A method for generating a three-dimensional printing path of a multi-branch tubular carbon fiber part and an additive manufacturing method

Abstract

This invention provides a method for generating a three-dimensional printing path and an additive manufacturing method for multi-branched tubular carbon fiber parts. The path generation method includes: (1) splitting the three-dimensional mesh model of the part into multiple branches and generating layer lines for each independent branch surface; (2) generating fiber routing lines for each branch based on the layer lines of each branch; (3) connecting the fiber routing lines of each branch at the branch joints to form a fiber routing line covering the entire surface of the part; (4) integrating the path based on surface spreading and phase adjustment to obtain a continuous, uninterrupted three-dimensional fiber routing line on the three-dimensional substrate surface that matches the range of motion of the main robotic arm. The printing path of this invention has high forming quality and high printing efficiency. The multi-branched tubular carbon fiber parts obtained based on this printing path have good quality and strong load-bearing capacity.

Description

Technical Field

[0001] This invention belongs to the field of 3D printing manufacturing technology, specifically relating to a method for generating 3D printing paths and an additive manufacturing method for multi-branched tubular carbon fiber parts. Background Technology

[0002] Continuous carbon fiber reinforced composites, with their excellent specific strength and specific stiffness, have become key materials for achieving lightweighting and high performance in high-end manufacturing fields such as aerospace, automotive, and robotics. Additive manufacturing (3D printing) technology, especially the additive manufacturing process for continuous fiber composites, provides unprecedented design freedom and molding possibilities for the manufacture of structural components with complex geometries, opening up new avenues for low-cost and rapid prototyping of complex composite structures.

[0003] Existing technologies (such as patents CN120206806B and CN120206807B) have proposed 3D printing path planning methods and dual-robotic arm collaborative printing systems for multi-branched substrates, providing a basic approach for automated winding. However, existing solutions still have significant shortcomings when dealing with practical engineering constraints: Uninterrupted path solutions under robotic arm motion limit constraints are complex: the rotation axes of industrial robotic arms (especially the A6 axis of the end flange) typically have physical angular limitations. Existing fiber arrangement rules for uninterrupted sequencing often involve comparing and calculating numerous flange phases, which is quite complex.

[0004] The lack of high-precision constant pressure contact control: In the process of complex curved surface winding printing, the contact pressure between the print head and the substrate surface is a key process parameter affecting fiber wetting, interlayer bonding strength, and the final mechanical properties of the part. Existing methods lack a real-time, closed-loop constant pressure control mechanism when the robotic arm's posture changes rapidly over a wide range. Simple open-loop control or path following without pressure feedback cannot guarantee maintaining a constant optimal clamping force throughout the printing process, which can easily lead to poor interlayer bonding and increased porosity, thereby impairing the consistency and reliability of the part's mechanical properties.

[0005] Furthermore, the challenge of demolding cannot be ignored: traditional processes often employ metal or insoluble rigid mandrels. For structures with complex internal cavities, especially those with branches or bends, the mandrel is difficult or even impossible to remove from closed or narrow channels after the composite material has cured. This severely limits the design and manufacturing freedom of such high-performance complex hollow structures.

[0006] Therefore, there is an urgent need to propose a better method for generating three-dimensional continuous fiber printing paths and additive manufacturing methods for multi-branched tubular carbon fiber parts, so as to achieve high-quality, high-efficiency and repeatable manufacturing of such complex structures. Summary of the Invention

[0007] The purpose of this invention is to provide a method for generating a three-dimensional continuous fiber printing path and an additive manufacturing method for multi-branched tubular carbon fiber parts with better printing effect and higher printing efficiency.

[0008] To achieve the above objectives, the technical solution adopted by the present invention is as follows: The first aspect of this invention provides a method for generating paths in three-dimensional continuous fiber printing of multi-branched tubular carbon fiber parts on a three-dimensional matrix, characterized by comprising the following steps: (1) The three-dimensional mesh model of the multi-branched tubular carbon fiber part is split into multi-branch surfaces, and layer lines are generated for each independent branch surface; (2) Generate fiber routing lines for each branch based on the layered lines of each branch; (3) Connect the fiber routing lines of each branch at the branch joint to form a fiber routing line covering the entire surface of the part; (4) Path integration based on surface unfolding and phase adjustment, which includes: Surface spreading mapping steps: Construct a mapping relationship between the roll surface (Surf0) and the two-dimensional spreading surface (Surf1), wherein the U-parameter direction of the roll surface corresponds to the rotation angle of the end flange of the main robotic arm that controls the attitude of the three-dimensional substrate, and the outer surface contour of the three-dimensional substrate is adapted to the inner cavity contour of the multi-branched tubular carbon fiber part; the V-parameter direction of the roll surface corresponds to the geodesic distance field from the near edge to the far edge on the three-dimensional mesh model, the near edge is the opening edge on the three-dimensional mesh model that docks with the end flange of the main robotic arm, and the far edge is the opening edge that is farthest from the end flange of the main robotic arm; the mapping relationship maps the U-parameter direction of the roll surface to the width direction of the two-dimensional spreading plane, and maps the V-parameter direction of the roll surface to the length direction of the two-dimensional spreading plane; Generation of inter-fiber connection path network: Select at least one cutting plane, and map the intersection line of the cutting plane and the 3D mesh model and all edge lines of the 3D mesh model from the roll surface to the spreading surface according to the mapping relationship to obtain the inter-fiber connection path network. The cutting plane is determined through the following steps: Obtain the geometric center points of each opening edge of the three-dimensional mesh model; A set of candidate planes is generated based on the geometric center point; Close all openings in the three-dimensional mesh model to form a closed mesh model; From the candidate plane set, select a plane that intersects with the closed mesh model and forms a line with a value of 1. Spreading surface sorting and connection steps: Map the fiber routing line covering the entire part surface in step (3) onto the two-dimensional spreading surface to obtain a set of planar curves; based on the coordinate values ​​of the starting point of each planar curve in the width direction of the two-dimensional spreading surface, alternately interweave and reorder; according to the sorting order, using the connecting path network between the spreading fibers as the connecting carrier, connect the beginning and end endpoints of adjacent planar curves to form a single, uninterrupted planar routing line on the two-dimensional spreading surface; Inverse mapping step: The planar wiring is inversely mapped back to the scroll surface to obtain a continuous three-dimensional fiber wiring on the three-dimensional substrate surface.

[0009] This invention, when planning the winding path, first breaks down the complex multi-branch model into independent, simple branches. Next, high-quality local fiber paths are generated on each independent branch. Then, an intelligent algorithm smoothly connects these branch paths into a unified fiber routing network covering the entire model. Finally, through surface unfolding and phase adjustment processes, this fiber routing network is transformed into a final printing path that satisfies hardware motion constraints and is absolutely continuous and uninterrupted. This invention transforms the path planning problem in three-dimensional space into a two-dimensional planar process, solving the bottleneck of robotic arm rotation limitations through mapping-adjustment-inverse mapping, thus ensuring the physical executability of the path.

