficients, representing the degree of importance attached to path length, temperature constraints, and surface features, respectively.

[0117] 33) Use the A* algorithm to iteratively optimize the path. In each iteration, select the path expansion direction with the lowest cost until a termination condition is met. The termination condition is reaching a predetermined path length, the number of optimization iterations, or satisfying all constraints (such as a temperature threshold T). i ≤T max Finally, the final path is generated, and the optimal cutting path is produced based on the optimization results. The path is overlaid on a dynamic 3D model, and key path information (such as path length, temperature distribution, surface roughness, etc.) is displayed through color coding or labels. By combining the geometric information, temperature distribution, and cutting quality parameters provided by the dynamic 3D model, a heuristic algorithm (A* algorithm) is used for path planning, which can effectively balance processing efficiency, cutting quality, and safety. This not only improves the intelligence level of path planning but also significantly enhances the overall efficiency and accuracy of laser cutting.

[0118] 3. Real-time correction of path deviation caused by mechanical vibration using encoder and inertial measurement unit measurement data: Vibration compensation model:

[0119]

[0120] a i : Vibration amplitude (X / Y / Z axis components). ω i : Vibration angular frequency (obtained through FFT analysis). φ i Phase angle. ΔP(t) represents the vibration offset of the laser cutting head in three-dimensional space (X / Y / Z axes), which is the dynamic position error caused by mechanical vibration.

[0121] S4. Establish a regression relationship between cutting speed and kerf width, and fit the coefficients using experimental data. Define an objective function to evaluate the cutting effect, and use an optimization algorithm to find the optimal cutting speed and laser power. Specifically, this includes:

[0122] S41. Establish the regression relationship between cutting speed v and seam width W: W=k1×v+k2×P+b.

[0123] Where: W is the kerf width; v is the cutting speed; P is the laser power; k1, k2, b are fitting coefficients, obtained by fitting experimental data. Kerf width W data were collected under different cutting speeds v and laser powers P, and the coefficients k1, k2, b were fitted using the least squares method.

[0124] S42. Define an objective function to evaluate the cutting effect; J(v,P)=λ1·W+λ2·C(p)+λ3·T where: J(v,P) is the objective function value. W is the kerf width. C(p) is the surface roughness. T is the temperature distribution. λ1, λ2, λ3 are weighting coefficients, set according to actual needs.

[0125] S43. Use an optimization algorithm (particle swarm optimization algorithm) to find the optimal cutting speed v and laser power P. During the cutting process, continuously monitor changes in geometric information, temperature distribution and cutting quality parameters, and adjust the cutting speed v and laser power P in a timely manner.

[0126] [v * ,P * The expression ] = arg min J(v,P) is used to find the optimal cutting speed v and laser power P. Where, [v... * ,P * ] is the optimal solution vector, representing the cutting speed v and laser power P that minimize the objective function J(v,P).

[0127] Particle Swarm Optimization (PSO) simulates the foraging behavior of bird flocks and finds the global optimum through group cooperation.

[0128] The steps are as follows:

[0129] 1) Initialize a group of particles, each particle representing a candidate solution (v, P);

[0130] 2) Calculate the objective function value J(v,P) for each particle;

[0131] 3) Update the velocity and position of each particle to move it toward a better solution;

[0132] 4) Repeat steps 2 and 3 until the termination condition is met (such as reaching the maximum number of iterations or the objective function value no longer changing significantly).

[0133] Finally, a PID controller is used to adjust the laser power and cutting speed in real time.

[0134] The automatic control method for laser cutting in the embodiments of the present invention has been described above. The automatic control system for laser cutting in the embodiments of the present invention is described below. One embodiment of the automatic control system for laser cutting in the embodiments of the present invention includes: a data acquisition module, a data preprocessing and modeling module, a path planning and obstacle avoidance module, a vibration compensation module, and a parameter optimization and control module.

[0135] 1. Data acquisition module;

[0136] 3D laser scanner: acquires high-precision point cloud data of the metal surface to be cut in real time.

[0137] Provides metal surface features (such as undulations, holes, and edges) to help identify and locate key features on the cutting path, improving the accuracy of cutting path planning.

[0138] Infrared thermal imager: Monitors the temperature distribution in the cutting area to prevent localized overheating that could lead to material deformation.

[0139] By mapping temperature field data onto a geometric model, temperature constraint information is provided for path planning, thus avoiding high-temperature areas.

