Cutting layout planning method and system for aluminum plate workpieces

Abstract

This invention relates to the field of aluminum plate cutting and processing technology, specifically to a method and system for cutting and layout planning of aluminum plate parts. The method includes: acquiring the contour data and process parameters of the workpiece; calculating an eigenvector containing a concavity descriptor and an inherent thermal load reference, the latter calculated based on the cutting path length, number of sharp corners, and unit heat input; constructing a heterogeneity metric by weighting normalized concavity difference and thermal load difference, with the thermal load difference weight increasing with the average thermal load of the two workpieces; dividing initial clusters based on this and iteratively optimizing them, calculating the minimum circumscribed polygon area and cluster thermal density after pre-layout; repeatedly executing boundary workpiece migration that maximizes the decrease in the objective function until convergence to generate the final cluster; and laying out the final clusters and controlling the cutting speed or cooling time using cluster thermal density. This invention avoids warping of the sheet metal caused by the accumulation of high-heat-generating parts, improves processing accuracy, and shortens the cycle time.

Description

Technical Field

[0001] This invention relates to the field of aluminum plate cutting and processing technology, and specifically to a method and system for cutting and layout planning of aluminum plate processed parts. Background Technology

[0002] Aluminum alloys, due to their high thermal conductivity and coefficient of thermal expansion, are sensitive to heat input when subjected to high-energy beam processing such as laser cutting and plasma cutting. When the workpiece has a complex shape with many concave features or dense sharp corners, heat accumulation and thermal stress are easily generated in localized areas during the cutting process. Layout planning focuses on pure geometric nesting, that is, arranging the workpieces as closely as possible through various nesting algorithms to pursue material utilization. However, a layout strategy that only fits around geometric boundaries will result in a high concentration of cutting trajectories in a small area of ​​the sheet material, causing a surge in localized heat input and leading to localized heat accumulation. This can cause the entire aluminum sheet to warp, deform thermally, or even be scrapped during the cutting process.

[0003] Current layout schemes lack the ability to comprehensively assess the differences in geometric contour properties and thermal response of workpieces in the initial layout stage. They cannot measure the attribute relationships between dissimilar components with different shape characteristics, nor can they proactively avoid local heat accumulation during the macro-layout stage. Existing planning systems typically treat layout and underlying processing control separately, failing to adjust cutting speed or cooling time parameters based on the heat density distribution of local layout areas. This disconnect between geometric layout and thermodynamic constraints limits the overall processing efficiency of the cutting system. Therefore, there is an urgent need for an aluminum plate cutting layout planning method that can integrate the geometric features of the workpiece with its inherent thermal load characteristics, collaboratively optimize layout efficiency and heat density balance from an overall structural perspective, and achieve intelligent control by linking underlying process parameters. This would break down the technical barriers that prevent the simultaneous achievement of high material utilization and high processing quality.

[0004] Chinese patent application CN121659786A discloses a reinforcement learning-based method for co-optimizing sheet metal utilization and energy consumption. This method acquires the contour data of the workpieces to be processed, constructs a feature vector containing an energy consumption component, and weights the vectors according to the feature differences between the workpieces. It then uses a policy network and reinforcement learning to iteratively optimize the layout, aiming to balance sheet metal utilization and cutting energy consumption during the layout process. However, this method, guided by the overall optimization of sheet metal utilization and energy consumption, lacks targeted characterization of local heat distribution caused by differences in the contour shape and heating tendency of the workpieces. Under conditions of concentrated heat or significant differences in heat generation among the workpieces, it struggles to suppress thermal deformation caused by local heat accumulation in the sheet metal at the layout stage, and its ability to adaptively control the cutting process is also insufficient.

[0005] Chinese patent document CN120940879B discloses a method for suppressing thermal deformation during laser cutting with integrated temperature control feedback. This method uses real-time detected molten pool temperature and feedback from the local temperature field of the workpiece to correct and adjust laser power, cutting speed, auxiliary gas flow rate, and cooling time, suppressing thermal deformation during the cutting process through closed-loop control. While this method can dynamically adjust the cutting process using real-time temperature feedback, its control relies on online temperature acquisition and response during the cutting execution phase. It lacks pre-planning of the heat generation tendency of the workpiece and the overall heat distribution during the layout planning stage, making it difficult to balance the heat distribution of the sheet metal from the source. In conditions where the workpieces are densely packed and have significant differences in heat generation, localized heat concentration can still occur, affecting dimensional accuracy and processing stability. Summary of the Invention

[0006] To improve the dimensional accuracy and overall yield of aluminum plate cutting, this invention provides a method and system for cutting and layout planning of aluminum plate parts.

[0007] In a first aspect, the present invention provides the following technical solution: a cutting and layout planning method for aluminum plate processed parts, comprising: acquiring contour data and process parameters of the processed parts to be laid out; calculating feature vectors for each processed part, wherein the feature vectors include a concavity descriptor representing the concavity feature of the contour, and an inherent thermal load reference calculated based on the cutting path length, the number of sharp corners, and the unit heat input; constructing a heterogeneity metric between processed parts based on the feature vectors, wherein the heterogeneity metric is obtained by weighting the difference value of the normalized concavity descriptor and the difference value of the normalized inherent thermal load reference, and the weight of the difference value of the normalized inherent thermal load reference is positively correlated with the mean of the inherent thermal load references of the two processed parts; and dividing the processed parts into [aspects] according to the heterogeneity metric. Multiple initial clusters are used; the initial clusters are iteratively optimized, the workpieces within each initial cluster are pre-arranged, and the minimum bounding polygon of each initial cluster is calculated to obtain the polygon area representing the layout efficiency, and the ratio of the sum of the inherent thermal load references of the workpieces within the initial cluster to the polygon area is used as the cluster heat density; the change in the system objective function caused by the migration of boundary workpieces to adjacent initial clusters is calculated, and the system objective function is positively correlated with the total polygon area of ​​each initial cluster and the variance of the cluster heat density; the migration operation that maximizes the decrease in the system objective function is repeatedly executed until convergence, generating the final cluster; the final clusters are laid out, and the cutting speed or cooling time within the corresponding final clusters is adjusted based on the cluster heat density.