[0010] In some specific embodiments, the multi-branched tubular carbon fiber component includes a main pipe extending along a first axis, the main pipe having a first opening and a second opening disposed opposite to each other along the first axis; at least one branch pipe extending outward from the main pipe and communicating with the main pipe, the branch pipe having a third opening away from the main pipe; wherein the at least one branch pipe is disposed at the same axial position as the main pipe; the first opening, the second opening and the third opening are all open ends, together forming at least three openings.

[0011] More preferably, the edge of the opening at one end of the main pipe corresponds to the proximal edge, and the edge of the opening at the other end of the main pipe corresponds to the distal edge.

[0012] More preferably, the branch pipe extends along a second axis intersecting the first axis; even more preferably, the branch pipe extends along a second axis perpendicular to the first axis. When there are multiple branch pipes, the multiple second axes may overlap (collinear) or not overlap (distributed at an angle).

[0013] More specifically, the multi-branched tubular carbon fiber parts are cylindrical bodies with the same or similar cross-sectional shapes, including but not limited to circular, square, and elliptical shapes. The multi-branched tubular carbon fiber parts form multi-branched structures with the same joint through the cylindrical bodies, including but not limited to T-shaped, T-shaped, and cross-shaped structures.

[0014] In some specific embodiments, in the step (4) of determining the cutting plane, the step of generating a set of candidate planes includes: A reference plane is determined based on any three of the geometric center points; For each of the reference planes, three associated planes are generated, wherein each associated plane is perpendicular to the reference plane and intersects the line determined by any two of the three points; The candidate plane set is composed of the reference plane and all the associated planes.

[0015] The intersection lines of the cutting plane and the 3D mesh model, along with all the edge lines of the 3D mesh model, form an inter-fiber connection path network (Net0). This network is then mapped from the roll surface (Surf0) to the spreading surface (Surf1) according to the mapping relationship, resulting in a spread inter-fiber connection path network (Net1). This ensures both the uninterrupted spreading inter-fiber connection path network and the matching range of rotation of the end flange of the main robotic arm.

[0016] In some specific embodiments, the surface unfolding mapping step (4) specifically includes: Define the proximal edge and the distal edge: with the plane where the end flange of the main robotic arm is located as a reference, select the proximal edge and the distal edge on the three-dimensional mesh model; Generate a basic parametric surface: Calculate the geodesic distance field from the near edge to each point on the surface of the 3D mesh model. The geodesic distance field is 0 at the near edge and the maximum geodesic distance at the far edge. Based on the geodesic distance field, extract closed contour lines distributed along the model axis. Generate a NURBS surface as the basic surface based on the closed contour lines. The U parameter direction of the basic surface corresponds to the wrapping direction of the contour lines, and its parameter range is set to [0, 1] to linearly map the actual rotation phase interval of the main robotic arm from -180° to +180°. The V parameter direction corresponds to the axis from the near edge to the far edge, and its parameter range is set to [0, 1] to linearly map the normalized geodesic distance from the near edge to the far edge. Constructing a complete scroll surface: Periodically extend the range of U parameters of the base surface from [0, 1] to [-0.5, 1.5] to construct a complete scroll surface with a parameter domain of [-0.5, 1.5] × [0, 1], where the U parameters in the sub-interval of [0, 1] correspond to the actual rotation phase interval of the robotic arm; Generate a two-dimensional spreading plane and establish a mapping: Create a rectangular plane as the spreading surface, whose length direction is defined as mapping the maximum geodesic distance, and whose width direction is defined as mapping the virtual rotation range of the main robotic arm from -360° to +360° to provide the periodic margin required for phase adjustment; establish the mapping relationship from the scroll surface to the spreading plane.

[0017] In some specific embodiments, in the spreading surface sorting and connection step of step (4), the phase adjustment is: the phase adjustment is achieved by translating the planar curve along the width direction of the spreading surface by an integer multiple of its width, so as to adapt to the actual rotation working range of the main robotic arm, specifically including: The width direction of the spread surface is defined as mapping the virtual rotation range of the main robotic arm from -360° to +360° to provide a periodic margin, wherein the range [0.25W, 0.75W] corresponds to the actual rotation working range of the main robotic arm from -180° to +180°. For each planar curve, calculate its coordinate range [Y] along the width direction on the spread surface. min ,Y max ]; If Y min < 0.25W or Y max If the value is greater than 0.75W, then the curve is shifted along its width by a distance of k×0.5W, where k is an integer, such that the proportion of the length of the shifted curve within the interval [0.25W, 0.75W] is not less than a preset threshold. More specifically, the preset threshold is preferably 50%.

[0018] The physical limitations of the robotic arm's rotation axis (A6 axis) are resolved by adjusting the phase as described above. This allows path segments in three-dimensional space that may cause the robotic arm to rotate beyond the work area to be pre-adjusted before printing, which is equivalent to physically performing a rotation reset and circumventing hardware limitations.

[0019] In some specific embodiments, in the spreading surface sorting and connection step of step (4), the alternating interlacing reordering specifically includes: Arrange all planar curves in ascending order according to the coordinates of their starting points on the unfolded surface to obtain the initial sequence; Perform n rounds of alternating interleaving operations on the initial sequence, where n > 1. In the k-th round, the sequence obtained in the previous round will be used to perform the operation on the 2-th sequence.k-1 Each ordered sublist is split into two new sublists; After completing n rounds of splitting, we get 2. n A final sublist; By using a polling extraction method, the first element of the current element is extracted from each of the final sublists in turn until all sublists are empty, thus forming the reordered target sequence.

[0020] By alternating and rearranging the paths as described above, the paths initially arranged in the order of the starting point can be broken up again, achieving a gradual change in the density of the filaments at the start of printing, thus optimizing the process quality of the initial printing segment.

[0021] In some specific implementations, n = 3 in the reordering operation; in other implementations, this can be adjusted according to actual needs.

[0022] In some embodiments, in the spreading and connecting steps of step (4), the specific steps for connecting the beginning and end endpoints of adjacent planar curves using the interconnecting path network between the spreading fibers as the connecting carrier are as follows: Sequential connection: Process each pair of adjacent planar curves in the order determined by the sorting. Generating a transition segment: For a pair of adjacent planar curves after sorting, the end point of the first curve and the start point of the second curve are used as the start and end points of the transition segment. In the network of connecting paths between the spread fibers, the shortest path that can connect the start and end points is found, and the shortest path is fitted into a smooth transition curve. Connecting the first and last curves: Connect the first curve, the transition curve, and the second curve in sequence to form a longer continuous curve; Iterative integration: For all adjacent curve pairs after sorting, repeat the steps from generating transition segments to iterative integration until all planar curves are integrated into a single continuous and uninterrupted planar cabling line on the spread surface.