[0140] Visual camera: Captures real-time cutting footage and analyzes kerf quality (such as cutting width and slag residue). Evaluates cutting results and provides feedback on cutting quality for subsequent parameter optimization.

[0141] Encoder and Inertial Measurement Unit (IMU): Track the three-dimensional motion trajectory of the laser cutting head and compensate for mechanical vibration errors. Improve cutting accuracy and reduce dynamic position errors caused by mechanical vibration.

[0142] 2. Data preprocessing and modeling module;

[0143] Data preprocessing: Performing preprocessing operations such as filtering, denoising, and outlier removal on multi-source data.

[0144] Improve the stability and accuracy of point clouds, ensure data quality, and provide a reliable foundation for subsequent modeling.

[0145] Dynamic 3D Model Generation: Preprocessed data is mapped to 3D space to generate a dynamic 3D model containing geometric information, temperature distribution, and cutting quality. This achieves spatiotemporal alignment of multi-source data, generates a continuous surface model, and integrates temperature field and cutting quality parameters into the model, supporting real-time updates and visualization.

[0146] 3. Path planning and obstacle avoidance module: Based on a dynamic 3D model, it generates the optimal cutting path in real time, avoiding surface defects (such as holes) and high-temperature areas. By comprehensively considering geometric distance, temperature constraints, and surface features, it balances processing efficiency, cutting quality, and safety, significantly improving the overall efficiency and precision of laser cutting.

[0147] 4. Vibration compensation module;

[0148] Vibration compensation model: Utilizes encoder and IMU measurement data to correct path deviations caused by mechanical vibration in real time. This reduces dynamic position errors due to mechanical vibration and improves cutting accuracy.

[0149] 5. Parameter optimization and control module;

[0150] Regression relationship establishment: Establish the regression relationship between cutting speed v and kerf width W, and fit the coefficients using experimental data. Provide a predictive model for cutting process parameters, offering a basis for parameter optimization.

[0151] Objective function definition and optimization: Define the objective function to evaluate the cutting effect, and use optimization algorithms (such as particle swarm optimization) to find the optimal cutting speed v and laser power P. The optimization algorithm finds the best combination of process parameters to improve cutting quality and efficiency.

[0152] PID controller: Adjusts laser power and cutting speed in real time. Ensures parameters remain optimal during the cutting process, adapting to dynamically changing environmental conditions.

[0153] The effects of this invention can be summarized as follows:

[0154] 1) High-precision data acquisition: Through multiple sensors such as 3D laser scanner, infrared thermal imager, vision camera and IMU, comprehensive monitoring of the cutting process is achieved, providing high-quality raw data.

[0155] 2) Precise data preprocessing and modeling: Through preprocessing steps such as filtering, noise reduction, and outlier removal, as well as dynamic 3D model generation technology, the quality of the data and the accuracy of the model are ensured.

[0156] 3) Intelligent path planning and obstacle avoidance: Combining the A* algorithm and multi-source data, intelligent path planning is realized, which effectively avoids high-temperature areas and surface defects, and improves cutting efficiency and quality.

[0157] 4) Effective vibration compensation: The vibration compensation model corrects the path deviation caused by mechanical vibration in real time, improving the cutting accuracy.

[0158] 5) Optimized parameter control: By establishing regression relationships, defining objective functions, and applying optimization algorithms, the optimal configuration of cutting speed and laser power was achieved, further improving cutting quality and efficiency.

[0159] In summary, this automatic control method and system for laser cutting achieves fully automated control from data acquisition to final cutting effect evaluation through key technologies such as multi-source data fusion, intelligent path planning, vibration compensation, and parameter optimization, significantly improving the accuracy and efficiency of laser cutting.

[0160] This invention also provides a computer device, which may be a server. The computer device includes a processor, memory, display screen, input device, network interface, and database connected via a system bus. The processor in this computer design provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database of the computer device is used to store the data corresponding to this embodiment. The network interface of the computer device is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements the above-described method.