[0008] By simultaneously introducing a concavity descriptor representing the degree of contour concavity into the feature vector, along with an inherent thermal load benchmark that integrates cutting path length, number of sharp corners, and unit heat input, and constructing a heterogeneous metric by weighting normalized concavity difference and normalized thermal load difference, the similarity judgment between processed parts reflects both geometry and heating tendency. Furthermore, because the weight of thermal load difference increases with the mean of the inherent thermal load benchmarks of the two processed parts, it avoids high-heat-generating processed parts being concentrated in the same area, preventing localized heat accumulation and warping of the sheet metal. Based on this, layout efficiency is measured by polygon area, and cluster heat density is measured by the ratio of the sum of thermal loads to the area. The system objective function is simultaneously correlated with the total polygon area and the variance of cluster heat density. Repeated execution of boundary processed part migration that maximizes the decrease in the objective function ensures that the heat distribution of each cluster tends to be more balanced while maintaining compact material cutting. Finally, the cutting speed or cooling time is adjusted based on cluster heat density, improving processing quality problems caused by thermal deformation and enhancing the coordination and reliability of layout planning and cutting execution.

[0009] Furthermore, obtaining the contour data and process parameters of the workpiece to be processed includes: obtaining a processing task package containing three-dimensional model data from the production management server, extracting the two-dimensional contour point set mapped from the three-dimensional model to the cutting plane; and extracting the corresponding cutting process parameters from the equipment database, wherein the cutting process parameters include the maximum cutting speed, constant output laser power, and safe operation threshold parameters.

[0010] Furthermore, the process of obtaining a machining task package containing 3D model data from the production management server and extracting a 2D contour point set includes: downloading a 3D model machining task package in STEP or IGES format via a communication interface connection to the server, extracting model edge data, projecting the 3D boundary curve into a 2D contour according to the set cutting plane, and generating a sequence of 2D contour point sets containing X and Y coordinates using discretization processing, wherein the sampling interval between adjacent coordinate points is 0.1mm to 0.5mm; the safe operation threshold parameters include the maximum acceleration limit of the machine tool and the safe avoidance lifting height of the follow-up cutting head during idle movement.

[0011] Furthermore, the inherent thermal load benchmark calculated based on the cutting path length, the number of sharp corners, and the unit heat input includes: traversing the contour data of the workpiece, extracting the turning points where the included angle between adjacent line segments is less than a preset acute angle threshold, and counting the total number of sharp corners of the workpiece; calculating the sum of the lengths of all line segments of the two-dimensional contour of the workpiece to obtain the total length of the cutting path; multiplying the total length of the cutting path and the total number of sharp corners by the preset unit heat input and the sharp corner heat concentration coefficient, respectively, and summing them to obtain the inherent thermal load benchmark of the workpiece.

[0012] By traversing the two-dimensional contour point set, the turning points where the included angle between adjacent line segments is less than a preset acute angle threshold are extracted, and the total number of sharp corners is counted. The total length of the cutting path is obtained by summing the lengths of all line segments. Then, the lengths are multiplied by the unit heat input and the heat concentration coefficient of the sharp corners respectively, and then summed. This ensures that the inherent heat load benchmark includes both the linear energy output on the conventional cutting path and the concentrated additional heat energy generated by the interpolation deceleration and dwell at the sharp corners. This avoids ignoring the local heat accumulation caused by shape changes when estimating heat generation based solely on the contour size. This allows the heat generation tendency of each processed part to be measured more accurately, providing a reliable basis for subsequent clustering and heat density control.

[0013] Further, constructing a heterogeneity metric between processed parts based on the feature vector includes: calculating the absolute value of the difference between the concavity descriptors of two processed parts and normalizing it as a first metric; calculating the absolute value of the difference between the inherent thermal load references of two processed parts and normalizing it as a second metric; calculating the average value of the inherent thermal load references of the two processed parts; when the average value is less than a set material thermal sensitivity threshold, using a fixed weight combination of the first and second metric; when the average value is greater than or equal to the material thermal sensitivity threshold, the weight of the second metric increases exponentially with the ratio of the difference between the average value and the material thermal sensitivity threshold to the material thermal sensitivity threshold, thus calculating the heterogeneity metric.

[0014] By normalizing the difference in concavity descriptors and the difference in inherent thermal load benchmarks, respectively, and using them as the first and second metric indicators, a two-level weighted control logic is established based on the comparison between the average value of the thermal load benchmarks of the two processed parts and the material's thermistor threshold. When the average thermal load is low, a fixed weight is used to take into account both shape and heat generation. When the average thermal load reaches or exceeds the material's thermistor threshold, the weight of the thermal load difference index increases exponentially with the excess ratio, thereby generating greater metric repulsion between high-heat-generating processed parts and preventing large-heat-generating parts from being classified into the same cluster during the clustering stage. This reduces the possibility of board warping due to heat concentration at the source of cluster division.