[0023] In some implementations, the multi-branch surface splitting in step (1) includes: Obtain the edge lines of the openings at the ends of each branch of the three-dimensional mesh model; A dynamic model is constructed, and collision constraints, adsorption forces that make it close to the surface of the three-dimensional mesh model, elastic constraints that maintain its length, bending constraints that maintain the angle of its adjacent line segments, and gravitational constraints that point to other edge lines are applied to each of the said end opening edge lines. Smooth branch cutting lines are generated through iterative solutions using physical simulations. The three-dimensional mesh model is divided along each of the branch cutting lines to obtain multiple independent branch tubular meshes.

[0024] In some implementations, generating layer lines for each individual branch surface in step (1) includes: For each branched tubular grid, calculate the geodesic distance field on its surface from the dividing line to the end edge line; Using a preset average layer height as the spacing, a series of closed contour lines are extracted from the geodesic distance field as initial layer lines; The initial layer lines are aligned by seam alignment: taking the first layer line as a reference, the seam points of the remaining layer lines are adjusted to the positions that are closest to the reference seam point along the curve arc length. The initial layer lines are then subjected to orientation unification: the wrapping direction of all layer lines is unified.

[0025] This invention utilizes dynamic physics simulation software (Grasshopper / Kangaroo2) to automatically and optimally generate smooth and reasonable branching lines by applying various physical constraints (collision, adhesion, length preservation, angle preservation, and gravity) to the model's end edge lines. This replaces traditional manual or simple geometric segmentation, achieving high-quality automated preprocessing. The generation of layered lines is based on a geodesic distance field, ensuring that the layered lines are distributed along the shortest path on the model surface, which is more accurate than simple axial projection. Unified seam alignment and consistent orientation operations establish a standardized and parametric geometric foundation for the subsequent generation of accurate spiral paths.

[0026] Furthermore, the specific steps for unifying the seam position and generation direction are as follows: Establish a baseline: Select a baseline layer line and record the location of its joint points and the direction of circumference; Aligning seams: Adjust the seam points of the remaining layer lines to the positions on their respective curves that are closest to the reference seam point. More specifically, for each layer line to be adjusted, calculate the arc length of all points on its closed curve from the curve to the reference seam point, and select the point with the shortest arc length as the new seam point. Unify direction: Adjust the wrapping direction of all layer lines to be consistent with the wrapping direction of the reference. More specifically, compare the tangent direction of each layer line at its joint point with the tangent direction of the reference, and reverse the layer lines with opposite directions.

[0027] Further, in step (2), generating the fiber routing for each branch includes: Calculate the first x The formula for calculating the offset of the seam point of the layered line is as follows: ,in, Representing the i The layered line and the first iThe average floor height between +1 floor lines i <x , This is the preset offset angle value; Move all the seam points of the layered lines along their own curve direction by the corresponding offset amount. After offsetting, the same number of equal division points are generated on each layer line, and the equal division points with the same index on adjacent layer lines are connected sequentially with a smooth curve to form the fiber busbar wiring for that branch. Compared with traditional methods, the above fiber busbar wiring generation method has higher quality and controllability, through parameter... The helix angle can be directly controlled, and in different implementations, specific values ​​can be set according to their respective needs.

[0028] The offset angle The value can be set according to actual process requirements. From a geometric kinematics perspective, this ensures that the fiber routing is continuous and uninterrupted within the rotation limit (±180°) of the main robotic arm. In a specific implementation, The upper limit of the value can approximately satisfy , This represents the maximum geometrically permissible angle of inclination of the fiber path direction relative to the V-parameter direction (axial direction) in the two-dimensional spreading mapping of the reel surface, when the fiber path is constrained within the actual rotational working range of the main robotic arm (corresponding to the spreading surface width range [0.25W, 0.75W]). Let U be the length of the shortest U-shaped isoparametric curve of this branch on the scroll surface (Surf0). This is the length of the shortest V-parametric curve for this branch on the scroll surface (Surf0). When When = 0°, the wiring is perpendicular to the layering line and is wrapped in a loop; when When ≠0°, the wiring is spiral-shaped, and The larger it is, the stronger its spirality.

[0029] In some specific implementations, the division points with the same index are specifically: division points are generated on each offset layer line, and the number of division points required for each layer line is... n Calculated by the following formula: , in, For the first i The length of the new layer line, This represents the total number of layer lines. d is the desired fiber spacing, and [] is the floor function.

[0030] In some specific embodiments, the connection smoothness metric is obtained by calculating the sum of the products of the squares of the curvature of the discrete sampling points of the projection curve and the corresponding arc length micro-element.

[0031] In some specific embodiments, the smooth geometric transition curve is a C1 continuous third-order B-spline curve.

[0032] In some specific embodiments, the step (3) of connecting the fiber routing lines of each branch at the branch joint includes: Filtering candidate connection pairs: For the target fiber busbar, among all the other branches of the fiber busbar, search for fiber busbars whose three-dimensional spatial distance between their starting point and the starting point of the target fiber busbar is less than a preset threshold, and form a candidate set; Evaluate connection quality: For each fiber busbar in the candidate set, construct a spatial bridging curve connecting the starting point of the target fiber busbar and the starting point of the candidate fiber busbar, and project the spatial bridging curve vertically onto the surface of the three-dimensional mesh model. Calculate the total bending energy of the projected curve as a measure of connection smoothness. Selection and Connection: Select the fiber busbar with the smallest smoothness metric value from the candidate set as the optimal connection pair, and construct a smooth geometric transition curve to connect the two fiber busbars in the optimal connection pair into a continuous curve.

[0033] This invention abandons the traditional approach of pre-setting connection relationships and adopts a process of active search, quantitative evaluation, and optimal connection selection. First, candidate connection pairs are screened based on a distance threshold. Then, a transition curve is constructed for each candidate pair, and its smoothness metric f (defined as the total bending energy projected onto the model surface) is calculated. Finally, the candidate pair with the smallest smoothness f value is selected for geometric connection. This automates and optimizes the connection process, ensuring a natural and smooth transition path at the joints, thereby forming high-quality fiber routing.

[0034] In some implementations, the path generation method further includes a fiber arrangement optimization step based on a specific field: The strain energy density distribution field of the multi-branched tubular part under the target stress condition is obtained as the density field of the fiber arrangement. The fiber routing obtained in step (3) is iteratively mapped and optimized based on the density field to make the fiber paths more dense in the region of high strain energy density and more sparse in the region of low strain energy density.

[0035] In some specific implementations, the iterative mapping optimization specifically includes: Construct mesh mapping pairs: Using the density field as the input color, obtain a variable density mesh whose vertex density is proportional to the density field value; then re-homogenize the variable density mesh to obtain a new uniform mesh, thus forming a mesh pair; Perform iterative mapping: Map the set of points on the fiber cabling line between the uniform grid and the variable density grid. During mapping, the path points are directly mapped to the same parameter coordinates on the variable density grid according to the parameter coordinates of the path points on the uniform grid, so that they gather in the high density area. The mapping process is iterated multiple times until a preset number of iterations is reached.

