Double-head ultraviolet laser cutting control method and system

Abstract

The invention provides a double-head ultraviolet laser cutting control method and system, and relates to the technical field of laser cutting, and the method comprises the following steps: obtaining geometrical characteristics of a workpiece, constructing a three-dimensional constraint model, and forming a multi-dimensional hierarchical decomposition tree; generating an initial cutting path and evaluating the priorities of the subtasks; in the execution process, a dynamic interference trend is pre-judged, and an obstacle avoidance rule set is established for path fine adjustment and working area reconstruction; and according to the machining difficulty, partition planning is conducted, partition coupling relation parameters are derived, and cutting parameters and movement speed are configured. Intelligent cooperative control of the double-end ultraviolet laser is achieved, the cutting efficiency and quality are improved, and interference and collision are avoided.

Description

Technical Field

[0001] This invention relates to laser cutting technology, and more particularly to a dual-head ultraviolet laser cutting control method and system. Background Technology

[0002] Ultraviolet laser cutting technology is widely used in precision machining fields such as electronics, semiconductors, and medical devices due to its high precision, low heat-affected zone, and good material adaptability. As industrial production demands for processing efficiency continue to increase, dual-head ultraviolet laser cutting systems are gradually becoming an important solution for improving production efficiency. This technology, by having two laser cutting heads work simultaneously, can significantly reduce processing time and improve production efficiency.

[0003] Traditional dual-head ultraviolet laser cutting systems primarily employ static work area division or simple path planning methods. These methods typically plan a pre-defined cutting path based on the workpiece's geometry, fixing the work area into two parts, each processed by a separate laser head. However, this static cutting approach cannot adapt to the dynamic changes required during the processing of complex workpieces.

[0004] Existing technologies lack the ability to adaptively plan cutting paths based on workpiece geometry. Traditional methods often employ fixed cutting path patterns and cannot perform intelligent path optimization based on the complex geometry of the workpiece, resulting in unstable cutting quality and insufficient precision when machining complex-shaped workpieces.

[0005] Dual-head lasers are prone to dynamic interference problems during simultaneous operation. Existing technologies mainly rely on preset safety distances and simple obstacle avoidance strategies, lacking the ability to monitor and predict interference trends in real time. This makes it impossible to effectively avoid the risk of collision between the two laser heads during high-speed processing, affecting the safety and reliability of the system. Summary of the Invention

[0006] The present invention provides a dual-head ultraviolet laser cutting control method and system, which can solve the problems in the prior art.

[0007] A first aspect of the present invention provides a dual-head ultraviolet laser cutting control method, comprising: Obtain the geometric feature information of the workpiece to be processed, construct a three-dimensional constraint model of the processing space based on the geometric feature information, and construct a multi-dimensional hierarchical decomposition tree of the cutting task of the workpiece to be processed based on the three-dimensional constraint model. Based on the spatial distribution characteristics of the sub-task units in the multi-dimensional hierarchical decomposition tree and their positional relationship in the three-dimensional constraint model, the initial cutting path of the dual-head ultraviolet laser is generated, and the priority of each sub-task unit is evaluated according to the workpiece processing characteristics. During the execution of the initial cutting path, the dynamic interference trend between the dual-headed ultraviolet lasers is predicted by the multi-dimensional hierarchical decomposition tree. An obstacle avoidance rule set is established in combination with the priority evaluation results of each sub-task unit. The cutting path of the dual-headed ultraviolet lasers is fine-tuned in real time and the working area is dynamically reconstructed according to the obstacle avoidance rule set. The reconstructed working area is divided into zones according to processing difficulty. The zone coupling relationship parameters are derived from the operating status of the dual-head ultraviolet laser. Cutting parameters and movement speed are configured according to the zone coupling relationship parameters to guide the dual-head ultraviolet laser to complete the collaborative cutting of each zone under quality constraints.

[0008] Based on the geometric feature information, a three-dimensional constraint model of the processing space is constructed, and based on the three-dimensional constraint model, a multi-dimensional hierarchical decomposition tree for the cutting task of the workpiece to be processed is constructed, including: Extract the set of cutting contour segments and the set of hole position coordinates from the geometric feature information, map the set of cutting contour segments and the set of hole position coordinates to the motion coordinate system of the dual-head ultraviolet laser, and determine the independent reachable spatial boundary of each laser based on the motion range of the dual-head ultraviolet laser; Based on the independent reachable space boundary, the spatial overlap region of the dual-head ultraviolet laser during its movement is calculated, and the spatial overlap region and the independent reachable space boundary together constitute a three-dimensional constraint model of the processing space. In the three-dimensional constraint model, the spatial overlapping area is identified as the interference risk constraint area, and the independent reachable spatial boundary is identified as the laser partition constraint boundary. Based on the interference risk constraint area and the laser partition constraint boundary, the set of cutting contour lines is hierarchically decomposed according to the spatial position belonging relationship to construct a multi-dimensional hierarchical decomposition tree of the cutting task of the workpiece to be processed.

[0009] The spatial overlap region of the dual-head ultraviolet laser during its movement is calculated based on the independent reachable space boundary, and the three-dimensional constraint model of the processing space is formed by the spatial overlap region and the independent reachable space boundary together, including: Obtain the motion stroke parameters and installation position parameters of each laser in the dual-head ultraviolet laser. Based on the motion stroke parameters and installation position parameters, calculate the maximum spatial range that the motion endpoint of each laser can reach in the processing space, and define the maximum spatial range as the independent reachable spatial boundary of each laser. Extract the spatial geometric description information of the independent reachable space boundary, and identify the intersection of the two independent reachable space boundaries of the dual-head ultraviolet laser during its movement through spatial geometric operations based on the spatial geometric description information, and determine the intersection as the spatial overlapping region; Obtain the spatial coordinate range and volume parameters of the spatially overlapping region, associate and map the spatial coordinate range with the spatial geometric description information, and based on the association mapping, construct a three-dimensional constraint model of the processing space together with the spatially overlapping region and the independent reachable spatial boundary.

[0010] Based on the spatial distribution characteristics of the sub-task units in the multidimensional hierarchical decomposition tree and their positional relationships in the three-dimensional constraint model, the initial cutting path of the dual-head ultraviolet laser is generated. Simultaneously, priority evaluation of each sub-task unit is performed according to the workpiece processing characteristics, including: Extract the center point coordinates and contour envelope size of each sub-task unit from the multi-dimensional hierarchical decomposition tree; perform spatial position matching between the center point coordinates and the spatial overlapping area in the three-dimensional constraint model to determine whether each sub-task unit is located within the spatial overlapping area. Based on the spatial location matching results, for sub-task units located within the spatial overlap region, the shortest distance between them and the boundary of the spatial overlap region is calculated as an interference risk measurement parameter; for sub-task units not located within the spatial overlap region, the spatial distance from their center point coordinates to the current position of the dual-head ultraviolet laser is calculated. The spatial distribution feature vector of each subtask unit is constructed by combining the interference risk measurement parameter with the spatial distance value and the contour envelope size. The priority of each subtask unit is evaluated based on the spatial distribution feature vector. At the same time, the subtask units with the interference risk measurement parameter less than the preset interference threshold are assigned priority values. Based on the priority values, each subtask unit is assigned to the working sequence corresponding to the dual-head ultraviolet laser, and the center point coordinates of each subtask unit are connected sequentially according to the priority values ​​of the subtask units in the working sequence to generate the initial cutting path of the dual-head ultraviolet laser.

[0011] The dynamic interference trend between the dual-headed ultraviolet lasers is predicted by the multi-dimensional hierarchical decomposition tree. An obstacle avoidance rule set is established based on the priority evaluation results of each sub-task unit. Real-time fine-tuning of the cutting path of the dual-headed ultraviolet lasers and dynamic reconstruction of the working area are performed according to the obstacle avoidance rule set, including: The dynamic interference trend between the dual-headed ultraviolet lasers is predicted by the multidimensional hierarchical decomposition tree. Based on the dynamic interference trend, a relative motion feature sequence of the dual-headed ultraviolet lasers is constructed. The relative motion feature sequence is quantified into an interference risk space region in the processing space coordinate system. Based on the relative motion feature sequence, the minimum safe distance between the dual-headed ultraviolet lasers is deduced. The minimum safe distance is used to derive the judgment criterion for the interference risk space region. The sequence of sub-task units associated with the spatial region of interference risk is analyzed, and the spatial distribution characteristics of the sub-task unit sequence are integrated with the minimum safe distance to generate a set of potential conflict points prediction. The hierarchical weight allocation scheme of the sub-task unit sequence is constructed by combining the priority evaluation results of each sub-task unit. An obstacle avoidance rule set including spatial avoidance strategy and temporal avoidance strategy is arranged according to the hierarchical weight allocation scheme. The cutting path of the subordinate sub-task unit is optimized in real time according to the spatial avoidance strategy. When the optimized cutting path still has the risk of interference, the execution sequence of the subordinate sub-task unit is reorganized according to the temporal avoidance strategy. If the interference risk cannot be eliminated by both the spatial avoidance strategy and the temporal avoidance strategy, the working area of ​​the dual-head ultraviolet laser is dynamically reconstructed based on the obstacle avoidance rule set.

[0012] The reconstructed work area is divided into zones according to processing difficulty. Zone coupling parameters are derived from the operating status of the dual-head ultraviolet laser. Cutting parameters and motion speed are configured based on these parameters to guide the dual-head ultraviolet laser in completing the collaborative cutting of each zone under quality constraints. Obtain the geometric complexity features and material property features of each sub-task unit within the reconstructed work area. Analyze the geometric complexity features and material property features through a feature fusion network to derive the processing difficulty assessment value of each sub-task unit. Based on the processing difficulty assessment value, construct the partitioning planning result of the difficulty level of the reconstructed work area. Based on the partitioning planning results, the operating load state and temporal distribution state of the dual-head ultraviolet laser are deduced. According to the feature fusion network, the operating load state and the temporal distribution state are respectively matched with the processing difficulty assessment value of each partition to generate partition coupling relationship parameters. The cutting parameters and motion speed of each partition are quantified based on the partition coupling relationship parameters. At the same time, the combination configuration scheme of laser power and pulse frequency is derived based on the results of the depth feature matching. The combination configuration scheme is then transformed into a collaborative execution command of the dual-head ultraviolet laser, guiding the dual-head ultraviolet laser to complete the collaborative cutting of each partition under quality constraints.

[0013] Based on the partitioning planning results, the operating load state and temporal distribution state of the dual-head ultraviolet laser are deduced. Using the feature fusion network, the operating load state and temporal distribution state are respectively matched with the processing difficulty assessment value of each partition using deep feature matching, generating partition coupling relationship parameters including: Based on the zoning planning results, the operating load state sequence and the time-series distribution state sequence of the dual-head ultraviolet laser are generated. The operating load state sequence and the time-series distribution state sequence are respectively constructed as operating state feature vectors within a dynamic time window. The spatiotemporal dimension of the operating state feature vectors is decoupled by a feature fusion network. The processing difficulty assessment value of each partition is extracted from the decoupled running state feature vector, and the mapping relationship matrix between the processing difficulty assessment value and the running state feature vector is established using the feature fusion network. The mapping relationship matrix is ​​iteratively optimized based on the deep feature matching algorithm to obtain the degree of mutual influence between each partition. The calculation results of the deep feature matching algorithm are marked as the significantly influential associated partitions, and partition coupling relationship parameters are generated.