[0015] Further, the workpieces are divided into multiple initial clusters based on the heterogeneous metric and iteratively optimized, including: inputting the heterogeneous metric distance matrix into a clustering algorithm, setting the number of cluster centers to complete the initial cluster division; establishing a system objective function, which is composed of the sum of the polygon areas of all initial clusters and the variance of the cluster heat density of all initial clusters through a linear combination of area penalty factor and variance penalty factor; using a greedy local search algorithm to traverse the workpieces at the boundary of each initial cluster, simulating moving the workpieces at the boundary into adjacent initial clusters and re-executing pre-layout and polygon area calculation to obtain the change in the system objective function; selecting the migration operation that maximizes the absolute value of the decrease in the system objective function to update the cluster member allocation, until the decrease in the system objective function for a set number of consecutive iterations is less than the convergence threshold, stopping the iteration and outputting the final cluster.

[0016] The initial clustering is completed by inputting the heterogeneous metric distance matrix into the clustering algorithm. The system objective function is constructed by linearly combining the sum of polygon areas and the variance of cluster heat density with area penalty factors and variance penalty factors, so that layout compactness and heat distribution uniformity are incorporated into a unified optimization criterion. Then, a greedy local search is used to traverse the processing parts at the cluster boundary, tentatively migrate them to adjacent clusters, and re-pre-layout and calculate the area. The migration of the cluster members that maximizes the absolute value of the decrease in the objective function is selected to update the cluster members. The process stops when the decrease is less than the convergence threshold after a set number of consecutive iterations. This avoids the local imbalance caused by relying solely on the initial partitioning, so that the final clusters achieve a more uniform heat density distribution while taking into account the material cutting efficiency, thus improving the stability of the partitioning results.

[0017] Furthermore, the pre-layout of the processed parts within each initial cluster and the calculation of the minimum bounding polygon of each initial cluster are performed, including: after pre-layouting the processed parts within the same initial cluster, extracting the outer boundary contour vertices of all processed parts within the initial cluster to construct a two-dimensional discrete point set of the initial cluster; using the convex hull algorithm to calculate the convex hull polygon of the two-dimensional discrete point set; using the convex hull polygon as the minimum bounding polygon of the initial cluster, and calculating the coverage area of ​​the minimum bounding polygon.

[0018] Furthermore, the final clusters are laid out, including: arranging them in descending order of the area of ​​the smallest bounding polygon of each final cluster; within the blank aluminum plate area, arranging the smallest bounding polygons of each final cluster as a whole in a two-dimensional, non-overlapping, compact arrangement according to the descending area order; after determining the global position of each final cluster on the blank aluminum plate area, maintaining the pre-determined relative position within the smallest bounding polygon area corresponding to each final cluster, performing a local non-overlapping layout of the processed parts within the final cluster.

[0019] By arranging the final clusters in descending order of their smallest circumscribed polygon area and prioritizing the placement of larger clusters, and then using the whole as a unit to perform a two-dimensional, non-overlapping, compact arrangement within the blank aluminum plate area, larger contour units occupy the plate surface first, reducing the waste of space in subsequent filling and improving the utilization rate of the material. After determining the global position of each cluster, the relative position determined by the pre-layout is maintained within the corresponding smallest circumscribed polygon area to complete the local layout within the cluster, avoiding conflicts between the global layout and the thermal distribution arrangement within the cluster. This ensures that the cluster pattern formed by the thermal load in the previous step is continued in the final blanking diagram, guaranteeing that the object of thermal density control is consistent with the layout result.

[0020] Furthermore, based on the cluster heat density, the cutting speed or cooling time within the corresponding final cluster is adjusted, including: determining whether the cluster heat density of the target final cluster exceeds a first set baseline; if it does, calculating the cutting speed increase based on the excess ratio; controlling the cutting head to run at the increased speed within the target final cluster when the increased speed does not exceed the maximum cutting speed; if it exceeds the maximum cutting speed, then running at the maximum cutting speed; determining whether the cluster heat density of the target final cluster exceeds a second set baseline, where the second set baseline is greater than the first set baseline; if it does, inserting a cooling waiting time proportional to the excess cluster heat density after completing the cutting of each workpiece within the target final cluster.

[0021] In a second aspect, the present invention provides a cutting and layout planning system for aluminum sheet processing parts, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the cutting and layout planning method for aluminum sheet processing parts of the present invention is implemented.

[0022] The beneficial effects of this invention are as follows: The concavity descriptor and the inherent thermal load benchmark are jointly incorporated into the feature vector, and a heterogeneous metric is constructed with a weight that increases with the mean thermal load. This allows for the reasonable differentiation of processed parts with similar shapes and heat generation, preventing board warping caused by the aggregation of high-heat-generating parts from the outset. After clustering, iterative migration is performed using a system objective function that correlates the total area of ​​polygons with the variance of cluster thermal density. This ensures that both compactness of material cutting and uniform heat distribution are achieved simultaneously, ultimately resulting in a more uniform cluster thermal density. During the layout stage, the materials are arranged compactly in descending order of circumscribed polygon area while maintaining relative positions within clusters. This maintains board utilization while continuing the pattern formed by thermal load. During the cutting execution stage, the cutting speed and cooling waiting time are controlled in two levels based on cluster thermal density. This reduces local heat accumulation and compensates for cooling delay with increased speed, thereby improving processing accuracy while shortening the processing cycle and enhancing the coordination and reliability of layout planning and cutting execution. Attached Figure Description

[0023] Figure 1 This is a flowchart of the cutting and layout planning method for aluminum plate processed parts according to an embodiment of the present invention;

[0024] Figure 2 This is a schematic diagram of the contour point set in an embodiment of the present invention; Figure 3 This is a bar chart comparing processing time in embodiments of the present invention; Figure 4 This is a structural block diagram of the aluminum plate processing part cutting and layout planning system according to an embodiment of the present invention. Detailed Implementation

[0025] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0026] Example of a method for cutting and layout planning of aluminum sheet processed parts: like Figure 1 As shown, the cutting and layout planning method for aluminum plate processed parts of the present invention includes: S1. Obtain the contour data and process parameters of the parts to be processed; calculate the feature vector of each part. The feature vector includes a concavity descriptor representing the concavity feature of the contour, and an inherent thermal load benchmark calculated based on the cutting path length, the number of sharp corners, and the unit heat input; construct a heterogeneity metric between the parts based on the feature vector. The heterogeneity metric is obtained by weighting the difference value of the normalized concavity descriptor and the difference value of the normalized inherent thermal load benchmark, and the weight of the difference value of the normalized inherent thermal load benchmark is positively correlated with the mean of the inherent thermal load benchmarks of the two parts.