[0036] This invention uses the strain energy density distribution field obtained from finite element analysis as the basis for optimization, driving the fiber path to gather in the key stress area of ​​the structure. This results in the final part no longer being a uniform fiber layout, but rather the material being distributed as needed according to the stress condition, strengthening in high stress areas and reducing weight in low stress areas, thereby significantly improving the mechanical properties and material utilization efficiency of the part.

[0037] In some specific implementations, the iterative mapping optimization specifically includes: Constructing mesh mapping pairs: Input the uniform triangular mesh of the 3D mesh model into the RemeshByColour component of Grasshopper, using the density field as the input color, to obtain a variable density mesh whose vertex density is proportional to the density field value; then re-homogenize the variable density mesh to obtain a new uniform mesh, thus forming a mesh pair; Perform iterative mapping: map the point set on the fiber busbar between the uniform grid and the variable density grid. The mapping process is iterated multiple times until a preset number of iterations is reached.

[0038] In each mapping, path points are mapped to the same parameter coordinates on the variable-density grid (new grid) based on their parameter coordinates on the uniform grid (original grid), causing them to cluster in high-density regions. After multiple iterations of the mapping process, fiber paths naturally become denser in high-energy regions, ensuring that the graphic relationships do not become misaligned.

[0039] In some implementations, a path point and attitude calculation step is also included, which includes: sampling path points on the three-dimensional fiber cabling according to an adaptive density, and calculating the target attitude of the robotic arm that controls the attitude of the print head for each sampling point. Wherein, the adaptive density is given by the formula de = de 0 ( a | n · m | + 1) Confirm, de To adapt to density, de 0 is the base point density. n Let be the surface unit normal vector of the sampling point on the 3D mesh model. m The normal vector is a preset fixed reference unit. mPreferably, the direction of the centerline of the end flange of the main robotic arm is preferred. a For a preset adjustment coefficient, in some specific implementations, 0 < a < 100, further preferably 2 < a < 15.

[0040] The target posture of the robotic arm is defined by the normal of the print head target posture plane, and the normal of the print head target posture plane is given by the formula N = (1- β ) n - βt Calculate, where N is the normal to the target orientation plane of the print head. n Let be the surface unit normal vector of the sampling point on the 3D mesh model. t Let be the unit tangent vector of the fiber cabling at the sampling point. β The preset tilt parameter, and 0 < β < 1.

[0041] Furthermore, the path point and attitude calculation steps also include velocity planning: Based on the attitude change rate between adjacent sampling points, the motion speed of the print head relative to the three-dimensional substrate between each sampling point is dynamically planned, and the motion speed is calculated according to the following formula: , in: For the print head from the first q -1 sampling point moved to the first q The velocity of movement at each sampling point; The preset base motion speed; For the first q The angle between the surface unit normal vector of the sampling point and the fixed reference unit normal vector, minus the angle between the sampling point and the fixed reference unit normal vector. q The absolute value of the difference between the included angles of -1 sampling points: ; For the first q The sampling point and the first q The angle between the projected components of the surface unit normal vector at -1 sampling point on the plane perpendicular to the fixed reference unit normal vector: ; No. q The angle between the surface unit normal vector at each sampling point and the fixed reference unit normal vector: ; , , These are the preset weighting coefficients.

[0042] In some specific implementations, the weighting coefficients in the formula The value range is [0, 150]. The value range is [0, 150]. The values ​​of are in the range of [0, 5].

[0043] More preferably, the weighting coefficients in the formula The value range is [50, 150]. The value range is [50, 150]. The values ​​of are in the range of [1, 5].

[0044] In some specific implementations, the path point and attitude calculation steps and the velocity planning steps are implemented in the Grasshopper environment of the Rhino platform through visual programming and custom script components.

[0045] By solving path points and attitudes and planning speed, it is possible to effectively prevent printing jitter, incomplete placement or quality degradation in complex curved areas due to excessively fast robotic arm movement. It achieves intelligent speed reduction that adapts to the geometric complexity of the path, thereby ensuring high printing quality.

[0046] A second aspect of the present invention provides an additive manufacturing method for a multi-branched tubular carbon fiber part. Based on a three-dimensional model of the multi-branched tubular carbon fiber part, a continuous fiber printing path is generated using the path generation method described above. Based on the continuous fiber printing path, a printing device is controlled to perform continuous carbon fiber winding printing on a three-dimensional substrate surface that matches the shape of the multi-branched tubular carbon fiber part, so as to prepare the multi-branched tubular carbon fiber part.

[0047] In some embodiments, the printing device is a cooperative industrial robot system, including a master robotic arm that controls the orientation of the three-dimensional substrate and a slave robotic arm that carries the print head.

[0048] According to some specific embodiments, the additive manufacturing method is as follows: Three-dimensional substrate preparation and installation steps: Prepare a three-dimensional substrate of water-soluble material, and install the three-dimensional substrate on the end of the main robotic arm through a fixed frame, so that one branch end face of the three-dimensional substrate corresponds to and is fixedly connected to the working end face of the flange at the end of the main robotic arm; Path planning and control command generation steps: Based on the three-dimensional model of the multi-branch tubular carbon fiber part, a continuous fiber printing path is generated using the path generation method described above. Based on the continuous fiber printing path and the model of the three-dimensional matrix, a control command is generated to coordinate the motion of the main robotic arm and the slave robotic arm carrying the print head. Winding and printing step: Control the main robotic arm and the slave robotic arm to move in coordination according to the coordinated motion command, and perform continuous winding and printing of carbon fibers on the three-dimensional substrate surface to form a composite material winding body; Post-processing and core removal steps: The wound body is post-processed and the three-dimensional matrix is ​​dissolved and removed to obtain the multi-branched tubular carbon fiber part.

[0049] Furthermore, the generation of coordinated motion commands for controlling the master and slave robotic arms includes combining the results of the path point and attitude calculation steps and the velocity planning steps with the kinematic models of the two robotic arms to calculate their respective end-effector pose and velocity command sequences.

[0050] In some specific embodiments, the winding printing step further includes synchronization control and force feedback adjustment: The host computer communicates with the two robotic arm controllers via the Real-time Data Exchange (RTDE) protocol to achieve motion synchronization; The printing pressure is collected in real time by the force sensor at the end of the robotic arm; When the pressure is greater than the first threshold, the tool coordinate system of the robotic arm is controlled to shift by a preset step along its positive Z-axis. When the pressure is less than the second threshold, the tool coordinate system is controlled to shift the preset step size along the negative Z-axis. After the offset, the entire tool coordinate system of the robotic arm is updated, and the pose of all subsequent path points is recalculated based on this. After generating the updated motion command, printing is resumed to maintain the printing pressure within the preset range.

[0051] In some specific implementations, the first threshold is 5.0N, the second threshold is 2.0N, and the preset step size is 0.001m.