[0014] A second aspect of the present invention provides a dual-head ultraviolet laser cutting control system, comprising: The acquisition module is used to acquire the geometric feature information of the workpiece to be processed, construct a three-dimensional constraint model of the processing space based on the geometric feature information, and construct a multi-dimensional hierarchical decomposition tree of the cutting task of the workpiece to be processed based on the three-dimensional constraint model. The evaluation module is used to generate the initial cutting path of the dual-head ultraviolet laser based on the spatial distribution characteristics of the sub-task units in the multi-dimensional hierarchical decomposition tree and their positional relationship in the three-dimensional constraint model, and to evaluate the priority of each sub-task unit according to the workpiece processing characteristics. The reconstruction module is used to predict the dynamic interference trend between the dual-head ultraviolet lasers through the multi-dimensional hierarchical decomposition tree during the execution of the initial cutting path, establish an obstacle avoidance rule set in combination with the priority evaluation results of each sub-task unit, and perform real-time fine-tuning of the cutting path of the dual-head ultraviolet laser and dynamic reconstruction of the working area according to the obstacle avoidance rule set. The cutting module is used to divide the reconstructed working area into zones according to the processing difficulty. The operating status of the dual-head ultraviolet laser is used to derive the zone coupling relationship parameters. The cutting parameters and movement speed are configured according to the zone coupling relationship parameters to guide the dual-head ultraviolet laser to complete the collaborative cutting of each zone under quality constraints.

[0015] A third aspect of the present invention provides an electronic device, comprising: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0016] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0017] The beneficial effects of this application are as follows: By constructing a three-dimensional constraint model of the processing space and forming a multi-dimensional hierarchical decomposition tree, accurate modeling of workpieces with complex geometric features and systematic decomposition of cutting tasks are achieved, thereby improving the planning accuracy of dual-head ultraviolet laser cutting.

[0018] An initial cutting path is generated based on spatial distribution characteristics and positional relationships. A priority evaluation mechanism is used to make the cutting process more targeted and efficient, significantly improving processing efficiency.

[0019] The innovative approach of using a multi-dimensional hierarchical decomposition tree to predict the dynamic interference trend between dual-head lasers, and combining the priority evaluation results to establish an obstacle avoidance rule set, enables real-time fine-tuning of the cutting path and dynamic reconstruction of the working area, effectively solving the interference problem of dual-head ultraviolet lasers in collaborative operation.

[0020] By partitioning the reconstructed working area and introducing partition coupling parameters, intelligent configuration of cutting parameters and motion speed is achieved, enabling the dual-head ultraviolet lasers to work efficiently and collaboratively while ensuring cutting quality, significantly improving the system's adaptability and cutting accuracy. Attached Figure Description

[0021] Figure 1 This is a flowchart illustrating the dual-head ultraviolet laser cutting control method according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating the generation of partitioned coupling relationship parameters for a dual-head ultraviolet laser according to an embodiment of the present invention. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0024] Figure 1This is a flowchart illustrating the dual-head ultraviolet laser cutting control method according to an embodiment of the present invention, as shown below. Figure 1 As shown, the method includes: Obtain the geometric feature information of the workpiece to be processed, construct a three-dimensional constraint model of the processing space based on the geometric feature information, and construct a multi-dimensional hierarchical decomposition tree of the cutting task of the workpiece to be processed based on the three-dimensional constraint model. Based on the spatial distribution characteristics of the sub-task units in the multi-dimensional hierarchical decomposition tree and their positional relationship in the three-dimensional constraint model, the initial cutting path of the dual-head ultraviolet laser is generated, and the priority of each sub-task unit is evaluated according to the workpiece processing characteristics. During the execution of the initial cutting path, the dynamic interference trend between the dual-headed ultraviolet lasers is predicted by the multi-dimensional hierarchical decomposition tree. An obstacle avoidance rule set is established in combination with the priority evaluation results of each sub-task unit. The cutting path of the dual-headed ultraviolet lasers is fine-tuned in real time and the working area is dynamically reconstructed according to the obstacle avoidance rule set. The reconstructed working area is divided into zones according to processing difficulty. The zone coupling relationship parameters are derived from the operating status of the dual-head ultraviolet laser. Cutting parameters and movement speed are configured according to the zone coupling relationship parameters to guide the dual-head ultraviolet laser to complete the collaborative cutting of each zone under quality constraints.

[0025] In one optional implementation, constructing a three-dimensional constraint model of the processing space based on the geometric feature information, and constructing a multi-dimensional hierarchical decomposition tree of the cutting task of the workpiece to be processed based on the three-dimensional constraint model includes: Extract the set of cutting contour segments and the set of hole position coordinates from the geometric feature information, map the set of cutting contour segments and the set of hole position coordinates to the motion coordinate system of the dual-head ultraviolet laser, and determine the independent reachable spatial boundary of each laser based on the motion range of the dual-head ultraviolet laser; Based on the independent reachable space boundary, the spatial overlap region of the dual-head ultraviolet laser during its movement is calculated, and the spatial overlap region and the independent reachable space boundary together constitute a three-dimensional constraint model of the processing space. In the three-dimensional constraint model, the spatial overlapping area is identified as the interference risk constraint area, and the independent reachable spatial boundary is identified as the laser partition constraint boundary. Based on the interference risk constraint area and the laser partition constraint boundary, the set of cutting contour lines is hierarchically decomposed according to the spatial position belonging relationship to construct a multi-dimensional hierarchical decomposition tree of the cutting task of the workpiece to be processed.

[0026] Geometric feature information is extracted from the design drawings or digital models of the workpiece to be processed, resulting in a set of cutting contour lines containing all straight and curved segments of the workpiece's outer contour boundary and internal hollow areas. Each line segment records its start-point coordinates, end-point coordinates, and required cutting depth. A set of hole coordinates contains the location information of all laser-drilled holes on the workpiece, recording the center coordinates, diameter, and machining accuracy requirements for each hole. The extraction accuracy of the geometric feature information is determined based on the workpiece material thickness and cutting accuracy requirements. For thin sheets of material with a thickness of 0.1 mm, the coordinate accuracy is maintained at 0.005 mm; for materials thicker than 1 mm, the coordinate accuracy can be relaxed to 0.02 mm.

[0027] Mapping the set of cutting contour lines and hole coordinates to the motion coordinate system of the dual-head ultraviolet laser is achieved through coordinate system transformation, establishing the correspondence between the workpiece coordinate system and the machine coordinate system. The workpiece coordinate system uses the geometric center of the workpiece as its origin, while the machine coordinate system uses the mechanical origin of the dual-head laser's motion platform as its origin. During the mapping process, the placement and orientation of the workpiece on the machining table must be considered. The coordinate transformation relationship is determined by measuring the actual positions of known feature points on the workpiece in the machine coordinate system. The transformation accuracy directly affects the cutting quality; the positional deviation should be controlled within 10% of the required machining accuracy.

[0028] The independent reachable space boundaries of each laser are determined based on the travel range of the dual-head ultraviolet laser. The independent reachable space of the first laser head is determined by the travel limitations of its mechanical structure and motion mechanism, including the maximum range of motion in the X, Y, and Z axes. The second laser head has the same motion capability but a different installation position; its independent reachable space in the machine coordinate system has a fixed offset relative to the first laser head. The independent reachable space boundary of each laser is represented by a three-dimensional region enclosed by six planes, which correspond to the positive and negative limit positions of motion in each axis. Safety margin settings ensure that the laser head will not exceed the safe range due to positioning errors or mechanical vibrations when moving near the boundaries.

[0029] The spatial overlap region of a dual-head ultraviolet laser during its motion is calculated based on the boundary of its independently accessible space. This is obtained by analyzing the geometric overlap of the independently accessible spaces of the two laser heads. The calculation of the spatial overlap region needs to consider the intersection of the two three-dimensional regions in space. The boundary of the intersection region is determined by the intersection line of the boundary planes of the two independently accessible spaces. The volume and shape of the overlap region directly affect the collaborative working capability of the dual-head laser; a larger overlap region results in greater flexibility in collaborative processing, but also increases the risk of interference.

[0030] The overlapping spatial regions and the independent reachable spatial boundaries together constitute a three-dimensional constraint model of the processing space. This model fully describes the workspace limitations and constraints of the dual-head ultraviolet laser system. The three-dimensional constraint model consists of three main components: the independent working region of the first laser, the independent working region of the second laser, and the overlapping working region of the two. Each region in the model has clearly defined geometric boundaries and access permissions, providing a spatial constraint basis for subsequent path planning and conflict avoidance. The establishment of the constraint model needs to consider the dynamic characteristics of the laser during its movement, including acceleration limits, velocity variations, and positioning accuracy requirements.

[0031] In the 3D constraint model, overlapping spatial regions are designated as interference risk constraint regions, where two laser heads may experience motion conflicts or mutual interference. Interference risk identification is based on the physical dimensions of the laser heads, their motion trajectories, and safety distance requirements. Interference risk is considered to exist when the distance between the two laser heads within the overlapping region is less than a safety threshold. The safety threshold is set by comprehensively considering the laser head's external dimensions, motion accuracy, and emergency stop distance, and is typically set to at least 1.5 times the maximum external dimensions of the laser head. The identification of interference risk constraint regions provides constraints for subsequent path planning and real-time obstacle avoidance.

[0032] The independent reachable space boundaries are designated as laser zone constraint boundaries, dividing the entire workspace into responsibility areas for different lasers. The first laser zone constraint boundary encloses the independent reachable space of the first laser, within which only the first laser is allowed to operate; a second laser is prohibited from entering. The second laser zone constraint boundary is defined in the same way, ensuring that the two lasers can operate independently within their respective zones without conflict. The establishment of zone constraint boundaries simplifies the complexity of multi-laser coordinated control, providing clear spatial constraints for task allocation and path planning.

[0033] Based on the interference risk constraint region and the laser partition constraint boundary, the set of cutting contour line segments is hierarchically decomposed according to their spatial location. The decomposition process first determines which partition each cutting contour line segment completely falls into; segments entirely within a single partition are directly assigned to that partition. Line segments crossing partition boundaries need to be segmented at the boundary, and each segment is assigned to its corresponding partition. Hole coordinates are assigned according to the partition where their center is located; holes located near partition boundaries need to consider the safety distance requirements during laser head operation.