[0027] The design file is read, and two-dimensional contour data consisting of a sequence of polygon vertex coordinates is extracted. Then, by parsing the Extensible Markup Language (XML) configuration file, process parameters including sheet thickness, laser power, and reference cutting speed are obtained. Based on the acquired two-dimensional contour data, the convex hull area of ​​the machined part is calculated. The difference between the actual contour area and the convex hull area is calculated, and then divided by the convex hull area to obtain the concavity descriptor.

[0028] The cutting path length is obtained by traversing vertices and accumulating distances. Using the vector dot product and inverse cosine formulas, the interior angles of adjacent line segments are identified as less than a certain value. The number of sharp corners is accumulated at each vertex. The unit heat input is obtained by dividing the laser power by the cutting speed. The inherent thermal load baseline is obtained by linearly weighting and summing the cutting path length, the number of sharp corners, and the unit heat input using a preset scaling factor.

[0029] The absolute values ​​of the differences between the concavity descriptors and the inherent thermal load references of the two processed parts are calculated separately and normalized to obtain the corresponding difference values. The arithmetic mean of the original thermal load references for the two processed parts is calculated. When this arithmetic mean is less than a set material thermal sensitivity threshold, a fixed weight is used as the weight for the difference value of the inherent thermal load reference and the difference value of the concavity descriptor. When the arithmetic mean is greater than or equal to the material thermal sensitivity threshold, the weight of the difference value of the inherent thermal load reference increases exponentially, while the weight of the difference value of the concavity descriptor is adaptively reduced. The two weighted difference values ​​are summed to obtain the heterogeneous metric distance matrix.

[0030] In some implementations, the contour data and process parameters of the workpiece to be laid out are obtained, including: Obtain the processing task package containing 3D model data from the production management server, and extract the 2D contour point set of the 3D model mapped to the cutting plane; Extract the corresponding cutting process parameters from the equipment database. The process parameters include the maximum cutting speed, constant output laser power, and safe operation threshold parameters.

[0031] Connect to the server via TCP / IP communication interface to download the STEP or IGES format 3D model processing task package, extract model edge data, and process it according to the set parameters. The cutting plane projects the three-dimensional boundary curve into a two-dimensional contour, and uses discretization to generate a contour containing... , A sequence of two-dimensional contour points with coordinates, wherein the preferred range for the sampling interval between adjacent coordinate points is [range missing]. to Example value The two-dimensional contour point set generated by the above projection and discretization process is as follows: Figure 2 As shown, the sampling points are evenly distributed along the closed contour of the workpiece, which can restore the geometric features of the outer boundary of the part.

[0032] By connecting to the parameter database of the laser cutting equipment via the ODBC interface, numerical process parameters are read according to the material and plate thickness mapping table, clarifying the relevant parameter variables and their optimal ranges, including: maximum cutting speed. Set the speed to the range of 8000mm / min to 15000mm / min, with an example value of 12000mm / min; set the constant output laser power P to the range of 3000W to 12000W, with an example value of 6000W.

[0033] Based on this, the safe operation threshold parameters are analyzed simultaneously, including the maximum acceleration limit of the machine tool. The set value is 1.5G, and the safety avoidance lifting height of the follow-up cutting head during idle movement is set to 2.0mm. The two-dimensional contour coordinate sequence obtained from the above analysis is jointly packaged with the corresponding energy and kinematic constraint values ​​and persistently stored in the system's structured database table.

[0034] This execution mechanism can transform an unprocessed 3D CAD model into a machining dataset that can be parsed by a 2D CNC platform and includes clear electromechanical and physical constraints.

[0035] In some implementations, the inherent thermal load reference calculated based on the cutting path length, the number of sharp corners, and the unit heat input includes: Traverse the two-dimensional contour point set of the workpiece, extract the turning points where the included angle between adjacent line segments is less than the preset acute angle threshold, and count the total number of sharp corners of the workpiece. Calculate the sum of the lengths of all line segments of the two-dimensional contour of the workpiece to obtain the total length of the cutting path; Multiply the total length of the cutting path and the total number of sharp corners by the preset unit path heat input and sharp corner heat concentration coefficient, respectively, and sum them to obtain the inherent thermal load reference of the workpiece.

[0036] Analyze the two vectors formed by three adjacent points in the contour set, and calculate the geometric angle using the dot product and cross product formulas. The preset acute angle threshold is set. The preferred range is between 30° and 60°, with an example value of 45°, which will be calculated. Local turning points are marked as heat-sensitive sharp corners. After traversing the closed contour, the total number of sharp corners of the current workpiece is obtained by accumulating the count. .

[0037] The length of the discrete line segment formed by adjacent coordinate points in the contour sequence is calculated one by one using the Euclidean distance formula. and all The total length of the cutting path of the workpiece is obtained by linear summation. .