[0052] In some specific implementations, the Z-axis offset of the tool coordinate system is constrained by an upper limit of +0.01m and a lower limit of -0.01m.

[0053] In some specific implementations, after the printing program starts, a preset time is delayed to allow the system to stabilize before activating the synchronization control and force feedback adjustment. Further, the preset time is 10 seconds.

[0054] Through the above closed-loop constant pressure control, the compaction pressure between the fiber and the matrix (or the printed layer) remains constant throughout the printing process, thereby significantly improving the interlayer bonding strength and the density and consistency of the printed parts, and realizing the molding of high-performance composite materials.

[0055] In some embodiments, during the three-dimensional substrate preparation and installation steps, the water-soluble material is polyvinyl alcohol, and the three-dimensional substrate and its edge extensions at both ends are integrally printed by fused deposition modeling and assembled into a rigid frame by connecting rods and flange fasteners. Further, the overall three-dimensional model formed by the three-dimensional substrate and the edge extensions is used as the path to generate the required three-dimensional model.

[0056] In some preferred embodiments, the three-dimensional substrate and its edge extensions at both ends are integrally printed by fused deposition modeling.

[0057] In some embodiments, the post-processing and core removal steps include removing the edge extension and the additive structure attached thereto, and the dissolution and removal of the three-dimensional matrix specifically involves immersing the entangled body in deionized water at 60°C to 90°C until the water-soluble material is completely dissolved.

[0058] In some implementations, the master robotic arm and the slave robotic arm are UR series collaborative robots, and the collaborative motion commands are synchronously sent to the controllers of the two robotic arms via a real-time data exchange protocol.

[0059] Due to the application of the above technical solution, the present invention has the following advantages compared with the prior art: This invention provides a method for generating three-dimensional continuous fiber printing paths for multi-branched tubular carbon fiber parts. Through physical simulation-based automatic multi-branch splitting, seam offset spiral generation, and intelligent branch connection based on quantified smoothness evaluation, it achieves automatic generation of high-quality fiber routing lines for arbitrary multi-branched tubular structures. In particular, the surface spreading and phase adjustment algorithms map the three-dimensional path to a two-dimensional plane for processing. By performing simple translation operations on the plane, it pre-solves the problem of physical rotation angle limitations of the rotation axis (A6 axis) when driving a robotic arm (such as a UR robot) to perform printing, thus ensuring that each generated path is "executable." This fundamentally avoids the common problem of path planning being disconnected from hardware execution, achieving fully automatic, high-performance, and physically executable continuous fiber laying planning for complex multi-branched structures. Furthermore, the path generation method of this invention also integrates topology optimization based on mechanical properties and adaptive motion speed planning, making the final path not only continuous and executable but also optimal in terms of mechanical properties and printing process quality.

[0060] The additive manufacturing method for multi-branched tubular carbon fiber parts provided by this invention is based on the three-dimensional continuous fiber printing path provided by this invention, resulting in excellent product quality. Furthermore, a precise inner core 3D printed from a water-soluble material (such as PVA) is used as the three-dimensional printing substrate for the multi-branched tubular carbon fiber parts. Through the high-precision coordinated movement of two robotic arms, the continuous fiber path obtained by the path generation method of this invention is executed. Combined with real-time, closed-loop force feedback constant pressure control, the interlayer bonding strength and the density of the composite material are significantly improved, ensuring a high degree of consistency and reliability in the mechanical properties of the parts. This is of great significance for the industrial manufacturing of high-performance complex structural components. Attached Figure Description

[0061] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0062] Figure 1 This is a schematic diagram of the dual-arm continuous carbon fiber winding system in Example 1; Figure 2 This is a schematic diagram of the three-dimensional substrate and fixing frame in Example 1; Figure 3 This is a schematic diagram of the overall process flow for the manufacturing method of the multi-branched tubular carbon fiber parts in Example 1. Figure 4 The output results of each processing stage of the fiber arrangement and connection scheme in the winding path planning of Example 1 are shown in the figure. Figure 5 This is an output diagram of the selective field-specific fiber arrangement optimization process in the winding path planning of Example 1. in, Figure 2 and Figure 3 In the middle, 3D substrate 00; substrate fixing frame 01; edge extension 011; connecting rod 012; flange fastener 013; continuous carbon fiber printhead 02; continuous carbon fiber feeding system 021; main robotic arm 03; main robotic arm control cabinet 031; slave robotic arm 04; slave robotic arm control cabinet 041; personal computer 05; 3D printer 06; PVA filament material spool 061. Detailed Implementation

[0063] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0064] It should be noted that the terms “comprising” and “having”, and any variations thereof, are intended to cover non-exclusive inclusion, for example, a process, method, apparatus, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to such process, method, product or device.

[0065] It should be noted that the term "multi-branched tubular carbon fiber part" specifically refers to a hollow component integrally formed from continuous carbon fiber composite material through additive manufacturing, characterized by a bifurcated topology. This part is not a single pipe, but rather includes a main pipe extending along a first axis and one or more branch pipes with the same joint, the main pipe and branch pipes together forming at least three openings. This application uses a typical three-branch structure as an example. The multi-branched tubular carbon fiber part to be manufactured is a T-shaped structure, including a main pipe extending along the first axis, the main pipe having a first opening and a second opening arranged opposite each other along the first axis; a branch pipe extending outward from the sidewall of the main pipe and communicating with the main pipe, the branch pipe having a third opening away from the main pipe, the branch pipe being located at the axial position in the middle of the main pipe, the first opening, the second opening, and the third opening being open ends, together forming three openings. The end of the main pipe closer to the end flange of the main robotic arm is defined as the proximal end (corresponding to the first opening), and the end farther from the end flange of the main robotic arm is defined as the distal end (corresponding to the second opening). The main pipe and branch pipe are both circular in shape in cross-sections perpendicular to their respective axes. It is understood that the present invention is not limited to the T-shaped structure described above. Based on the concept of the present invention, the part can also be a cross-shaped (with two opposing branch pipes) or a spherical (branch pipes near the end of the main pipe) or other joint-like structures. The cross-sections of the main pipe and branch pipes perpendicular to their respective axes can be regular shapes such as circles, squares, or ellipses.

[0066] Example 1: This example provides a manufacturing system and method for multi-branched tubular carbon fiber parts.