[0034] A hierarchical decomposition establishes a multi-level task organization structure. The top level represents the complete machining task for the entire workpiece, the middle level comprises sets of sub-tasks for each partition, and the bottom level consists of specific line segment cutting or hole machining tasks. Each level of task contains explicit geometric information, machining parameters, and constraints. The hierarchical relationship between tasks reflects the logical dependency and spatial constraints of the machining sequence. The decomposition results consider the interdependencies of tasks; some tasks can only be executed after other tasks are completed. This dependency is expressed in the hierarchical structure through the connections between task nodes.

[0035] The multi-dimensional hierarchical decomposition tree of the workpiece cutting task organizes the hierarchical decomposition results into a tree-like data structure. The root node of the tree represents the entire workpiece processing task, and the child nodes represent the individual subtasks after decomposition. Each node contains the spatial location information of the task, processing parameter requirements, priority settings, and dependencies with other tasks. The multi-dimensional characteristic is reflected in the fact that the tree structure not only decomposes according to spatial location but also considers constraints of multiple dimensions such as processing time, difficulty level, and quality requirements. The decomposition tree provides a structured task description for task scheduling and path optimization of the dual-head ultraviolet laser, supporting parallel task identification and conflict detection. The dynamic updating capability of the tree structure allows for adjustments to task allocation and execution order during processing based on actual conditions.

[0036] In one optional implementation, calculating the spatial overlap region of the dual-headed ultraviolet laser during its movement based on the independent reachable space boundary, and then combining the spatial overlap region with the independent reachable space boundary to form a three-dimensional constraint model of the processing space includes: Obtain the motion stroke parameters and installation position parameters of each laser in the dual-head ultraviolet laser. Based on the motion stroke parameters and installation position parameters, calculate the maximum spatial range that the motion endpoint of each laser can reach in the processing space, and define the maximum spatial range as the independent reachable spatial boundary of each laser. Extract the spatial geometric description information of the independent reachable space boundary, and identify the intersection of the two independent reachable space boundaries of the dual-head ultraviolet laser during its movement through spatial geometric operations based on the spatial geometric description information, and determine the intersection as the spatial overlapping region; Obtain the spatial coordinate range and volume parameters of the spatially overlapping region, associate and map the spatial coordinate range with the spatial geometric description information, and based on the association mapping, construct a three-dimensional constraint model of the processing space together with the spatially overlapping region and the independent reachable spatial boundary.

[0037] In the dual-head ultraviolet laser, the motion travel parameters of each laser are read from the laser controller's configuration file, which stores the maximum travel distance of each laser in the X, Y, and Z axes. The motion travel parameters of the first laser include an X-axis travel of 600 mm, a Y-axis travel of 400 mm, and a Z-axis travel of 100 mm. The second laser has the same motion travel parameters. The accuracy of acquiring the motion travel parameters is required to be 0.1 mm. The encoder feedback signal of the laser servo driver determines the actual usable travel range. The motion travel parameters also include the maximum speed and acceleration limits for each axis: a maximum speed of 500 mm / s for the X-axis, 300 mm / s for the Y-axis, and 100 mm / s for the Z-axis. These parameters affect the calculation of the reachability of the laser's motion endpoints.

[0038] The installation position parameters for each laser need to be obtained from mechanical structure measurements and coordinate calibration. These parameters describe the fixed installation coordinates of each laser in the machine coordinate system. The installation position parameters for the first laser are: X-axis -200 mm, Y-axis -150 mm, Z-axis 200 mm; for the second laser, X-axis +200 mm, Y-axis -150 mm, Z-axis 200 mm. The measurement accuracy of these installation position parameters must be ensured to reach 0.05 mm using a coordinate measuring machine. The stability of the installation position directly affects the laser's working accuracy and the accuracy of spatial calculations. The installation position parameters include the laser's three-dimensional coordinate offset relative to the machine coordinate origin and its rotation angles around each axis. The measurement accuracy of the rotation angles must reach 0.1 degrees.

[0039] The maximum spatial range that the moving endpoints of each laser can reach in the processing space needs to be calculated by adding the motion stroke parameters and the installation position parameters. For the first laser, the maximum spatial range of the moving endpoint is from -500 mm to +100 mm on the X-axis, from -350 mm to +50 mm on the Y-axis, and from 100 mm to 300 mm on the Z-axis. The calculation method for the maximum spatial range of the second laser's moving endpoint is the same: from -100 mm to +500 mm on the X-axis, from -350 mm to +50 mm on the Y-axis, and from 100 mm to 300 mm on the Z-axis. The calculation of the maximum spatial range needs to consider the dynamic constraints during laser movement, including a safety margin to avoid collisions with mechanical structures, with a 10 mm safety distance reserved in each axis direction.

[0040] The maximum spatial extent is defined as the independent reachable space boundary of each laser. Each independent reachable space boundary is represented by a cuboid region enclosed by six planes. The independent reachable space boundary of the first laser is determined by these six planes: the negative X-axis boundary plane is located at -500 mm, the positive X-axis boundary plane is located at +100 mm, the negative Y-axis boundary plane is located at -350 mm, the positive Y-axis boundary plane is located at +50 mm, the lower Z-axis boundary plane is located at 100 mm, and the upper Z-axis boundary plane is located at 300 mm. The independent reachable space boundary of the second laser is defined in the same way, but the X-axis boundary position is different. The independent reachable space boundary is represented using plane normal vectors and distance parameters for ease of subsequent geometric calculations.

[0041] The spatial geometric description information of the independently reachable space boundary includes the vertex coordinates, side lengths, volume, and center coordinates of the boundary region. For the first laser, this includes eight vertex coordinates, a boundary region with an X-axis length of 600 mm, a Y-axis length of 400 mm, a Z-axis length of 200 mm, a volume of 48,000,000 cubic millimeters, and a geometric center coordinate of -200 mm X-axis, -150 mm Y-axis, and 200 mm Z-axis. The second laser's independently reachable space boundary has the same dimensional parameters but different locations. The accuracy of the spatial geometric description information is consistent with the calculation accuracy of the independently reachable space boundary, ensuring the accuracy of subsequent geometric calculations.

[0042] During the movement of a dual-headed ultraviolet laser, the intersection of two independently accessible spatial boundaries requires identification through cuboid region intersection calculations. The spatial geometric calculation process involves determining the overlap of the projections of the two cuboid regions along the three coordinate axes. The two regions intersect when there is overlap in all three axial projections. The overlap range along the X-axis is from -100 mm to +100 mm, along the Y-axis from -350 mm to +50 mm, and along the Z-axis from 100 mm to 300 mm. The intersection is geometrically a cuboid, with its boundaries defined by the inner planes of the two independently accessible spatial boundaries. The numerical precision of the spatial geometric calculations must reach 0.01 mm to ensure the accuracy of the intersection boundary calculations.

[0043] The intersecting portion is defined as the spatial overlapping region, which inherits all geometric features of the intersecting portion. The X-axis range of the spatial overlapping region is from -100 mm to +100 mm, the Y-axis range is from -350 mm to +50 mm, and the Z-axis range is from 100 mm to 300 mm. The region has a length of 200 mm, a width of 400 mm, and a height of 200 mm, with a total volume of 16,000,000 cubic millimeters. The geometric center of the spatial overlapping region is located at the origin of the machine coordinate system. Both lasers can reach this region, requiring an access control mechanism to avoid conflicts. The boundary accuracy of the spatial overlapping region is consistent with the calculation accuracy of the independent reachable space boundary, providing an accurate geometric basis for subsequent constraint model construction.

[0044] The spatial coordinate range of the spatially overlapping region includes the minimum and maximum values ​​of the region along the three coordinate axes. The spatial coordinate range of the spatially overlapping region is from -100 mm to +100 mm on the X-axis, from -350 mm to +50 mm on the Y-axis, and from 100 mm to 300 mm on the Z-axis. The volume parameter is 16,000,000 cubic millimeters. The coordinate range is represented in interval form for easy boundary checks and collision detection. The volume parameter reflects the spatial range in which the two lasers interfere; the larger the volume, the higher the complexity of interference risk management. The accuracy requirements for the spatial coordinate range and volume parameter are consistent with the accuracy of the geometric calculations.

[0045] A mapping relationship needs to be established between spatial coordinate ranges and spatial geometric descriptions, establishing a correspondence between the numerical descriptions and geometric shape descriptions of overlapping spatial regions. Spatial coordinate ranges provide numerical boundary information for the regions, while spatial geometric descriptions provide shape and location information. This mapping ensures consistency and completeness between the two description methods. The data structure of the mapping relationship includes the correspondence between the six boundary values ​​of the spatial coordinate ranges and the twelve plane parameters of the independently reachable spatial boundaries. Establishing this mapping provides an efficient data access interface for the construction and querying of constraint models, supporting rapid spatial location determination and boundary checking operations.

[0046] The overlapping spatial regions and the independent reachable spatial boundaries together constitute the three-dimensional constraint model of the processing space. This model comprises three main components: the first independent reachable space boundary of the laser, the second independent reachable space boundary of the laser, and the overlapping spatial region between the two. The model's data structure adopts a hierarchical organization: the top layer provides an overall description of the entire processing space; the middle layer contains the geometric descriptions of each independent region and the overlapping region; and the bottom layer contains the specific coordinate parameters and constraints. The three-dimensional constraint model provides a complete geometric constraint foundation for laser path planning, conflict detection, and workspace management. The model's update and query operations support real-time requirements, with response times controlled within milliseconds.

[0047] In one optional implementation, based on the spatial distribution characteristics of the sub-task units in the multidimensional hierarchical decomposition tree and their positional relationships in the three-dimensional constraint model, an initial cutting path for the dual-head ultraviolet laser is generated. Simultaneously, priority evaluation of each sub-task unit is performed according to the workpiece processing characteristics, including: Extract the center point coordinates and contour envelope size of each sub-task unit from the multi-dimensional hierarchical decomposition tree; perform spatial position matching between the center point coordinates and the spatial overlapping area in the three-dimensional constraint model to determine whether each sub-task unit is located within the spatial overlapping area. Based on the spatial location matching results, for sub-task units located within the spatial overlap region, the shortest distance between them and the boundary of the spatial overlap region is calculated as an interference risk measurement parameter; for sub-task units not located within the spatial overlap region, the spatial distance from their center point coordinates to the current position of the dual-head ultraviolet laser is calculated. The spatial distribution feature vector of each subtask unit is constructed by combining the interference risk measurement parameter with the spatial distance value and the contour envelope size. The priority of each subtask unit is evaluated based on the spatial distribution feature vector. At the same time, the subtask units with the interference risk measurement parameter less than the preset interference threshold are assigned priority values. Based on the priority values, each subtask unit is assigned to the working sequence corresponding to the dual-head ultraviolet laser, and the center point coordinates of each subtask unit are connected sequentially according to the priority values ​​of the subtask units in the working sequence to generate the initial cutting path of the dual-head ultraviolet laser.