[0161] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0162] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0163] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An automatic control method for laser cutting, characterized in that, include: The S1 and 3D laser scanners generate three-dimensional point cloud data of the metal surface, the infrared thermal imager monitors the temperature field to obtain temperature field data, the vision camera captures the cutting scene to obtain visual image data of the cutting quality, and the encoder and inertial measurement unit monitor the position and vibration frequency of the cutting head to obtain multi-source data. S2. Preprocess the multi-source data, map the preprocessed data to three-dimensional space, and generate a dynamic three-dimensional model containing geometric information, temperature distribution and cutting quality. S3. Cutting path planning and obstacle avoidance based on dynamic 3D model; By combining a dynamic 3D model, the optimal cutting path is generated in real time, avoiding surface defects and high-temperature areas; S4. Establish the regression relationship between cutting speed and kerf width, fit the coefficients through experimental data, define the objective function to evaluate the cutting effect, and use optimization algorithms to find the optimal cutting speed and laser power. In step S2, the preprocessing includes filtering and denoising the 3D point cloud data, and using Kalman filtering to eliminate measurement noise from the 3D laser scanner; The preprocessed data is mapped to three-dimensional space to generate a dynamic three-dimensional model containing geometric information, temperature distribution, and cutting quality. Specifically, this includes: 1) Spatiotemporal alignment of multi-source data; unify the coordinate system and time reference of 3D point cloud data, temperature field data, and cutting quality visual image data, and align multi-source data through calibration parameters to ensure spatial consistency; 2) 3D geometric model reconstruction; In order to generate a continuous surface model from the filtered point cloud data, the Poisson surface reconstruction method combined with a dynamic update strategy is adopted; In the real-time point cloud data stream scenario, the surface model is dynamically updated to reflect the latest geometric changes. Incremental Delaunay triangulation is adopted to gradually update the mesh vertices V(t) and faces F(t). 3) Temperature distribution mapping; associating temperature field data with vertices / faces of the 3D geometric model, and using inverse distance weighted interpolation for areas not directly measured; 4) Integration of cutting quality parameters; quantification of visual image data on cutting quality and mapping it to a 3D geometric model; 5) Dynamic 3D model update, realizing the dynamic effect of model evolution over time, and updating the point cloud position x. t and temperature T t Perform a status update: The state prediction equation is: The observation equation is: in: This is the state estimate at time t. A is the state transition matrix, describing how the system state transitions from the previous time t-1 to the current time t. B is the control input matrix, describing the control input u. t Impact on system state; w t Process noise represents the random error present during the system's state transition process; v t Observational noise represents the random error present during sensor measurement. H The observation matrix maps the state variables to the observation space; z t This is the observed value at the current time t.

2. The automatic control method for laser cutting according to claim 1, characterized in that: In step S1, the 3D laser scanner is used to acquire three-dimensional point cloud data of the metal surface to be cut in real time, providing metal surface features, including undulations, holes and edge features, to help identify and locate key features on the cutting path; The 3D laser scanner and encoder work in tandem to ensure that the scanned data is strictly synchronized with the position of the cutting head. A high-speed processor module is added to achieve hardware-level trigger synchronization, unifying the clock signals of the 3D laser scanner, encoder, and inertial measurement unit to μs-level accuracy. A vision camera is used to capture real-time cutting images and analyze the kerf quality, including the cutting width and slag residue, to evaluate the cutting effect. An encoder and inertial measurement unit are also set up to track the three-dimensional motion trajectory of the laser cutting head, thereby compensating for mechanical vibration errors.

3. The automatic control method for laser cutting according to claim 2, characterized in that: Integration of cutting quality parameters; quantification of cutting quality visual image data and mapping to a 3D geometric model, specifically including: (1) Calculate surface roughness using the point cloud curvature formula C ( p The smoothness of a surface is measured by the difference in normal vectors. ;in: n ( p ) is a point p The normal vector, p i For point p Neighboring points; (2) Establish cutting speed v With seam width W The regression relationship is used to predict and optimize cutting process parameters; (3) Display cutting quality parameters on the geometric model by means of color coding or label overlay.