[0067] like Figure 1 and Figure 2As shown in the figure, a manufacturing system for multi-branched tubular carbon fiber parts includes a water-soluble three-dimensional matrix 00, a matrix fixing frame 01, a continuous carbon fiber printing component, and a matrix 3D printing component. Among them, the matrix fixing frame 01 includes an edge extension 011 extending outward from the end of the three-dimensional matrix 00, a connecting rod 012 threadedly connected to the edge extension 011, and a flange fixing member 013 threadedly connected to the connecting rod 012. The continuous carbon fiber printing component includes a main robotic arm for controlling the movement of the three-dimensional matrix 00, a sub-robotic arm with a continuous carbon fiber print head 02 mounted at the end and controlling the movement of the continuous carbon fiber print head 02, a continuous carbon fiber feeding system 021 supporting the continuous carbon fiber print head 02, a main robotic arm control cabinet 031 for driving the main robotic arm 03, a sub-robotic arm control cabinet 041 for driving the sub-robotic arm 04, and a personal computer 05 electrically connected to the main robotic arm control cabinet 031 and the sub-robotic arm control cabinet 041. The matrix 3D printing component includes a 3D printer 06 and its consumables (PVA filament cartridge 061). As Figure 3 shown, the specific steps of the manufacturing method for multi-branched tubular carbon fiber parts are as follows: S1: Design the matrix fixing frame According to the electronic model of the target multi-branched tubular carbon fiber part (such as a multi-branched hollow shell), reverse design a water-soluble three-dimensional matrix model geometrically matching it. Based on this water-soluble three-dimensional matrix model, design a rigid frame structure for fixing and controlling. The frame includes an edge extension, a flange fixing member, and a connecting rod. Among them, the edge extension is a boss structure axially extending outward from one or more ends of the water-soluble three-dimensional matrix (the extension length is 10 mm in this embodiment), which is used to provide a mechanical clamping and connection interface. The flange fixing member is provided with a standard interface for connecting to the flange at the end of the main robotic arm. The connecting rod is used to rigidly connect the edge extension and the flange fixing member.

[0068] In this embodiment, the edge extension and the three-dimensional matrix are integrally designed using the same material and process. The outer surface mesh model M of the obtained three-dimensional matrix-edge extension component model (without screw holes) is used as the three-dimensional matrix network model M for generating the printing path. Matching threaded connection structures are provided between the connecting rod and the edge extension and the flange fixing member.

[0069] S2: Manufacture the water-soluble matrix Adopt the fused deposition modeling (FDM) process, use a 3D printer and its consumables (PVA), and integrally print the water-soluble three-dimensional matrix designed in step S1 and the edge extensions at both ends thereof to ensure accurate geometric dimensions and structural integrity.

[0070] S3: Install the matrix fixing frame The 3D matrix-edge extension assembly printed in step S2 is connected to the flange fastener via connecting rods to form a rigid integral frame. The flange at the end of the main robotic arm is adjusted to zero, and the flange fastener is precisely aligned with it and tightened with bolts.

[0071] S4: Winding Path Planning This step is executed on a host computer (personal computer), using the Grasshopper platform and its Kangaroo2 physics simulation plugin on the Rhino platform. The input is the outer surface mesh model M (also known as the 3D triangular mesh model M or the single-layer triangular mesh model M on the outer surface) of the 3D matrix-edge extension component model (without screw holes), and the output is a sequence of pose and velocity commands to control the coordinated motion of the two robotic arms. The specific steps are as follows: S41. Multi-branch surface splitting and layering S411. Physical simulation generates branch cutting lines: In Grasshopper, first obtain the outer surface mesh model M of the 3D base-edge extension component model (without screw holes) (as in this embodiment). Figure 4 a), and the polylines E1, E2, …, E at the ends of its branches. N (N is the number of branches, which is 3 in this example). Then, a dynamic model is constructed using the Kangaroo2 plugin, generating smooth and reasonable branch cutting lines based on the polylines at the ends of each branch, including: Assign collision properties: Each segment and its endpoints of each polyline are designated as a "collider," specifically implemented using the Kangaroo2 Collider component. In this example, the collision radius (Radii) is set to 1, and the collision strength (Collider_Strength) is set to 1.0e+7.

[0072] Apply mesh adhesion force: Apply a force to all nodes on each polyline to make them adhere to the mesh model M, implemented through the OnMesh component. In this embodiment, the strength is set to 1.0e+6.

[0073] Imposing shape constraints on the curve: Length Elasticity: Applying elastic properties to all segments of each polyline to maintain their initial length, implemented through the Length component. In this embodiment, the Strength is set to 9.0e+4.

[0074] Angular elasticity: Applying angular elasticity to the angle formed by adjacent segments in each polyline, implemented through the Angle component. In this embodiment, the angle (RestAngle) is set to 180°, and the strength (Angle_Strength) is set to 1.0e+7.

[0075] Apply cleaving force to all nodes on each polyline: apply an attraction pointing to all other polylines, implemented using the OnCurve component. In this embodiment, the strength is set to 2.0e+2.

[0076] The iteration count of the ZombieSolver solver in Kangaroo2 is set to 5000. After starting the simulation, the initial polylines E1, E2, …, E N Evolving under multiple constraints, it eventually forms a smooth curve distributed along the model surface, which is recorded as branch cutting lines Cut1, Cut2, …, Cut N .

[0077] S412. Mesh Splitting and Processing: In Grasshopper, the acquired original mesh model M is cut along each cutting line. i Perform the cutting. Keep each cut. i With the corresponding end edge line E i The mesh between the branches. For each branch tubular mesh obtained, a homogenization process is performed: a mesh homogenization reconstruction operation is performed using the TriRemesh retriangulation component in the Grasshopper environment to optimize mesh quality and obtain smooth branches tubular surfaces. Figure 4 b).

[0078] For each independent branched tubular mesh surface, a set of uniformly distributed, seam-aligned, and directionally consistent closed layered lines are generated as a geometric reference for subsequent fiber path planning. The specific operation process is as follows: (1) Define the distance field: the currently processed independent branch tubular mesh. Define the end closest to the center of the model as the beginning (i.e., the cut line). i The end furthest from the center of the model is defined as the tail end (i.e., the end edge line E). i In the Grasshopper environment, using Shortest Walk, MeshGeodesic, or similar components, compute the distance from each node on the mesh to the head cut. i The geodesic distance (i.e., the shortest path length along the grid surface). and each node to the tail E i geodetic distance Y This calculation assigns a field value to each node of the grid. This forms a scalar distance field covering the entire branch surface. Within this field, Cut... i The distance between all points on the line is 0, Ei The distance between all points on the line is 1.

[0079] (2) Extracting and generating contour lines (layer lines): setting the target average layer height H (For example, 2mm), this is an adjustable parameter that determines the distance between adjacent layer lines. Based on the distance field generated in the previous step, a series of distance isopleths are extracted on the triangle edges of the grid using linear interpolation with an interval of h. Specifically, in Grasshopper, using IsoVist, Contour, or through Field Line with a custom script component, the distance field and a series of field values ​​are input. )(in , It corresponds to each grid point + The average value of , [] is the floor function. ), to generate multiple closed contour lines. These contour lines serve as the "layer lines" of the branch surface.

[0080] (3) Unifying the joint location and generation direction: After generating the initial multi-layer contour lines (layer lines), it is necessary to unify the joint location and generation direction to establish a consistent parameterized benchmark for subsequent steps. A joint refers to the starting point (and also the ending point) of each closed layer line; the direction refers to its circumferential direction (clockwise or counterclockwise). The specific operation process is as follows: a. Establish a baseline, using the first layering line (usually the cut line) i The seam point P0 of the first line generated by its offset or its surrounding direction (e.g., uniformly preset to counterclockwise) is used as the global reference.