[0048] The center point coordinates of each sub-task unit in the multi-dimensional hierarchical decomposition tree are read from the node data structure of the decomposition tree. Each sub-task unit node stores the geometric center coordinates of its corresponding cutting contour or hole position. The center point coordinates include values ​​in the X, Y, and Z axes, with a coordinate accuracy maintained at the 0.01 mm level. The contour envelope size describes the spatial occupancy of the sub-task unit, including the maximum size values ​​along the X, Y, and Z axes. For sub-task units of the cutting contour type, the contour envelope size is the size of the bounding rectangle of the contour; for sub-task units of the hole position type, the contour envelope size is the effective influence range after adding the hole diameter to the laser spot size. The extraction process traverses all leaf nodes of the decomposition tree to collect complete spatial information of the sub-task units, providing a data foundation for subsequent spatial analysis.

[0049] Spatial matching of the center point coordinates with the overlapping spatial regions in the 3D constraint model requires determining whether the coordinates fall within the 3D boundary of the overlapping region. The boundary of the overlapping region is enclosed by six planes, corresponding to the minimum and maximum values ​​of the X, Y, and Z axes, respectively. The matching process compares the center point coordinates of the subtask unit with the boundary coordinates of the overlapping region axis by axis. When the center point coordinates are within the boundary range in all three axes, the subtask unit is determined to be within the overlapping region. The determination accuracy is consistent with the coordinate extraction accuracy to avoid misjudgments in critical boundary cases. The boundary check uses an inclusion detection algorithm with constant time computation, supporting rapid determination of a large number of subtask units.

[0050] Sub-task units located within the spatial overlap region need to have their shortest distance to the boundary of the overlap region calculated as an interference risk metric parameter. The shortest distance calculation considers the distances from the center point of the sub-task unit to the six boundary planes of the spatial overlap region, and takes the minimum value as the shortest distance. The distance calculation uses the Euclidean distance formula from a point to a plane, maintaining a calculation accuracy of 0.001 mm. A smaller value for the interference risk metric parameter indicates that the sub-task unit is closer to the boundary of the overlap region, and the higher the risk of dual-laser collision. The valid range of the distance value is from 0 to the maximum inscribed sphere radius of the overlap region; sub-task units outside this range are not located within the overlap region.

[0051] Sub-task units not located within the spatial overlap area need to have their center point coordinates calculated to determine the spatial distance from the current position of the dual-head ultraviolet laser. The current position of the dual-head ultraviolet laser includes the real-time coordinates of the first and second lasers, obtained from the laser controller's position feedback system. The spatial distance is calculated using the three-dimensional Euclidean distance formula, calculating the distance from the sub-task unit's center point to the current positions of the two lasers separately, and taking the smaller value as the spatial distance for that sub-task unit. The distance calculation accuracy must reach 0.01 mm to ensure accurate path planning. The spatial distance reflects the distance the lasers need to travel to reach the sub-task unit, affecting processing efficiency and path optimization.

[0052] Interference risk measurement parameters and spatial distance values ​​are combined with contour envelope dimensions to construct spatial distribution feature vectors for each subtask unit. These feature vectors are three-dimensional vectors. The first component is the normalized interference risk measurement parameter or spatial distance value; the second component is the maximum value of the contour envelope dimension in the horizontal plane; and the third component is the vertical value of the contour envelope dimension. Normalization unifies parameters with different dimensions to a value range of 0 to 1, with the normalization baseline value determined based on the maximum size of the workspace. The construction of these feature vectors provides standardized input data for subsequent priority evaluation, and the weights of each component can be adjusted according to specific process requirements.

[0053] Spatial distribution feature vectors support priority assessment for each subtask unit. The assessment process comprehensively considers three factors: interference risk, distance, and processing difficulty. Priority assessment employs a weighted summation method, linearly combining the three components of the feature vector according to preset weights to obtain a priority assessment score. In the weight settings, the interference risk component has a maximum weight of 0.5, the distance component has a weight of 0.3, and the size component has a weight of 0.2. The priority assessment score ranges from 0 to 1, with higher scores indicating higher priority and requiring priority processing. The assessment result assigns a priority value to each subtask unit for subsequent task scheduling and path generation.

[0054] Subtask units whose interference risk metric parameters are less than a preset interference threshold need to be assigned a special priority value. The preset interference threshold is determined based on the laser's safe operating distance, with a default value of 20 mm, which can be adjusted according to specific equipment characteristics and process requirements. When a subtask unit's interference risk metric parameter is less than the preset interference threshold, the subtask unit is marked as a high-risk task, and its priority value is forcibly set to the highest value of 1.0 to ensure priority processing and avoid potential conflicts. The accuracy of threshold determination is consistent with the accuracy of distance calculation to avoid errors in handling critical situations. The identification of high-priority tasks provides crucial information for the coordinated control of the dual-head laser.

[0055] Priority values ​​assign each subtask unit to the corresponding working sequence of the dual-headed ultraviolet laser. The allocation process is determined based on the positional relationship of the subtask units in the 3D constrained model. Subtask units located within the independent reachable space of the first laser are assigned to the working sequence of the first laser, and those located within the independent reachable space of the second laser are assigned to the working sequence of the second laser. Subtask units located in the overlapping spatial regions are dynamically allocated based on their priority values ​​and the current load balance. The allocation algorithm employs a greedy strategy, prioritizing the allocation of high-priority tasks to lasers with lighter current loads. The working sequence data structure uses a priority queue, supporting dynamic task insertion and priority adjustment.

[0056] In a work sequence, subtasks are arranged in numerical priority, with higher priority subtasks at the beginning and lower priority subtasks at the end. The sorting algorithm is quicksort, with a time complexity of O(nlogn), where n is the number of subtasks. During sorting, subtasks with the same priority are further sorted by spatial distance, with smaller distance values ​​appearing first. The sorting result ensures that the task execution order within each work sequence meets both priority requirements and efficiency optimization goals.

[0057] The initial cutting path of the dual-head ultraviolet laser is generated by sequentially connecting the center point coordinates of each sub-task unit. The path generation process iterates through the sorted work sequence, extracts the center point coordinates of each sub-task unit in sequence, and connects adjacent coordinate points with straight line segments to form a continuous path. The path connection follows the shortest distance principle; the trajectory of the laser moving from its current position to the next sub-task unit is a straight line. The initial cutting path consists of two parts: a movement path and a processing path. The movement path is the laser's idle movement trajectory between sub-task units, and the processing path is the specific cutting trajectory within each sub-task unit. The path data format includes parameters such as coordinate sequence, movement speed, and laser power.

[0058] In one optional implementation, the dynamic interference trend between the dual-headed ultraviolet lasers is predicted using the multidimensional hierarchical decomposition tree. An obstacle avoidance rule set is established based on the priority evaluation results of each sub-task unit. Real-time fine-tuning of the cutting path of the dual-headed ultraviolet lasers and dynamic reconstruction of the working area are performed according to the obstacle avoidance rule set, including: The dynamic interference trend between the dual-headed ultraviolet lasers is predicted by the multidimensional hierarchical decomposition tree. Based on the dynamic interference trend, a relative motion feature sequence of the dual-headed ultraviolet lasers is constructed. The relative motion feature sequence is quantified into an interference risk space region in the processing space coordinate system. Based on the relative motion feature sequence, the minimum safe distance between the dual-headed ultraviolet lasers is deduced. The minimum safe distance is used to derive the judgment criterion for the interference risk space region. The sequence of sub-task units associated with the spatial region of interference risk is analyzed, and the spatial distribution characteristics of the sub-task unit sequence are integrated with the minimum safe distance to generate a set of potential conflict points prediction. The hierarchical weight allocation scheme of the sub-task unit sequence is constructed by combining the priority evaluation results of each sub-task unit. An obstacle avoidance rule set including spatial avoidance strategy and temporal avoidance strategy is arranged according to the hierarchical weight allocation scheme. The cutting path of the subordinate sub-task unit is optimized in real time according to the spatial avoidance strategy. When the optimized cutting path still has the risk of interference, the execution sequence of the subordinate sub-task unit is reorganized according to the temporal avoidance strategy. If the interference risk cannot be eliminated by both the spatial avoidance strategy and the temporal avoidance strategy, the working area of ​​the dual-head ultraviolet laser is dynamically reconstructed based on the obstacle avoidance rule set.

[0059] Predicting the dynamic interference trend between dual-headed ultraviolet lasers using a multi-dimensional hierarchical decomposition tree requires analyzing the execution timing and spatial relationships of adjacent sub-task units within the decomposition tree. The prediction process traverses all nodes of the decomposition tree, extracting the estimated start time, execution duration, and spatial location information of each sub-task unit to construct a spatiotemporal execution matrix. The prediction algorithm checks whether the two lasers will simultaneously execute tasks within the spatially overlapping region within the same time window; when spatiotemporal overlap is detected, it is marked as a potential interference point. The accuracy of the interference trend determination is based on the time estimation accuracy and position accuracy of the sub-task units, requiring a time accuracy of 0.1 seconds and a position accuracy of 1 millimeter. The prediction result includes the time point of interference occurrence, the involved sub-task units, and the interference intensity level.

[0060] A dynamic interferometric trend analysis constructs a relative motion feature sequence for the dual-headed ultraviolet lasers, recording the relative positional changes of the two lasers during the predicted execution process. The relative motion feature sequence includes timestamps, the position coordinates of the first laser, the position coordinates of the second laser, and the distance between them. The sampling frequency is set to 10 times per second to ensure the capture of critical state changes during motion. The data structure of the feature sequence adopts a time-series format, with each sampling point containing the differences in X, Y, and Z axis coordinates and the composite distance value. The sequence length is determined based on the estimated execution time of the entire processing task, typically containing hundreds to thousands of sampling points. The relative motion feature sequence provides detailed motion trajectory data for subsequent interferometric risk analysis.

[0061] The relative motion feature sequence is quantized into an interference risk spatial region in the processing space coordinate system. During quantization, points on the motion trajectory whose distance is less than a safety threshold and their surrounding areas are marked as high-risk regions. The division of the interference risk spatial region uses a three-dimensional grid method with a grid resolution of 5 mm cubes. Each grid cell is assigned a risk level based on the number of risk trajectory points passing through it. There are five risk levels: extremely low, low, medium, high, and extremely high, with risk values ​​ranging from 0.1 to 1.0. The quantization results form a three-dimensional risk distribution map, providing spatial constraint information for path planning and conflict avoidance. The grid update frequency is synchronized with the sampling frequency of the relative motion feature sequence to ensure real-time risk assessment.

[0062] The minimum safe distance between dual-headed ultraviolet lasers is derived from relative motion characteristic sequences. The derivation process analyzes historical motion data and current mission characteristics to determine the minimum distance requirement to ensure safe operation. The calculation of the minimum safe distance considers factors such as the physical size of the lasers, motion speed, braking distance, and positioning accuracy. The basic safe distance is set at twice the maximum size of the laser's external dimensions, and the speed compensation distance is dynamically adjusted according to the current motion speed and braking capability. The derivation algorithm uses a sliding window analysis method with a window length of 5 seconds to analyze the motion trend and acceleration changes within the window. The numerical range of the minimum safe distance is 50 mm to 200 mm, and the specific value is determined based on the real-time motion status and mission complexity.