[0038] Entering the thermophysical quantization stage, based on the obtained constant output laser power P and the reference cutting speed... Calculate the heat input per unit path. In one embodiment, the reference cutting speed is divided by 60s and then substituted into the previous formula. The preferred range for this is 10J / mm to 50J / mm for a 3mm aluminum plate, with an example value of 45J / mm. Since excessive radiation dwell occurs when the machine tool performs interpolation speed reduction at sharp corners, the heat concentration coefficient at sharp corners is utilized. The coefficient is used to represent the additional heat energy generated by a single deceleration stay. The preferred range for this coefficient is 50J / unit to 200J / unit, with an example value of 120J / unit.

[0039] The calculation engine uses algebraic equations Solve by summing the conventional linear energy output with the concentrated energy caused by abrupt shape changes, and output the inherent thermal load reference for each part to be laid out. value.

[0040] In some implementations, heterogeneity metrics between processed parts are constructed based on feature vectors, including: The absolute value of the difference between the concavity descriptors of the two machined parts is calculated and normalized as the first metric. The absolute value of the difference between the inherent thermal load references of the two processed parts is calculated and normalized as the second metric. Calculate the average value of the inherent thermal load reference of the two processed parts. When the average value is less than the set material thermal sensitivity threshold, the first and second measurement indicators are combined with fixed weights. When the average value is greater than or equal to the material thermal sensitivity threshold, the weight of the second measurement indicator increases exponentially with the ratio of the difference between the average value and the material thermal sensitivity threshold to the material thermal sensitivity threshold, and the heterogeneity measurement is calculated.

[0041] The calculation module extracts the processed parts. With the processed parts Pre-calculated concavity descriptor , The absolute value of the difference between the two is calculated and linearly mapped to the [0,1] interval using the Min-Max normalization method to generate the dimensionless first metric. Similarly, the inherent thermal load references for both are extracted. and The second metric is obtained by calculating the absolute value of the difference and performing the same normalization process. .

[0042] Calculate the average thermal load of the two parts And call the material thermal sensitivity threshold in the material property table. For conventional 6061 aluminum alloy materials, the preferred setting range for this threshold is 5000J to 15000J, with an example standard value of 8000J.

[0043] In the multi-dimensional feature fusion stage, a two-level control logic is established. When J is in the processing environment, it indicates a low risk of thermal deformation. In this case, the heterogeneity metric is calculated. Where the weights are fixed and The preferred ranges are all from 0.3 to 0.7 and the sum is 1. In the example, the values ​​are respectively... , When judged At time J, the heat-resistant clustering sensitive state is activated, and the formula is used. Increase the proportion of heat measurement, exponential growth factor The preferred range is set to 0.5 to 2.0, with an example value of 1.2.

[0044] Simultaneously, the shape metric weights are adaptively decayed. ,if Then The truncation limit is set at 0.9, retaining a shape optimization factor of 0.1. The new weights are then calculated... This functional relationship generates a large repulsive force between high-heat-generating workpieces, reducing the possibility of large parts clustering and causing board warping at the source of clustering.

[0045] S2, the workpieces are divided into multiple initial clusters based on the heterogeneity metric; iterative optimization is performed on the multiple initial clusters, the workpieces within each initial cluster are pre-arranged and the minimum bounding polygon of each initial cluster is calculated to obtain the polygon area representing the arrangement efficiency, and the ratio of the sum of the inherent thermal load references of the workpieces within the initial cluster to the polygon area is used as the cluster heat density; the change in the system objective function caused by the migration of boundary workpieces to adjacent initial clusters is calculated, and the system objective function is positively correlated with the total polygon area of ​​each initial cluster and the variance of the cluster heat density; the migration operation that maximizes the decrease in the system objective function is repeatedly executed until convergence, generating the final cluster.

[0046] The calculated heterogeneous metric distance matrix is ​​input into a machine learning algorithm library, such as the KMedoids clustering algorithm, and the expected number of cluster centers is set to complete the initial cluster division.

[0047] For each initial cluster, a 2D layout engine based on the NoFitPolygon algorithm is invoked, combined with the BottomLeftFill heuristic strategy to tightly arrange the processed parts within the cluster. The convex hull polygon algorithm from the computational geometry library is then used to obtain the area of ​​the outer contour of the current layout as the polygon area. The cluster thermal density is calculated by summing the intrinsic thermal load baselines of all processed parts within the cluster and dividing by the polygon area.

[0048] Establish a system objective function, which is composed of the sum of the polygon areas of all initial clusters and the variance of the cluster heat density of all initial clusters, through a linear combination of area penalty factor and variance penalty factor.

[0049] A greedy local search algorithm is used to traverse the processed parts at the boundary of each cluster, simulate moving them into all physical adjacent clusters, and re-execute the pre-layout and polygon area calculation to obtain a new total area and cluster heat density variance. For example, after normalizing the two, a weighted sum is obtained to obtain the change in the system objective function.

[0050] Select the single workpiece migration operation that maximizes the absolute decrease in the system objective function value, update the cluster member assignment, and continue to repeat this evaluation and migration loop until the decrease in the system objective function is less than the set convergence threshold for five consecutive iterations. Then stop the iteration and output the final cluster.

[0051] In some implementations, the workpieces within each initial cluster are pre-layouted and the minimum bounding polygon of each initial cluster is calculated, including: After pre-layouting the processed parts within the same initial cluster, extract the outer boundary contour vertices of all processed parts within the initial cluster to construct a two-dimensional discrete point set of the initial cluster; The convex hull algorithm is used to calculate the convex hull polygon of a two-dimensional discrete point set; Use the convex hull polygon as the minimum bounding polygon of the initial cluster, and calculate the coverage area of ​​the minimum bounding polygon.