[0081] b. Align the seams of the remaining layers: for each of the other layer lines C i (i > 0), calculate the distances along C for all points on it. i Calculate the arc length distance from the curve itself to the reference point P0, and locate the point Q that is closest to the reference point. i ; with Q i For the new seam point, close curve C at this point. i "Interrupt" and reconnect, making Q i This becomes its new starting point and end point, thus ensuring that the seams of all layer lines are roughly aligned in the circumferential direction.

[0082] c. Unify Generation Direction: Check the tangent direction of each layer line (including the baseline line) at its new seam point. If the dot product of the tangent direction of a line and the baseline tangent direction is negative (i.e., the angle is greater than 90°), it is determined that its direction is opposite to the baseline. In this case, the Reverse Curve component needs to be used to flip the direction of the line so that it is consistent with the baseline direction. Through the above steps, a set of layer lines with aligned seams and consistent direction can be obtained, laying the geometric foundation for the subsequent generation of precisely corresponding division points.

[0083] S42. Branched fiber optic cabling generation Based on the aligned layer lines, the fiber routing for each branch is generated, and the specific steps are as follows: (1) Joint offset: For a certain branch, set its first x The offset of the seam point of the layered line is: , in, Representing the i ( i <x ) layer lines and the first i The average floor height between +1 floor lines The preset offset angle (45° in this embodiment) is used. All seam points of the layered lines of this branch are moved along the curve by their corresponding offset amounts to obtain new layered lines. Figure 4 d).

[0084] (2) Division and Connection: Generate division points on each new layer line. The number of division points required for each layer line. n Calculated by the following formula: , in, For the first i The length of the new layer line, This represents the total number of layer lines. d The desired fiber spacing is set to 2.0 mm in this embodiment, and [] represents the round-down function. This quantity... n Divide each dividing line into equal parts to obtain a series of corresponding dividing points ( Figure 4 e).

[0085] (3) Connect the points with the same index on adjacent layer lines sequentially with a smooth curve (such as a B-spline curve) to form the preliminary fiber routing line of that branch. Figure 4 f).

[0086] S43. All branch fiber optic cables are connected as a whole: The fiber optic cabling generated on each independent branch is smoothly and continuously connected at the joint positions of the multi-branch substrate to form an uninterrupted fiber optic cabling covering the entire surface of the substrate. The specific steps are as follows: Filtering candidate connection pairs: For fiber optic cabling CRV on the target branch A (Starting point P) A In the fiber optic cabling of all other branches, search for the starting point Q. B With P A The curve Crv whose Euclidean distance is less than a preset threshold D (e.g., 30 mm in this embodiment) B , forming the candidate set {Crv B}

[0087] Evaluation and selection of optimal connection pairs: For the candidate set {Crv B Each curve Crv in} B Perform the following operations to quantitatively evaluate its comparison with CRV A Connection quality: Constructing a bridging curve: with P A and Q B As the endpoint, with Crv A In P A dot, CRV B In Q B Using the tangent direction of a point as a constraint, construct a C1 continuous (tangent continuous) 3rd-order B-spline curve Bridge. {AB} .

[0088] Calculate the smoothness metric f {AB} Bridge {AB} The projection curve Proj is obtained by vertically projecting it onto the mesh model M. {AB} Calculate Proj {AB} Total bending energy as a smoothness measure f {AB} Its discretization calculation formula is: f {AB} ≈Σ (κ i 2 × Δs i ), where κ i For Proj {AB} The curvature of the i-th sampling point, Δs i This corresponds to the arc length infinitesimal element.

[0089] Select the optimal pair: Compare all candidate f pairs. {AB} Value, select f {AB} The candidate curve Crv with the smallest value B As CRV A The optimal connection pair.

[0090] Perform smooth geometric joins: for the selected optimal join pairs (Crv) A , Crv B Perform geometric join operations to form a continuous path: Trimming the original curve: To ensure a natural transition at the connection points, trim the curves from Crv... A P A Points and CRV B Q B A small distance (e.g., 2mm in this embodiment) is cut backward from the point to obtain the new endpoint P. A' and Q B' .

[0091] Construct the final bridging curve: with P A' and Q B' As the endpoint, with its position in Crv A and CRV B Using the tangent line as a constraint, reconstruct a C1 continuous B-spline bridge curve, Bridge. {final} .

[0092] Connection: Connect CRV A (After trimming) Bridge {final} ,Crv B (After cutting) they are connected end to end in sequence to form a continuous three-dimensional curve.

[0093] Iterative integration: Repeat steps (1) to (3) for all unconnected fiber optic cabling until a fiber optic cabling covering the entire substrate is formed. Figure 4 f).

[0094] S44. Path integration based on surface unfolding and phase adjustment To resolve the conflict between the "uninterrupted" requirement of continuous fiber printing and the limited rotation angle range (-360° ~ +360°) of the robotic arm's rotation axis (A6 axis), this step transforms the complex fiber arrangement path in three-dimensional space into a path planning problem on a two-dimensional plane through parametric surface unfolding and mapping. After completing the continuous sorting and connection of the paths on the plane, it is then inversely mapped back to three-dimensional space, thus obtaining a physically printable, continuous, and uninterrupted complete fiber arrangement. The specific process is as follows: S441. Construct the mapping relationship between the roll surface Surf0 and the unfolded surface Surf1. To handle the rotational limitation of the robotic arm, a deployable, parametric surface model (Surf0) and its mapping to a plane (Surf1) are constructed. This step is implemented in the Grasshopper visual programming environment, and the specific implementation steps are as follows: Selecting a reference edge: Using the plane where the end flange of the main robotic arm is located as a reference, automatically identify or manually select the opening edge closest to this plane on the mesh model M as the near-end edge E. a And the farthest opening edge from the plane is designated as the farthest edge E. b .

[0095] Generate basic parametric surfaces: Calculate the axial distance field: Similar to the S421(1) method, on the original mesh model M, calculate the distance field from the near edge E. a To the far edge E b The geodesic distance field. This distance field is a scalar field, E. a The distance between all points on E is 0. b The distance between all points is 1. Calculate E. a To E b Maximum geodetic distance L max .

[0096] Generate basic parametric surfaces: Set an axial segment number N seg (For example, N) seg = ceil(L max / 5mm), ceil is the round-up function). Using the IsoVist or Contour component, based on the above distance field, the distance values ​​are extracted as follows: The contour lines are a series of approximately parallel closed loops on the model surface. This series of contour lines is then input into Grasshopper's Loft component, and a lofting operation is performed to generate a smooth, single NURBS surface, denoted as the base surface S. base The U direction (lateral parameter) of this basic surface corresponds to the wrapping direction of each contour line, and the V direction (longitudinal parameter) corresponds to the direction from E... a To E b The axial direction. Its U parameter range [0, 1] corresponds to the actual rotation phase range of the robotic arm flange from -180° to +180°.