[0063] The minimum safe distance is used to determine the interference risk zone, defining the risk level and corresponding response strategies for different distance ranges. When the distance between two lasers is less than the minimum safe distance, the corresponding area is marked as a prohibited area, and the system forcibly stops related movements. When the distance is between the minimum safe distance and 1.5 times the minimum safe distance, it is marked as a warning area, requiring deceleration or coordinated movement. When the distance is greater than 1.5 times the minimum safe distance, it is marked as a safe area, allowing normal speed operation. The parameters of the determination criteria can be dynamically adjusted and optimized according to process requirements and equipment characteristics. The response time for criterion execution is required to be less than 100 milliseconds to ensure real-time safety protection.

[0064] The sequence of sub-task units associated with the spatial region of interference risk needs to be identified and extracted. Association analysis is based on the overlap between the spatial location of the sub-task units and the risk region. The analysis process checks the intersection of the contour envelope boundary of each sub-task unit with the spatial region of interference risk. A sub-task unit is considered associated with a risk region if the intersection volume is greater than 10% of the total volume of the sub-task units. The associated sub-task unit sequence is sorted according to risk level and execution sequence to form a priority task list. Each sub-task unit in the sequence records its risk level, the volume proportion of the associated risk region, and the estimated execution time. The analysis results provide a set of target tasks for obstacle avoidance strategy formulation.

[0065] The spatial distribution characteristics of the sub-task unit sequence are fused with the minimum safe distance to generate a predicted set of potential conflict points. The fusion process comprehensively considers task location, execution sequence, and safe distance constraints. The potential conflict point prediction employs a spatiotemporal collision detection algorithm, which expands the execution path of the sub-task unit into a three-dimensional pipeline with a radius equal to the minimum safe distance, detecting the intersection of pipelines corresponding to different lasers. The predicted set includes information such as the time and location of the conflict, the tasks involved, and the severity of the conflict. The severity of the conflict is calculated based on the overlap volume and duration. Prediction accuracy depends on time estimation and location accuracy; typical prediction errors are controlled within 0.5 seconds in time and 5 millimeters in location.

[0066] The priority evaluation results of each sub-task unit construct a hierarchical weight allocation scheme for the sub-task unit sequence. This scheme determines the priority order of different tasks in conflict resolution. The hierarchical weight allocation adopts a multi-level priority mechanism, primarily considering factors such as the technological importance, completion difficulty, and time constraints of the tasks. High-priority tasks are weighted from 0.8 to 1.0, medium-priority tasks from 0.4 to 0.7, and low-priority tasks from 0.1 to 0.3. The weight allocation scheme supports dynamic adjustment; weight values ​​can be recalculated when task execution changes. The allocation results form a weight matrix, providing a decision-making basis for subsequent conflict resolution strategies.

[0067] The hierarchical weight allocation scheme includes a set of obstacle avoidance rules encompassing spatial and temporal avoidance strategies. These rules define specific responses to different conflict scenarios. Spatial avoidance strategies include path replanning, detour route selection, and safety distance adjustment; strategy selection is based on the geometric characteristics of the conflict area and available space analysis. Temporal avoidance strategies include task delay execution, execution order adjustment, and parallelism control; strategy selection is based on task dependencies and time constraints. The obstacle avoidance rule set is organized using a decision tree structure, with the root node representing the conflict detection result, intermediate nodes representing judgment conditions, and leaf nodes representing specific avoidance measures. The execution priority of the rule set follows the hierarchical weight allocation scheme, with higher-weighted tasks receiving priority resource and path allocation.

[0068] A spatial avoidance strategy guides real-time optimization of the cutting paths of subordinate sub-task units. The optimization process dynamically adjusts the task execution path based on the current conflict situation. The path optimization algorithm employs a local search method to find alternative paths that avoid conflict areas while satisfying process constraints. Optimization parameters include path offset distance, detour angle, and speed adjustment range. The offset distance ranges from 5 mm to 50 mm, and the detour angle ranges from 15 degrees to 45 degrees. The convergence condition for the optimization process is finding a conflict-free path or reaching the maximum number of searches, set to 100. The real-time optimization response time is required to be less than 1 second to ensure timely handling of dynamic conflicts.

[0069] If interference risks still exist in the optimized cutting path, a timing avoidance strategy reorganizes the execution sequence of subordinate subtask units. The sequence reorganization process adjusts the start time and execution order of tasks to avoid simultaneous time and space conflicts. The reorganization algorithm employs task scheduling optimization methods, considering factors such as task dependencies, resource constraints, and deadlines. The timing precision of the sequence adjustment is 0.1 seconds, and the adjustment range is determined based on the conflict duration and task buffer time. The reorganization results need to be verified to ensure that the adjusted execution sequence meets process requirements and quality standards; adjustment schemes that do not meet the requirements will be rejected, and other solutions will be sought.

[0070] When neither spatial nor temporal avoidance strategies can eliminate interference risks, the obstacle avoidance rule set drives the dynamic reconstruction of the working area of ​​the dual-headed ultraviolet laser. The working area reconstruction process involves re-dividing the responsibility areas of the two lasers and adjusting the boundaries and access permissions of the spatially overlapping areas. Based on the current task distribution and conflict characteristics, the reconstruction algorithm calculates the new region boundary positions and dimensions, ensuring that the reconstructed working area eliminates major conflict sources. The adjustment range of the region reconstruction is limited by mechanical constraints and process requirements, with the boundary adjustment range being ±30% of the original boundary position. The reconstruction process requires updating the 3D constraint model and related data structures to ensure that subsequent operations are based on the new region division.

[0071] In one optional implementation, the reconstructed working area is divided into zones according to processing difficulty. Zone coupling parameters are derived from the operating status of the dual-head ultraviolet laser. Cutting parameters and motion speed are configured based on these zone coupling parameters to guide the dual-head ultraviolet laser to complete the collaborative cutting of each zone under quality constraints. Obtain the geometric complexity features and material property features of each sub-task unit within the reconstructed work area. Analyze the geometric complexity features and material property features through a feature fusion network to derive the processing difficulty assessment value of each sub-task unit. Based on the processing difficulty assessment value, construct the partitioning planning result of the difficulty level of the reconstructed work area. Based on the partitioning planning results, the operating load state and temporal distribution state of the dual-head ultraviolet laser are deduced. According to the feature fusion network, the operating load state and the temporal distribution state are respectively matched with the processing difficulty assessment value of each partition to generate partition coupling relationship parameters. The cutting parameters and motion speed of each partition are quantified based on the partition coupling relationship parameters. At the same time, the combination configuration scheme of laser power and pulse frequency is derived based on the results of the depth feature matching. The combination configuration scheme is then transformed into a collaborative execution command of the dual-head ultraviolet laser, guiding the dual-head ultraviolet laser to complete the collaborative cutting of each partition under quality constraints.

[0072] The geometric complexity features of each sub-task unit within the reconstructed work area are extracted from the geometric shape description of the sub-task unit, including parameters such as the rate of change of contour curvature, the number of corners, the variation in line segment length, and geometric symmetry. The extraction process analyzes the boundary contour of the sub-task unit, calculates the curvature value of each point on the contour, and obtains the rate of change of curvature by the difference between the curvature values ​​of adjacent points; a larger rate of change of curvature indicates higher geometric complexity. The number of corners is counted by determining the number of turning points on the contour with an angle change exceeding 30 degrees; the number of corners is positively correlated with processing difficulty. The variation in line segment length is obtained by calculating the standard deviation of the contour segment length; a larger standard deviation indicates a more irregular geometric shape. Geometric symmetry is quantified by calculating the symmetry of the contour about the center point; lower symmetry indicates a higher processing difficulty.

[0073] Material properties for each sub-task unit are obtained from the workpiece material database, including physical parameters such as material thickness, melting point temperature, thermal conductivity, reflectivity, and hardness. The data precision of these material properties must reach two decimal places to ensure the accuracy of subsequent calculations. Material thickness affects the penetration depth and power requirements of laser cutting, ranging from 0.05 mm to 5 mm; greater thickness increases processing difficulty. Melting point temperature determines the minimum laser power requirement, ranging from 200°C to 3000°C; higher melting points require greater laser power. Thermal conductivity affects heat diffusion and cutting quality; materials with high thermal conductivity are prone to generating a heat-affected zone. Reflectivity affects the effective utilization of laser energy; materials with high reflectivity require higher laser power. Hardness affects cutting speed and tool wear; materials with high hardness have relatively slower cutting speeds.

[0074] This feature fusion network analyzes geometric complexity features and material property features. The network structure consists of four main parts: an input layer, a feature extraction layer, a feature fusion layer, and an output layer. The input layer receives four parameters from the geometric complexity features and five parameters from the material property features, for a total of nine input features. The feature extraction layer normalizes and transforms the input features. Normalization maps each feature value to the range of 0 to 1, while the feature transformation uses a non-linear activation function to enhance feature expressiveness. The feature fusion layer weights the extracted geometric and material features. The fusion weights are learned through training data, with a weight of 0.6 for geometric features and 0.4 for material features. The output layer generates a processing difficulty assessment value, ranging from 0.1 to 1.0, with higher values ​​indicating greater processing difficulty.

[0075] The derivation process of the processing difficulty assessment value for each sub-task unit comprehensively considers the combined effects of geometric complexity and material properties. The assessment value is calculated using a weighted summation method, multiplying the fused feature values ​​by preset weight coefficients and then summing them. The weight coefficients are determined based on historical processing data and expert experience, with a weight coefficient of 0.7 for geometric complexity and 0.3 for material properties. The accuracy of the assessment value is maintained to three decimal places to ensure fine-grained differentiation of difficulty. The assessment process includes a boundary check mechanism; when the input feature value exceeds the normal range, anomaly handling is triggered, and in anomaly cases, the assessment value is set to the default value of 0.5. The assessment results provide a quantitative basis for subsequent zoning planning.

[0076] The difficulty assessment values ​​are used to construct a difficulty level zoning plan for the reconstructed work area, dividing it into three levels: high difficulty, medium difficulty, and low difficulty. The zoning threshold is set as follows: sub-task units with an assessment value greater than 0.7 are assigned to the high difficulty zone, those between 0.4 and 0.7 to the medium difficulty zone, and those less than 0.4 to the low difficulty zone. The zoning plan is optimized using a spatial clustering algorithm, merging spatially close sub-task units with the same difficulty level into contiguous zones to avoid overly fragmented zoning. The distance threshold for the clustering algorithm is set to 50 mm; adjacent sub-task units with a distance less than the threshold and the same difficulty level are merged. The zoning plan results form a zoning map, recording the boundary coordinates of each zone, the number of sub-task units it contains, and the average difficulty level.