[0052] For several workpieces that have been clustered into the same initial cluster, the number is usually in the range of 2 to 10, with 5 workpieces as an example. The bottom-left heuristic translation and discrete angle rotation method is used to complete the compact pre-layout within the cluster and lock the pose of each workpiece in the local coordinate system of the cluster.

[0053] Iterate through the contour data objects of these 5 processed parts, extract all contour vertex data on their outer boundaries, filter out the point sets of internal hole topological voids, remove overlapping and collinear redundant points caused by splicing, and merge to generate a global two-dimensional discrete point set with a size between 100 and 500 nodes, with an example size of 240 nodes. .

[0054] During the spatial envelope model construction phase, the algorithm engine imports the Graham scan method to compute the set. Convex polygon bounding box: Obtaining the point set The minimum point of the axial coordinates is used as the origin of the initial polar axis. The remaining coordinate points in the set are sorted in ascending order according to the polar angle of the line connecting them to the origin. The cross product is then calculated iteratively using a data stack containing three points. If a clockwise deflection occurs, a stack pop operation is performed to extract the outer convex hull vertex sequence sorted counterclockwise.

[0055] The convex hull polygon is determined as the smallest bounding polygon of the initial cluster in the overall layout. Using the Shoelace formula, following the logic of calculating the determinant of polygon vertices: Calculate the area of ​​the closed geometric region enclosed by all convex hull boundaries.

[0056] S3 arranges the final clusters and, based on the cluster heat density, adjusts the cutting speed or cooling time within the corresponding final clusters.

[0057] A genetic algorithm is used to optimize the global optimal position and rotation angle of the workpieces within each final cluster. A nested collision detection algorithm based on the separating axis theorem is called to verify the validity of the positions, and the two-dimensional coordinate transformation matrix of the compact layout scheme is output.

[0058] During the CNC code (GCode) generation stage, the baseline cutting speed and standard cooling time are set. The cluster heat density values ​​of each final cluster calculated in the previous step are read, and a piecewise linear interpolation algorithm is used to establish the mapping relationship between cluster heat density and control coefficient.

[0059] When the cluster heat density exceeds the preset safety threshold limit, the cutting speed command F value of the CNC machining spindle is increased proportionally to the excess difference, or a delay pause command G04 is inserted in the middle of the cutting transition path segment of adjacent processed parts in the same final cluster, and the delay seconds are calculated according to the positive mapping of the cluster heat density, thereby increasing the cooling time, reducing local heat accumulation, and preventing thermal deformation of the aluminum plate.

[0060] In some implementations, the final clusters are formatted and laid out, including: Arrange the final clusters in descending order of their minimum bounding polygon area; Within the blank aluminum plate area, the smallest bounding polygons of each final cluster are arranged in a two-dimensional, non-overlapping, compact manner according to the descending order of area. After determining the global position of each final cluster on the blank aluminum plate area, the local non-overlapping layout of the processed parts within the final cluster is carried out within the smallest bounding polygon area corresponding to each final cluster, while maintaining the relative position determined by the pre-layout.

[0061] The execution scheduler extracts all generated final cluster entity data structures, iterates through and reads the pre-stored minimum bounding polygon area parameter of the convex hull of each cluster object, and calls the quicksort algorithm to reconstruct these cluster lists in descending order of absolute area, obtaining clusters starting from the largest area of ​​0.5. To the minimum area A descending order processing queue.

[0062] Within a pre-defined two-dimensional blank aluminum plate grid, the standard example dimension is the width. × Length The global origin is located at the lower left corner coordinate (0,0). The convex hull polygon with the largest area at the head of the queue is taken out for material arrangement. Based on the critical polygon layout algorithm, the polygon is used as a rigid motion unit to find the placement point in the available area of ​​the board.

[0063] When placing macroscopically, by setting A fixed rotation angle step is used for circumferential pose probing, and anti-interference constraints are added to keep the normal boundary gap between any two convex hull polygons within a safe buffer distance threshold, with a preferred range. to Example value Inside, the components are sequentially attached to the interior of the aluminum plate, with the goal of locking the polygonal position by occupying the minimum area increase of the top outline of the plate, thus achieving non-overlapping and compact splicing at the plate level.

[0064] After mapping the origin and tilt angle of all outer polygon coordinates to the absolute coordinate system of the aluminum plate, the underlying system removes the local envelope constraint corresponding to each polygon and reads the local offset matrix and rotation quaternion generated during pre-layout of the five parts in the corresponding cluster. Using a standard affine homogeneous transformation matrix, the workpiece is analytically unfolded into the internal physical mesh defined by the current convex hull fence, completing the reconstruction and generation of the nested blanking diagram while ensuring that the local relative positions remain unchanged.

[0065] In some implementations, the cutting rate or cooling time within the final cluster is controlled based on the cluster heat density, including: Determine whether the cluster heat density of the target final cluster exceeds the first set baseline. If it does, calculate the cutting speed increase based on the excess ratio. If the increased speed does not exceed the maximum cutting speed, control the cutting head to run at the increased speed when running within the target final cluster. If it exceeds the maximum cutting speed, then run at the maximum cutting speed. Determine whether the cluster heat density of the target final cluster exceeds the second set baseline. The second set baseline is greater than the first set baseline. If it exceeds the second set baseline, then after the cutting of each processed part in the target final cluster is completed, a cooling waiting time proportional to the amount of cluster heat density exceeding the standard is inserted.

[0066] The parameter evaluation module calls the sum of the inherent thermal load references of all parts within the target final cluster, divides it by the area of ​​the smallest circumscribed polygon, and calculates the measured value of the cluster thermal density. A two-stage threshold is preset based on the measured thermal expansion curve of a plate of a specific thickness, wherein the first threshold is set as a baseline. The preferred range is between 10 Up to 25 The example value is 15. Second, set the baseline. Greater than the first set baseline, the preferred range is between 30. Up to 50 The example value is 35. .