4. The automatic control method for laser cutting according to claim 3, characterized in that: Step S3 specifically includes: S31. The optimal cutting path is generated using the A* algorithm, with the following formula: in: c ( s i , s i+1 ): From node s i arrive s i+1 The cost function, where Path is the optimal path; S32. Taking into account both cutting efficiency and cutting quality, the cutting efficiency is determined by minimizing the total path length. L path Cutting quality is achieved by avoiding high-temperature and high-roughness areas; specifically, this is implemented as follows: 31) Initialization; Define start and end points: Determine the start and end points based on the cutting task, and set the initial path: Use a straight line to connect the start and end points as the initial path; 32) Path evaluation; Cost function calculation: Calculate the total cost of the current path based on geometric distance, temperature constraints, and surface feature factors; J = w 1 · L path + w 2 · J T + w 3 · J C in: w 1, w 2, w 3: Weighting coefficients, representing the degree of importance attached to path length, temperature constraints, and surface features, respectively. J T For temperature constraints, J C This is the cumulative cost of surface roughness; 33) Use the A* algorithm to optimize the path step by step; in each iteration, select the path expansion direction with the minimum cost until the termination condition is met; S33. Real-time correction of path deviation caused by mechanical vibration using encoder and inertial measurement unit measurement data: Vibration compensation model: : Vibration amplitude; ω i : Angular frequency of vibration; : Phase angle; ΔP(t) represents the vibration offset of the laser cutting head in three-dimensional space.

5. The automatic control method for laser cutting according to claim 4, characterized in that: Step S4 specifically includes: S41. Establish the regression relationship between cutting speed v and seam width W: W = k1 × v + k2×P + b; Where: W is the kerf width; v is the cutting speed; P is the laser power; k1 , k2 , b is the fitting coefficient; S42. Define an objective function to evaluate the cutting effect; J (v,P)=λ1·W + λ2·C(p) + λ3·T; where: J ( v , P ) is the objective function value, W is the kerf width, C(p) is the surface roughness, T is the temperature distribution, and λ1 is the kerf width. , λ2 , λ3 is the weighting coefficient; S43. Use particle swarm optimization algorithm to find the optimal cutting speed v and laser power P. During the cutting process, continuously monitor the changes in geometric information, temperature distribution and cutting quality parameters, and adjust the cutting speed v and laser power P in a timely manner. [ v ∗ , P ∗ ]=arg min J ( v , P The laser is used to find the optimal cutting speed v and laser power P, and finally a PID controller is used to adjust the laser power and cutting speed in real time.

6. The automatic control method for laser cutting according to claim 5, characterized in that: In step S43, the particle swarm optimization algorithm is used to simulate the foraging behavior of bird flocks, and the global optimum is found through group cooperation. The steps are as follows: 1) Initialize a group of particles, each particle representing a candidate solution (v, P); 2) Calculate the objective function value J(v,P) for each particle; 3) Update the velocity and position of each particle to move it toward a better solution; 4) Repeat steps 2 and 3 until the termination condition is met. The termination condition includes reaching the maximum number of iterations or the objective function value no longer changing significantly.

7. A control system employing the automatic control method for laser cutting as described in claim 6, characterized in that: The system includes a data acquisition module, a data preprocessing and modeling module, a path planning and obstacle avoidance module, a vibration compensation module, and a parameter optimization and control module. The data acquisition module generates 3D point cloud data of the metal surface using a 3D laser scanner, obtains temperature field data by monitoring the temperature field with an infrared thermal imager, captures the cutting scene using a vision camera to obtain visual image data of the cutting quality, and monitors the position and vibration frequency of the cutting head using an encoder and an inertial measurement unit, thus obtaining multi-source data. The data preprocessing and modeling module preprocesses the multi-source data, mapping the preprocessed data to 3D space to generate a dynamic 3D model containing geometric information, temperature distribution, and cutting quality. The path planning and obstacle avoidance module plans the cutting path and avoids obstacles based on the dynamic 3D model. Combining a dynamic 3D model, the system generates the optimal cutting path in real time, avoiding surface defects and high-temperature areas. The vibration compensation module uses encoder and IMU measurement data to correct path deviations caused by mechanical vibration in real time. The parameter optimization and control module establishes a regression relationship between cutting speed and kerf width, fits coefficients through experimental data, defines an objective function to evaluate the cutting effect, and uses optimization algorithms to find the optimal cutting speed and laser power.

8. The control system of the automatic control method for laser cutting according to claim 7, characterized in that: The 3D laser scanner acquires high-precision point cloud data of the metal surface to be cut in real time; it provides metal surface features to help identify and locate key features on the cutting path, thereby improving the accuracy of cutting path planning; the infrared thermal imager monitors the temperature distribution in the cutting area and maps the temperature field data onto the geometric model to provide temperature constraint information for path planning, thus avoiding high-temperature areas. A vision camera captures real-time images of the cutting process and analyzes the cut quality; an encoder and an inertial measurement unit track the three-dimensional motion trajectory of the laser cutting head, compensating for mechanical vibration errors and reducing dynamic position errors caused by mechanical vibration.

Images