[0077] The zoning planning results are used to extrapolate the operational load status of the dual-head ultraviolet laser. This extrapolation is based on the number of sub-task units in each zone, processing time estimation, and resource requirement analysis. The operational load status includes parameters such as CPU utilization, memory usage, laser power load, and motion axis load. CPU utilization is estimated based on the computational complexity of the path planning and control algorithms; complex tasks can achieve up to 80% CPU utilization, while simple tasks can achieve around 30%. Memory usage is calculated based on the storage requirements of path data and status information, with typical utilization ranging from 40% to 70% of available memory. Laser power load is determined based on material properties and cutting parameters, ranging from 20% to 90% of rated power. Motion axis load is calculated based on path complexity and motion speed; high-speed, complex motion can result in loads reaching 85% of the rated load.

[0078] The temporal distribution state of the dual-headed ultraviolet laser describes the execution time arrangement and sequence of tasks in each partition. The derivation of the temporal distribution state considers factors such as task dependencies, resource constraints, and parallelism limitations. The temporal distribution is represented in Gantt chart form, with the horizontal axis representing time and the vertical axis representing task identifiers. The chart displays the start time, duration, and end time of each task. The task scheduling algorithm employs the critical path method, prioritizing high-difficulty tasks on the critical path to ensure the shortest overall execution time. The temporal distribution state includes a time buffer setting, reserving 5% to 10% buffer time after each task is completed to handle uncertainties during execution. The time accuracy of the distribution state is 0.1 seconds, meeting the requirements of real-time control.

[0079] The feature fusion network performs deep feature matching between the runtime load status and temporal distribution status and the processing difficulty assessment value of each partition. The deep feature matching process establishes the correlation between load status parameters, temporal parameters, and difficulty assessment values, and the matching algorithm employs correlation analysis. Matching the runtime load status with the processing difficulty assessment value is achieved by calculating the correlation coefficient between each load parameter and the difficulty value; parameter pairs with a correlation coefficient greater than 0.6 are considered to have a strong correlation. Matching the temporal distribution status with the processing difficulty assessment value analyzes the correspondence between task execution time and difficulty level; higher-difficulty tasks typically require longer execution times and more preparation time. The matching results are quantified into a feature weight matrix, with matrix element values ​​ranging from 0 to 1, representing the correlation strength between different features.

[0080] Deep feature matching generates partition coupling parameters, which describe the mutual influence and coordination needs between different partitions. These parameters consist of three main components: resource contention coefficient, temporal dependency coefficient, and spatial constraint coefficient. The resource contention coefficient quantifies the degree of competition between different partitions for resources such as laser power and motion axes; partitions with intense competition require more refined resource allocation strategies. The temporal dependency coefficient describes the execution order constraints between partitions; some partitions can only begin execution after other partitions have completed. The spatial constraint coefficient reflects the degree of spatial interference between partitions; partitions with overlapping spatial locations need to coordinate their execution timing to avoid conflicts. The numerical range of the coupling parameters is from 0 to 1, with larger values ​​indicating stronger coupling.

[0081] The partition coupling relationship parameters quantify the cutting parameters and movement speed of each partition. The quantification process adjusts the execution parameters of each partition based on the coupling strength. Cutting parameters include key parameters such as laser power, pulse frequency, cutting speed, and focal position. The laser power adjustment range is 30% to 95% of the rated power; higher laser power is required for more complex partitions. The pulse frequency adjustment range is 1 kHz to 100 kHz, with the frequency selection determined based on material properties and cutting quality requirements. The cutting speed adjustment range is 1 mm / s to 50 mm / s; speed settings need to balance processing efficiency and cutting quality. The focal position adjustment accuracy is 0.01 mm to ensure effective focusing of laser energy. Movement speed includes rapid traverse speed and positioning speed; the rapid traverse speed can reach up to 500 mm / s, while the positioning speed is controlled below 100 mm / s to ensure positioning accuracy.

[0082] The deep feature matching results are used to derive a combination configuration scheme for laser power and pulse frequency. This derivation process is based on a comprehensive analysis of material properties, geometric complexity, and quality requirements. The combination configuration scheme is organized using a parameter mapping table, which contains the optimal parameter combinations corresponding to different material types, thickness ranges, and complexity levels. The combination configuration of laser power and pulse frequency considers the synergistic effect between the two: high power and low frequency are suitable for cutting thick materials, while low power and high frequency are suitable for precision machining. Ten typical combinations are included, covering common processing scenarios and material types. Scheme selection uses a decision tree algorithm to automatically select the best-matching configuration combination based on the current task characteristics. The configuration accuracy requirements are: laser power error less than 5%, and pulse frequency error less than 2%.

[0083] The combined configuration scheme is interpreted as coordinated execution commands from the dual-head ultraviolet lasers. This process transforms parameter configurations into executable control commands. The coordinated execution commands include device identification, parameter settings, execution timing, and synchronization signals. The command format uses a structured data format, comprising a command header, parameter segment, and checksum segment. The command header contains metadata such as command type, priority, and timestamp; the parameter segment contains specific device parameter settings; and the checksum segment contains a data integrity checksum. Command transmission employs a real-time communication protocol, with transmission latency controlled within 10 milliseconds. Synchronization accuracy for command execution is required to reach the 1-millisecond level to ensure the precise coordinated action of the dual-head lasers.

[0084] To achieve coordinated cutting of different zones using dual-headed ultraviolet lasers under quality constraints, a quality monitoring and feedback mechanism is necessary. Quality constraints include indicators such as cutting accuracy, surface roughness, heat-affected zone width, and edge perpendicularity. Cutting accuracy must be controlled within ±0.02 mm, surface roughness less than 1.6 micrometers, heat-affected zone width within 0.1 mm, and edge perpendicularity error less than 2 degrees. Quality monitoring employs real-time detection methods to monitor key quality parameters online during the cutting process. The feedback mechanism adjusts cutting parameters based on quality monitoring results, automatically optimizing parameter settings when quality indicators deviate from target values. Coordinated control of the collaborative cutting ensures the synchronization and consistency of the two lasers' actions, avoiding quality problems caused by incoordination.

[0085] In one optional implementation, the operating load state and temporal distribution state of the dual-head ultraviolet laser are deduced based on the partitioning planning results. Then, according to the feature fusion network, the operating load state and the temporal distribution state are respectively matched with the processing difficulty assessment value of each partition using deep feature matching to generate partition coupling relationship parameters, including: Based on the zoning planning results, the operating load state sequence and the time-series distribution state sequence of the dual-head ultraviolet laser are generated. The operating load state sequence and the time-series distribution state sequence are respectively constructed as operating state feature vectors within a dynamic time window. The spatiotemporal dimension of the operating state feature vectors is decoupled by a feature fusion network. The processing difficulty assessment value of each partition is extracted from the decoupled running state feature vector, and the mapping relationship matrix between the processing difficulty assessment value and the running state feature vector is established using the feature fusion network. The mapping relationship matrix is ​​iteratively optimized based on the deep feature matching algorithm to obtain the degree of mutual influence between each partition. The calculation results of the deep feature matching algorithm are marked as the significantly influential associated partitions, and partition coupling relationship parameters are generated.

[0086] like Figure 2 As shown, the method includes: The generation of the operational load state sequence for the dual-head ultraviolet laser from the zoning planning results requires extracting time-series data from the task allocation and equipment operating parameters of each zone. The operational load state sequence includes time-series data for five key parameters: CPU utilization, memory usage, laser power load, motion axis load, and temperature status. The data acquisition frequency is set to 100 times per second to ensure the capture of transient changes during operation. CPU utilization is obtained by monitoring the processor usage of the path planning and control algorithms, with a value range of 0 to 100% and an accuracy maintained at 1%. Memory usage monitors the memory usage of path data, state cache, and temporary variables, with a value range of 0 to 95%, triggering a memory cleanup mechanism when it exceeds 95%. Laser power load reflects the ratio of the laser's actual output power to its rated power, ranging from 10% to 90%, with an accuracy requirement of 0.1%. Motion axis load includes the torque load and speed status of the X, Y, and Z axes, with a load range of 0 to 85% of the rated load.

[0087] The time-series distributed state sequence records the dynamic changes in the execution time arrangement and resource allocation of tasks in each partition. The time-series distributed state sequence includes time-series parameters such as task start time, estimated duration, actual execution time, resource occupation time, and waiting time. The task start time is represented by an absolute timestamp with millisecond precision to ensure accurate time synchronization. The estimated duration is estimated based on the partition difficulty assessment value and historical execution data, with the estimation error controlled within 10%. The actual execution time is obtained by real-time monitoring of task status changes and includes three components: preparation time, processing time, and completion time. The resource occupation time statistics show the actual usage time of critical resources such as lasers and motion axes in each partition. The waiting time records the delay time caused by resource contention or dependencies; if the waiting time is too long, a scheduling optimization mechanism is triggered.

[0088] The runtime load state sequence and the time-series distribution state sequence are used to construct a runtime state feature vector within a dynamic time window. The time window length is set to 10 seconds, and the sliding step is 1 second. The construction process of the runtime state feature vector involves feature extraction and dimensionality reduction of the raw data within the time window. The extracted features include statistical characteristics such as mean, variance, maximum, minimum, and trend. The dynamic time window is implemented using a circular buffer data structure with a buffer size of 1000 data points, supporting real-time processing of high-frequency data. The feature vector is 20-dimensional, containing four statistical features for each of the five load parameters and five time-series parameters. Vector normalization uses a minimum-maximum normalization method, mapping all feature values ​​to the range of 0 to 1, avoiding mutual interference between features of different dimensions. The feature vector update frequency is synchronized with the data acquisition frequency to ensure the real-time performance and accuracy of the features.

[0089] The feature fusion network decouples the runtime feature vectors in terms of spatiotemporal dimensions. The network structure comprises two main components: a temporal decoupling module and a spatial decoupling module. The temporal decoupling module uses a recurrent neural network structure to extract time-series patterns and trend changes from the feature vectors. The network has three layers, each containing 64 neurons. The spatial decoupling module uses a convolutional neural network structure to analyze the spatial correlation and influence patterns across different intervals. The convolutional kernel size is set to 3×3, the stride is 1, and zero-padding is used. The activation function for the decoupling process is a modified linear unit function, and the optimization algorithm is an adaptive moment estimation algorithm with a learning rate of 0.001. The network is trained using supervised learning, with training data derived from historical processing task records. The ratio of training set to validation set is 8:2. Model convergence is determined based on the trend of the loss function on the validation set; convergence is considered achieved when the change in the loss function is less than 0.001 over 10 consecutive rounds.

[0090] The decoupled operational status feature vectors are used to extract the processing difficulty assessment value for each partition. The extraction process utilizes the load and timing features in the feature vectors to calculate the comprehensive difficulty index for each partition. The calculation of the processing difficulty assessment value considers factors such as peak CPU utilization, memory usage duration, laser power variation, average motion axis load, and execution time deviation. The weights for each factor are: CPU utilization 0.2, memory usage 0.15, laser power 0.3, motion axis load 0.25, and execution time 0.1. The assessment value is calculated using a weighted average method, with a value range of 0.1 to 1.0 and an accuracy maintained at 0.001. The assessment value is dynamically updated every 5 seconds, recalculated based on the latest feature vector data, ensuring that the assessment result reflects the current actual difficulty status.