[0067] Execution layer comparison and The difference, when When the thermal density is determined to be too high, the excess ratio is calculated in real time. Based on the preset acceleration response coefficient The preferred range is 0.1 to 0.5, with an example value of 0.2, combined with the baseline cutting velocity of the current mapping in this region. For example, 8000 mm / min, calculate the compensated acceleration increment. If the speed of the output commands is increased after the speed-up correction... Less than or equal to the maximum cutting speed limited by the machine tool hardware For example, 12000 mm / min will be achieved by overwriting the F instruction word in the G code of the target cluster at the bottom layer of the CNC system to make it follow the command. Execute. If exceeded, output via hard saturation. .

[0068] If, during the decision-making process for implementing acceleration actions, the system detects that the heat density in the area meets the requirements... If the reading exceeds the second set baseline, it indicates that simply increasing the speed is no longer sufficient to alleviate the regional overheating. The CNC post-processing module will then enter a light-off intervention mode to calculate the absolute value of the cluster heat density exceeding the standard. And multiply by the value set to 1.2 The residence time is determined by a cooling conversion coefficient preferably ranging from 0.5 to 2.0. .

[0069] At this point, after the "M05 Turn Off Laser" command line at the end of all cut-off paths within the cluster, a delayed G-code such as "G04X[T_{cool}]" is implanted. This mechanical pause, where the actuator remains stationary and gas is continuously purged for several seconds, slows down the transfer and dissipation of heat within the cluster.

[0070] The experiment selected the maximum output power as The maximum acceleration limit for the laser cutting machine tool is set to The processing object is long ,Width And the thickness is The 6061 aluminum alloy sheet maintains a constant indoor temperature. The test task included 100 machined parts of different sizes and complexities, including 20 easily deformable parts with high-heat sharp corners. The experiment was divided into three groups: the control group used a traditional layout algorithm based on pure geometric contour matching; the ablation group, based on the inherent thermal load benchmark, performed thermal clustering and spatial layout; and the experimental group, based on the ablation group, fully implemented a cutting speed and cooling time control mechanism based on cluster heat density.

[0071] The measured data extracted after the test run showed that the utilization rate of the sampled material in the control group was [missing information]. The average local thermal deformation is as high as The overall scrap rate reached The total processing time was 42 minutes. The ablation group used heat-resistant clustering sensitive states to physically isolate and arrange the high-heat-processing parts, slightly reducing the overall sheet metal utilization rate. However, the average thermal deformation decreased to The scrap rate fell back to The total processing time was 43 minutes. After implementing coordinated hardware and software control, the utilization rate of the board material in the experimental group remained at [percentage missing]. The average local thermal deformation was compressed to The scrap rate dropped to Meanwhile, because the acceleration mechanism offset the delay caused by cooling intervention, the overall processing time was reduced to 38 minutes. The processing times of the three experimental schemes are compared as follows: Figure 3 As shown, the differences in the impact of different typesetting strategies and thermal control mechanisms on the overall processing cycle are presented.

[0072] The ablation group's results demonstrate that heterogeneity measurement can avoid large-area accumulation of high heat sources in a planar plane, thereby reducing the risk of primary thermal deformation. Building upon this, the experimental group utilized speed regulation and dwell time delay mechanisms to intervene when the local heat density exceeds the warning line, blocking the cumulative effect along the heat conduction path and reducing the product scrap rate compared to a purely geometric algorithm. This control solution not only improves product processing precision with minimal material utilization loss, but also shortens the system's working cycle through adaptive compensation of cutting speed, achieving a dual improvement in production efficiency and quality yield.

[0073] On the other hand, the present invention also provides a cutting and layout planning system for aluminum sheet processed parts. For example... Figure 4 The aluminum plate processing part cutting and layout planning system shown includes a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement the aluminum plate processing part cutting and layout planning method according to the first aspect of the present invention.

Claims

1. A method for cutting and layout planning of aluminum plate processed parts, characterized in that, include: Obtain the contour data and process parameters of the parts to be processed; calculate the feature vector of each part, the feature vector including a concavity descriptor representing the concavity feature of the contour, and an inherent thermal load reference calculated based on the cutting path length, the number of sharp corners and the unit heat input. Based on the feature vector, a heterogeneity metric is constructed between the processed parts. The heterogeneity metric is obtained by weighting the difference value of the normalized concavity descriptor and the difference value of the normalized inherent thermal load benchmark. The weight of the difference value of the normalized inherent thermal load benchmark is positively correlated with the mean of the inherent thermal load benchmark of the two processed parts. The workpiece is divided into multiple initial clusters based on the heterogeneity metric. Multiple initial clusters are iteratively optimized. The workpieces in each initial cluster are pre-arranged and the minimum bounding polygon of each initial cluster is calculated. The area of ​​the polygon representing the arrangement efficiency is obtained, and the ratio of the sum of the inherent thermal load references of the workpieces in the initial cluster to the area of ​​the polygon is used as the cluster thermal density. Calculate the change in the system objective function caused by migrating the boundary workpiece to the adjacent initial cluster. The system objective function is positively correlated with the total area of ​​the polygons of each initial cluster and the variance of the cluster heat density. Repeatedly execute the migration operation that maximizes the decrease in the system objective function until convergence is achieved, generating the final cluster. The final clusters are arranged and laid out, and the cutting speed or cooling time within the corresponding final clusters is adjusted based on the cluster heat density.