[0091] A feature fusion network establishes a mapping matrix between processing difficulty assessment values ​​and operational status feature vectors. The matrix construction process analyzes the correlation and dependency between the assessment values ​​and each dimension of the feature vectors. The mapping matrix has a size of 20×1, corresponding to the mapping from 20-dimensional feature vectors to 1-dimensional assessment values. Matrix element values ​​represent the contribution of the corresponding feature dimension to the assessment value, with values ​​ranging from -1 to 1; positive values ​​indicate positive correlation, and negative values ​​indicate negative correlation. The matrix is ​​initialized using a random initialization method, with initial values ​​ranging from -0.1 to 0.1. The mapping relationship is learned using the gradient descent algorithm, with the mean squared error loss function and L2 regularization to prevent overfitting. The learning rate for matrix updates is set to 0.01, the batch size is 32 samples, and the training epochs are 1000.

[0092] The deep feature matching algorithm iteratively optimizes the mapping matrix, aiming to minimize the error between the predicted and actual evaluation values. An adaptive learning rate adjustment strategy is employed during the iterative optimization process; the learning rate is halved if the loss function does not significantly decrease within five consecutive iterations. A momentum mechanism is introduced, with a momentum coefficient set to 0.9, to accelerate convergence and avoid local optima. The iteration process terminates when the maximum number of iterations (2000) is reached or the loss function converges, with a convergence threshold of 0.0001. A model snapshot is saved every 100 iterations for model rollback and performance comparison. The degree of mutual influence between partitions is calculated by analyzing the interaction terms of corresponding features in different partitions within the optimized mapping matrix. The influence value ranges from 0 to 1, with higher values ​​indicating more significant influence.

[0093] The results of the deep feature matching algorithm are labeled as significantly influential associated partitions. The labeling process is based on a threshold judgment of the degree of influence and a statistical significance test. The threshold for judging the significance of influence is set to 0.6; when the degree of influence between partitions is greater than 0.6, they are labeled as significantly associated. The statistical significance test uses the t-test method with a significance level set to 0.05. Only associations that pass the significance test are confirmed as valid associations. The labeling results of associated partitions form an association matrix. The matrix is ​​a symmetric matrix, and the matrix elements are 0 or 1, where 1 indicates that there is a significant association between the corresponding partitions, and 0 indicates that there is no significant association. The identification results of associated partitions are used to guide subsequent collaborative optimization and resource allocation. Partitions with high correlation require tighter coordination and control.

[0094] The generation of partition coupling parameters comprehensively considers the degree of influence, significance of association, and physical constraints between partitions. Coupling parameters include three main attributes: coupling strength, coupling type, and coupling direction. The coupling strength value is equal to the influence degree value of the corresponding partition, ranging from 0.6 to 1.0. Coupling type is categorized into three types based on the nature of the influence: competitive, cooperative, and dependent. Competitive indicates resource competition between partitions, cooperative indicates that partitions need to work together, and dependent indicates that partitions have execution order dependencies. The coupling direction describes the directionality of the influence, divided into unidirectional and bidirectional coupling. Unidirectional coupling indicates that one partition influences another, while bidirectional coupling indicates that partitions influence each other. The parameter data structure uses a graph structure, where nodes represent partitions, edges represent coupling relationships, and edge weights represent coupling strength.

[0095] The method further includes: The dual-head ultraviolet laser equipment is upgraded by adding laser heads to the existing single-head laser equipment to create a dual-head structure. The addition of the laser heads adopts a modular installation method. The new laser heads are parallel to the existing laser heads within the equipment's working plane, with a spacing of 150 mm between them. This spacing was optimized and calculated based on the common processing material size of 800 mm × 600 mm and the density of typical processing patterns (3 feature points per square centimeter). The laser heads are physically mounted using an 8 mm thick aluminum alloy bracket, connected to the equipment body with M6 bolts. The bolt torque is set at 12 Nm to ensure the stability of the laser heads during high-speed movement. Each laser head weighs 2.3 kg, and the bracket is designed to bear a single-point load of 10 kg with a safety factor of 4.3. The vertical height adjustment range of the laser heads is ±5 mm, with an adjustment accuracy of 0.1 mm, achieved through fine-tuning nuts for precise positioning.

[0096] The custom-designed control board enables independent control of the dual laser heads. It employs a 32-bit ARM processor architecture with a 168 MHz clock speed, 512 kilobytes of RAM, and 2 megabytes of flash memory. The board integrates dual pulse-width modulation (PWM) signal generators, each supporting a frequency range of 10 Hz to 100 kHz, a duty cycle adjustment accuracy of 0.1%, and an output voltage range of 0 to 5 volts. Laser emission frequency control is achieved through a digital signal processor, with a frequency setting range of 1 kHz to 50 kHz and a frequency stability better than 0.01%. Power control utilizes analog voltage regulation, with a linear relationship between the control voltage and laser power, achieving a control accuracy of 1% of the rated power. Scanning path control is based on G-code instruction parsing, supporting three trajectory types: linear interpolation, circular interpolation, and spline curve interpolation, with a position resolution of 1 micrometer and a speed control range of 1 mm / min to 3000 mm / min.

[0097] The compatibility between the control board and the existing control system of the equipment is achieved through a standardized interface protocol, using the CAN bus communication protocol with a baud rate of 1 megabit per second, a standard frame format, and an 11-bit identifier length. The communication data packet structure includes six fields: frame header, device address, function code, data length, data content, and checksum. The maximum data packet length is 64 bytes. In the device address allocation, the existing controller address is 0x01, the new control board address is 0x02, and the broadcast address is 0xFF. The function code definition covers four types of operations: parameter setting, status query, instruction execution, and error reporting, with an encoding range of 0x10 to 0x4F. The communication timeout is set to 500 milliseconds. After a timeout, a retransmission mechanism is initiated, with a maximum of 3 retransmissions. A fault protection mode is triggered if a retransmission fails.

[0098] The selection and configuration of optical path components are based on market-standard components. The reflector is a 25mm diameter dielectric film reflector with a reflectivity greater than 99.5% and a damage threshold of 5 joules per square centimeter. The focusing lens is a 100mm focal length plano-convex lens with a transmittance greater than 98% and a focal diameter less than 50 micrometers. The optical path components are installed using a precision optical adjustment frame with an adjustment accuracy of 1 arcsecond in the angular direction and 1 micrometer in the displacement direction. The laser beam transmission path is calibrated using an optical power meter and a beam analyzer, with optical power loss controlled within 5% and beam circularity error less than 5%. Optical alignment uses a He-Ne laser as the indicator light source, with a wavelength of 632.8 nm, a power of 1 milliwatt, and a coaxiality error with the ultraviolet laser less than 50 micrometers.

[0099] The development of specialized software controls the coordinated operation of the two laser heads. The software architecture employs a multi-threaded design, with the main control thread responsible for task allocation and status monitoring, and laser head control threads managing the operating parameters of each laser head. The task allocation algorithm is based on a work area partitioning strategy, dividing the material to be processed into multiple sub-regions according to the number of laser heads. The size of each sub-region is calculated and determined based on the total material size and the laser head spacing. Boundary processing for the region partitioning uses an overlapping scanning method, with an overlap width set at 0.5 mm to ensure processing quality at the joints. The processing sequence is arranged using a path optimization algorithm to minimize the laser head travel distance and idle time. The algorithm convergence criteria are a path length change of less than 1% and more than 100 iterations.

[0100] Parallel operations are implemented using a time-slice round-robin scheduling strategy. The time slice length for each laser head is set to 10 milliseconds, and the scheduling cycle is 20 milliseconds, ensuring synchronized execution of the two laser heads. A double-buffering mechanism is used for the data buffer, with each buffer having a capacity of 1024 instructions. When the current buffer finishes execution, it automatically switches to the backup buffer, with a switching time of less than 1 millisecond. Instruction preprocessing includes coordinate transformation, velocity planning, and acceleration / deceleration control. Coordinate transformation is calculated based on the origin offset of each laser head, and velocity planning uses a trapezoidal acceleration / deceleration curve, with a maximum acceleration set at 500 mm / s².

[0101] Real-time monitoring is achieved through sensor data acquisition. Laser power monitoring uses a photodiode power sensor with a response time of less than 1 microsecond, a power detection range of 0.1 watt to 20 watts, and a detection accuracy of 2% of the reading. Temperature monitoring uses a thermocouple temperature sensor with a measurement range of -40°C to 150°C, a measurement accuracy of ±0.5°C, and a response time of 500 milliseconds. The sensor data acquisition frequency is set to 1 kHz, and data filtering uses a moving average algorithm with a sliding window length of 10 data points. Monitoring data is acquired through an analog-to-digital converter with a 16-bit resolution and an input voltage range of 0 to 5 volts.

[0102] The feedback protection mechanism is triggered based on a preset threshold comparison. Laser power anomalies are determined when the actual power deviates from the set power by more than 10% for a duration greater than 100 milliseconds. Temperature anomalies are determined when the laser head temperature exceeds 70 degrees Celsius or the control board temperature exceeds 60 degrees Celsius. Anomaly detection uses a logic where three consecutive sampled values ​​exceed the threshold to avoid false alarms caused by occasional interference. Protection measures include three levels: laser power reduction, equipment shutdown, and alarm notification. Power reduction adjusts the laser power to 50% of the rated power; shutdown immediately stops all laser output and mechanical movement; and alarm notifications are provided through an audible and visual alarm and text prompts on the display screen.

[0103] A second aspect of the present invention provides a dual-head ultraviolet laser cutting control system, comprising: The acquisition module is used to acquire the geometric feature information of the workpiece to be processed, construct a three-dimensional constraint model of the processing space based on the geometric feature information, and construct a multi-dimensional hierarchical decomposition tree of the cutting task of the workpiece to be processed based on the three-dimensional constraint model. The evaluation module is used to generate the initial cutting path of the dual-head ultraviolet laser based on the spatial distribution characteristics of the sub-task units in the multi-dimensional hierarchical decomposition tree and their positional relationship in the three-dimensional constraint model, and to evaluate the priority of each sub-task unit according to the workpiece processing characteristics. The reconstruction module is used to predict the dynamic interference trend between the dual-head ultraviolet lasers through the multi-dimensional hierarchical decomposition tree during the execution of the initial cutting path, establish an obstacle avoidance rule set in combination with the priority evaluation results of each sub-task unit, and perform real-time fine-tuning of the cutting path of the dual-head ultraviolet laser and dynamic reconstruction of the working area according to the obstacle avoidance rule set. The cutting module is used to divide the reconstructed working area into zones according to the processing difficulty. The operating status of the dual-head ultraviolet laser is used to derive the zone coupling relationship parameters. The cutting parameters and movement speed are configured according to the zone coupling relationship parameters to guide the dual-head ultraviolet laser to complete the collaborative cutting of each zone under quality constraints.

[0104] A third aspect of the present invention provides an electronic device, comprising: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0105] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0106] This invention can be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of the invention.

[0107] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; 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 or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A dual-head ultraviolet laser cutting control method, characterized in that, include: Obtain the geometric feature information of the workpiece to be processed, construct a three-dimensional constraint model of the processing space based on the geometric feature information, and construct a multi-dimensional hierarchical decomposition tree of the cutting task of the workpiece to be processed based on the three-dimensional constraint model. Based on the spatial distribution characteristics of the sub-task units in the multi-dimensional hierarchical decomposition tree and their positional relationship in the three-dimensional constraint model, the initial cutting path of the dual-head ultraviolet laser is generated, and the priority of each sub-task unit is evaluated according to the workpiece processing characteristics. During the execution of the initial cutting path, the dynamic interference trend between the dual-headed ultraviolet lasers is predicted by the multi-dimensional hierarchical decomposition tree. An obstacle avoidance rule set is established in combination with the priority evaluation results of each sub-task unit. The cutting path of the dual-headed ultraviolet lasers is fine-tuned in real time and the working area is dynamically reconstructed according to the obstacle avoidance rule set. The reconstructed working area is divided into zones according to processing difficulty. The zone coupling relationship parameters are derived from the operating status of the dual-head ultraviolet laser. Cutting parameters and movement speed are configured according to the zone coupling relationship parameters to guide the dual-head ultraviolet laser to complete the collaborative cutting of each zone under quality constraints.

2. The method according to claim 1, characterized in that, Based on the geometric feature information, a three-dimensional constraint model of the processing space is constructed, and based on the three-dimensional constraint model, a multi-dimensional hierarchical decomposition tree for the cutting task of the workpiece to be processed is constructed, including: Extract the set of cutting contour segments and the set of hole position coordinates from the geometric feature information, map the set of cutting contour segments and the set of hole position coordinates to the motion coordinate system of the dual-head ultraviolet laser, and determine the independent reachable spatial boundary of each laser based on the motion range of the dual-head ultraviolet laser; Based on the independent reachable space boundary, the spatial overlap region of the dual-head ultraviolet laser during its movement is calculated, and the spatial overlap region and the independent reachable space boundary together constitute a three-dimensional constraint model of the processing space. In the three-dimensional constraint model, the spatial overlapping area is identified as the interference risk constraint area, and the independent reachable spatial boundary is identified as the laser partition constraint boundary. Based on the interference risk constraint area and the laser partition constraint boundary, the set of cutting contour lines is hierarchically decomposed according to the spatial position belonging relationship to construct a multi-dimensional hierarchical decomposition tree of the cutting task of the workpiece to be processed.

3. The method according to claim 2, characterized in that, The spatial overlap region of the dual-head ultraviolet laser during its movement is calculated based on the independent reachable space boundary, and the three-dimensional constraint model of the processing space is formed by the spatial overlap region and the independent reachable space boundary together, including: Obtain the motion stroke parameters and installation position parameters of each laser in the dual-head ultraviolet laser. Based on the motion stroke parameters and installation position parameters, calculate the maximum spatial range that the motion endpoint of each laser can reach in the processing space, and define the maximum spatial range as the independent reachable spatial boundary of each laser. Extract the spatial geometric description information of the independent reachable space boundary, and identify the intersection of the two independent reachable space boundaries of the dual-head ultraviolet laser during its movement through spatial geometric operations based on the spatial geometric description information, and determine the intersection as the spatial overlapping region; Obtain the spatial coordinate range and volume parameters of the spatially overlapping region, associate and map the spatial coordinate range with the spatial geometric description information, and based on the association mapping, construct a three-dimensional constraint model of the processing space together with the spatially overlapping region and the independent reachable spatial boundary.

4. The method according to claim 1, characterized in that, Based on the spatial distribution characteristics of the sub-task units in the multidimensional hierarchical decomposition tree and their positional relationships in the three-dimensional constraint model, the initial cutting path of the dual-head ultraviolet laser is generated. Simultaneously, priority evaluation of each sub-task unit is performed according to the workpiece processing characteristics, including: Extract the center point coordinates and contour envelope size of each sub-task unit from the multi-dimensional hierarchical decomposition tree; perform spatial position matching between the center point coordinates and the spatial overlapping area in the three-dimensional constraint model to determine whether each sub-task unit is located within the spatial overlapping area. Based on the spatial location matching results, for sub-task units located within the spatial overlap region, the shortest distance between them and the boundary of the spatial overlap region is calculated as an interference risk measurement parameter; for sub-task units not located within the spatial overlap region, the spatial distance from their center point coordinates to the current position of the dual-head ultraviolet laser is calculated. The spatial distribution feature vector of each subtask unit is constructed by combining the interference risk measurement parameter with the spatial distance value and the contour envelope size. The priority of each subtask unit is evaluated based on the spatial distribution feature vector. At the same time, the subtask units with the interference risk measurement parameter less than the preset interference threshold are assigned priority values. Based on the priority values, each subtask unit is assigned to the working sequence corresponding to the dual-head ultraviolet laser, and the center point coordinates of each subtask unit are connected sequentially according to the priority values ​​of the subtask units in the working sequence to generate the initial cutting path of the dual-head ultraviolet laser.

5. The method according to claim 1, characterized in that, The dynamic interference trend between the dual-headed ultraviolet lasers is predicted by the multi-dimensional hierarchical decomposition tree. An obstacle avoidance rule set is established based on the priority evaluation results of each sub-task unit. Real-time fine-tuning of the cutting path of the dual-headed ultraviolet lasers and dynamic reconstruction of the working area are performed according to the obstacle avoidance rule set, including: The dynamic interference trend between the dual-headed ultraviolet lasers is predicted by the multidimensional hierarchical decomposition tree. Based on the dynamic interference trend, a relative motion feature sequence of the dual-headed ultraviolet lasers is constructed. The relative motion feature sequence is quantified into an interference risk space region in the processing space coordinate system. Based on the relative motion feature sequence, the minimum safe distance between the dual-headed ultraviolet lasers is deduced. The minimum safe distance is used to derive the judgment criterion for the interference risk space region. The sequence of sub-task units associated with the spatial region of interference risk is analyzed, and the spatial distribution characteristics of the sub-task unit sequence are integrated with the minimum safe distance to generate a set of potential conflict points prediction. The hierarchical weight allocation scheme of the sub-task unit sequence is constructed by combining the priority evaluation results of each sub-task unit. An obstacle avoidance rule set including spatial avoidance strategy and temporal avoidance strategy is arranged according to the hierarchical weight allocation scheme. The cutting path of the subordinate sub-task unit is optimized in real time according to the spatial avoidance strategy. When the optimized cutting path still has the risk of interference, the execution sequence of the subordinate sub-task unit is reorganized according to the temporal avoidance strategy. If the interference risk cannot be eliminated by both the spatial avoidance strategy and the temporal avoidance strategy, the working area of ​​the dual-head ultraviolet laser is dynamically reconstructed based on the obstacle avoidance rule set.

6. The method according to claim 1, characterized in that, The reconstructed work area is divided into zones according to processing difficulty. Zone coupling parameters are derived from the operating status of the dual-head ultraviolet laser. Cutting parameters and motion speed are configured based on these parameters to guide the dual-head ultraviolet laser in completing the collaborative cutting of each zone under quality constraints. Obtain the geometric complexity features and material property features of each sub-task unit within the reconstructed work area. Analyze the geometric complexity features and material property features through a feature fusion network to derive the processing difficulty assessment value of each sub-task unit. Based on the processing difficulty assessment value, construct the partitioning planning result of the difficulty level of the reconstructed work area. Based on the partitioning planning results, the operating load state and temporal distribution state of the dual-head ultraviolet laser are deduced. According to the feature fusion network, the operating load state and the temporal distribution state are respectively matched with the processing difficulty assessment value of each partition to generate partition coupling relationship parameters. The cutting parameters and motion speed of each partition are quantified based on the partition coupling relationship parameters. At the same time, the combination configuration scheme of laser power and pulse frequency is derived based on the results of the depth feature matching. The combination configuration scheme is then transformed into a collaborative execution command of the dual-head ultraviolet laser, guiding the dual-head ultraviolet laser to complete the collaborative cutting of each partition under quality constraints.

7. The method according to claim 6, characterized in that, Based on the partitioning planning results, the operating load state and temporal distribution state of the dual-head ultraviolet laser are deduced. Using the feature fusion network, the operating load state and temporal distribution state are respectively matched with the processing difficulty assessment value of each partition using deep feature matching, generating partition coupling relationship parameters including: Based on the zoning planning results, the operating load state sequence and the time-series distribution state sequence of the dual-head ultraviolet laser are generated. The operating load state sequence and the time-series distribution state sequence are respectively constructed as operating state feature vectors within a dynamic time window. The spatiotemporal dimension of the operating state feature vectors is decoupled by a feature fusion network. The processing difficulty assessment value of each partition is extracted from the decoupled running state feature vector, and the mapping relationship matrix between the processing difficulty assessment value and the running state feature vector is established using the feature fusion network. The mapping relationship matrix is ​​iteratively optimized based on the deep feature matching algorithm to obtain the degree of mutual influence between each partition. The calculation results of the deep feature matching algorithm are marked as the significantly influential associated partitions, and partition coupling relationship parameters are generated.

8. A dual-head ultraviolet laser cutting control system for implementing the method described in any one of claims 1-7, characterized in that, include: The acquisition module is used to acquire the geometric feature information of the workpiece to be processed, construct a three-dimensional constraint model of the processing space based on the geometric feature information, and construct a multi-dimensional hierarchical decomposition tree of the cutting task of the workpiece to be processed based on the three-dimensional constraint model. The evaluation module is used to generate the initial cutting path of the dual-head ultraviolet laser based on the spatial distribution characteristics of the sub-task units in the multi-dimensional hierarchical decomposition tree and their positional relationship in the three-dimensional constraint model, and to evaluate the priority of each sub-task unit according to the workpiece processing characteristics. The reconstruction module is used to predict the dynamic interference trend between the dual-head ultraviolet lasers through the multi-dimensional hierarchical decomposition tree during the execution of the initial cutting path, establish an obstacle avoidance rule set in combination with the priority evaluation results of each sub-task unit, and perform real-time fine-tuning of the cutting path of the dual-head ultraviolet laser and dynamic reconstruction of the working area according to the obstacle avoidance rule set. The cutting module is used to divide the reconstructed working area into zones according to the processing difficulty. The operating status of the dual-head ultraviolet laser is used to derive the zone coupling relationship parameters. The cutting parameters and movement speed are configured according to the zone coupling relationship parameters to guide the dual-head ultraviolet laser to complete the collaborative cutting of each zone under quality constraints.

9. An electronic device, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by the processor, they implement the method described in any one of claims 1 to 7.

Images