2. The method for cutting and layout planning of aluminum plate processed parts according to claim 1, characterized in that, Obtain the contour data and process parameters of the part to be processed, including: The processing task package containing 3D model data is obtained from the production management server, and the 2D contour point set mapped from the 3D model to the cutting plane is extracted. The corresponding cutting process parameters are extracted from the equipment database, including the maximum cutting speed, constant output laser power, and safe operation threshold parameters.

3. The method for cutting and layout planning of aluminum plate processed parts according to claim 2, characterized in that, Retrieve the machining task package containing 3D model data from the production management server and extract the 2D contour point set, including: The system connects to the server via a communication interface to download a 3D model processing task package in STEP or IGES format, extracts the model edge data, projects the 3D boundary curve into a 2D contour according to the set cutting plane, and uses discretization processing to generate a sequence of 2D contour points containing X and Y coordinates, where the sampling interval between adjacent coordinate points is 0.1mm to 0.5mm. The safe operation threshold parameters include the maximum acceleration limit of the machine tool and the safe avoidance lifting height of the follow-up cutting head during idle movement.

4. The method for cutting and layout planning of aluminum plate processed parts according to claim 1, characterized in that, The inherent thermal load reference, calculated based on the cutting path length, the number of sharp corners, and the unit heat input, includes: Traverse the contour data of the workpiece, extract the turning points where the included angle between adjacent line segments is less than the preset acute angle threshold, and count the total number of sharp corners of the workpiece. Calculate the sum of the lengths of all line segments of the two-dimensional contour of the workpiece to obtain the total length of the cutting path; The inherent thermal load reference of the workpiece is obtained by multiplying the total length of the cutting path and the total number of sharp corners by the preset unit heat input and the sharp corner heat concentration coefficient, respectively, and then summing them.

5. The method for cutting and layout planning of aluminum plate processed parts according to claim 1, characterized in that, Constructing a heterogeneity metric between processed parts based on the feature vectors includes: The absolute value of the difference between the concavity descriptors of the two machined parts is calculated and normalized as the first metric; the absolute value of the difference between the inherent thermal load references of the two machined parts is calculated and normalized as the second metric. The average value of the inherent thermal load reference of the two processed parts is calculated. When the average value is less than the set material thermal sensitivity threshold, the first and second metric indicators are combined with fixed weights. When the average value is greater than or equal to the material thermal sensitivity threshold, the weight of the second metric indicator increases exponentially with the ratio of the difference between the average value and the material thermal sensitivity threshold to the material thermal sensitivity threshold, and the heterogeneity metric is calculated.

6. The method for cutting and layout planning of aluminum plate processed parts according to claim 1, characterized in that, The workpiece is divided into multiple initial clusters based on the heterogeneity metric and iteratively optimized, including: Input the heterogeneous metric distance matrix into the clustering algorithm, set the number of cluster centers to complete the initial cluster division; establish the system objective function, which is composed of the sum of the polygon areas of all initial clusters and the variance of the cluster heat density of all initial clusters through a linear combination of area penalty factor and variance penalty factor; A greedy local search algorithm is used to traverse the processed parts at the boundary of each initial cluster. The algorithm simulates moving the processed parts at the boundary into the adjacent initial cluster and re-performing the pre-layout and polygon area calculation to obtain the change in the system objective function. The migration operation that maximizes the absolute value of the decrease in the system objective function is selected to update the cluster member allocation. The iteration stops when the decrease in the system objective function is less than the convergence threshold after a set number of iterations, and the final cluster is output.

7. The method for cutting and layout planning of aluminum plate processed parts according to claim 6, characterized in that, Pre-layout of the workpieces within each initial cluster and calculation of the minimum bounding polygon of each initial cluster, including: After pre-layouting the processed parts within the same initial cluster, extract the outer boundary contour vertices of all processed parts within the initial cluster to construct a two-dimensional discrete point set of the initial cluster; The convex hull algorithm is used to calculate the convex hull polygon of the two-dimensional discrete point set; The convex hull polygon is used as the minimum bounding polygon of the initial cluster, and the coverage area of ​​the minimum bounding polygon is calculated.

8. The method for cutting and layout planning of aluminum plate processed parts according to claim 1, characterized in that, The final clusters will be formatted and laid out, including: Arrange the final clusters in descending order of their minimum bounding polygon area; within the blank aluminum plate area, arrange the minimum bounding polygons of each final cluster as a whole in a two-dimensional, non-overlapping, compact arrangement according to the descending order of area. After determining the global position of each final cluster on the blank aluminum plate area, the local non-overlapping layout of the processed parts within the final cluster is carried out within the smallest bounding polygon area corresponding to each final cluster, while maintaining the relative position determined by the pre-layout.

9. The method for cutting and layout planning of aluminum plate processed parts according to claim 1, characterized in that, Based on the cluster heat density, the cutting speed or cooling time within the corresponding final cluster is adjusted, including: Determine whether the cluster heat density of the target final cluster exceeds the first set baseline. If it does, calculate the cutting speed increase based on the excess ratio. If the increased speed does not exceed the maximum cutting speed, control the cutting head to run at the increased speed when running within the target final cluster. If it exceeds the maximum cutting speed, then run at the maximum cutting speed. Determine whether the cluster heat density of the target final cluster exceeds the second set baseline, which is greater than the first set baseline. If it does, then after the cutting of each processed part in the target final cluster is completed, a cooling waiting time proportional to the amount of cluster heat density exceeding the standard is inserted.

10. A cutting and layout planning system for aluminum plate processed parts, characterized in that, It includes a processor and a memory, the memory storing computer program instructions, which, when executed by the processor, implement the cutting and layout planning method for aluminum plate processing parts according to any one of claims 1 to 9.

Images