[0097] Construct a complete scroll surface Surf0 (covering a virtual rotation range of -360° to +360°): For the base surface S... base Perform parameter extension to cover the complete rotation cycle. Specifically, within the parameter domain, extend S... base The range of the U-parameters is virtually extended from [0, 1] to [-0.5, 1.5]. Wherein: the surface with U-parameters in the interval [0, 1] corresponds to the actual rotation phase interval of the robotic arm. The surfaces with U-parameters in the intervals [-0.5, 0) and (1, 1.5] are S-parameters. base The periodic extension in the U direction corresponds to the parameterized position of the theoretical path on the surface during the reset period after the robotic arm rotates beyond its limit. In implementation, the parameter domain of the surface can be manipulated using components such as Construct Domain and Deconstruct / Reparameterize, and the S-axis can be copied using CopyTrim or scripts. basePeriodically extending the boundary states at U=0 and U=1, we finally obtain the complete scroll surface Surf0 with a parameter domain of [-0.5, 1.5] × [0, 1]. Figure 4 g).

[0098] Define the surface Surf1 and establish the mapping: Creating a Plane: Use the Rectangle component to create a virtual rectangular plane as the tiling plane Surf1. Its length 'a' is 2000 times the length of the shortest V-shaped isoparametric curve on the scroll surface (Surf0), and its width 'b' is 1000 times the length of the shortest U-shaped isoparametric curve on the scroll surface (Surf0). The size of this virtual rectangular plane has no special requirements, only that it be large enough to provide sufficient operational space for subsequent tiling plane sorting and connection algorithms, avoiding algorithmic misjudgments.

[0099] Establish a linear mapping from Surf0 to Surf1 based on parametric coordinates. This mapping can be implemented using Surface Closest Point, Evaluate Surface, and custom mathematical operation components in Grasshopper. The mathematical rule is: for any point P(u, v) on Surf0, it maps to a point P'(x, y) on Surf1. The mapping formulas are: x = v × a (axial distance mapped to plane length), y = (u - [u]) × b (fractional part of the U parameter mapped to plane width, [] is the floor function). This rule ensures that the periodicity of the U parameter in Surf0 corresponds correctly to the planar coordinates of Surf1.

[0100] S442. Generate the inter-fiber interconnection path network Net0 and spread the inter-fiber interconnection path network Net1: On the spreading plane Surf1, a set of key connecting channels are defined to guide the transition sections of subsequent different branch fiber paths. This ensures that the network of connecting paths between spreading fibers is uninterrupted and that the rotational range of the main robotic arm flange is matched.

[0101] (1) Define key cutting planes: Select a set of planes to generate Net0. These planes are usually determined based on the geometric features of the model and need to intersect all edges of the 3D mesh model M. They can be selected according to the complexity of the model, manually defined in 3D modeling software, or automatically calculated by an algorithm. The algorithm scheme for selecting cutting planes in this invention is as follows: Obtain the geometric center points of all edges, obtain all schemes of selecting any 3 points from the geometric center points, and obtain the plane Pln formed by the three points of each scheme. a Then select three straight lines formed by every two points from the three points, and select the line that intersects with Pln. a Three planes Pln that are perpendicular to and intersect the above linesb ,Pln c ,Pln d Let the set of all planes be {Pln}; close all the end openings of the 3D mesh model M with a new layer of facet mesh to form a complete closed mesh M. cap Of all planes {Pln}, select the one that is closest to M. cap A plane whose intersections form a plane with one line of intersection.

[0102] Generate the inter-fiber interconnection path network: Calculate the intersection lines between the selected key cutting plane and the original mesh model M to obtain a set of three-dimensional spatial curves. In Grasshopper, this can be achieved using the MeshXPlane component. Combine these intersection lines with the edge lines of the three-dimensional mesh model to form the path network; the resulting spatial curves are the inter-fiber interconnection path network (Net0).

[0103] Mapping to the spreading plane: Using the mapping relationship established in step S441, this set of spatial curves is mapped from the roll surface Surf0 to the spreading surface Surf1. In Grasshopper, the parametric coordinates (u,v) of the 3D curve points on Surf0 can be found using the Surface Closest Point component. Then, the corresponding points on Surf1 are calculated using the mapping formula in step S441. Connecting these points yields the network of interconnected paths between spreading fibers (Net1).

[0104] S443. Path Mapping, Sorting, and Plane Connectivity On the Surf1 laying plane, all mapped fiber paths are integrated. The main robotic arm's rotational limitations are addressed by adjusting the phase, the printing sequence is optimized by rearranging, and finally, they are connected into a continuous, uninterrupted path. The specific implementation steps are as follows: Path mapping: All three-dimensional continuous fiber routing lines generated in steps S41-S43 are mapped from Surf0 to Surf1 through the mapping relationship established in S441, resulting in a set of planar curves {Crv}. Plane} Phase adjustment: Let the width of Surf1 be W, corresponding to the virtual rotation range of the main robotic arm from -360° to +360°. Traverse {Crv Plane Each curve Crv in}: a. Calculate its Y-coordinate range on Surf1 [Y min , Y max ]; b. If Y min < 0.25W or Y max If the value is greater than 0.75W, then calculate an integer k such that the range of the translated Y coordinate is [Y]. min+ k×0.5W, Y max The main part of [+ k×0.5W] (e.g., more than 50% of the curve length) falls within the interval [0.25W, 0.75W]. c. Shift the curve Crv along the positive Y-axis of Surf1 by a distance of k×0.5W. Figure 4 h). This operation is equivalent to commanding the robotic arm's A6 axis to rotate k×360° before printing this path segment to perform a phase reset, thereby ensuring that the actual printing stroke is within the limits. Record the translation amount k for each curve.

[0105] (3) Interlacing and rearranging of planar paths: a. Initial sorting: All phase-adjusted planar curves {Crv Plane Sort the elements in ascending order according to their Y-coordinates on Surf1, resulting in an initial ordered list List0.

[0106] b. Alternating and rearranging (this example uses n=3): Round 1: Insert the elements from List0 into two new lists, List0 and List1, in sequence, alternating between them. A1 and List B1 (That is, the first element is put into A1, the second into B1, the third into A1, and so on).

[0107] Round 2: For each List A1 and List B1 The internal elements are then subjected to the alternating splitting operation described above again, resulting in two sublists for each element, for a total of four lists.

[0108] Round 3: For each of these four sublists, perform internal alternating splitting again, ultimately resulting in 2. 3 = 8 ordered sublists.

[0109] Merge sequences: Create a new empty list List Sorted Then iterate through these 8 sublists: in turn, take the first element of each sublist, remove it from the atomic list, and then append it to the List. Sorted End. Repeat this polling process until all sublists are empty. The final list is... Sorted This is the rearranged target sequence. This algorithm changes the distribution of the curve's starting point in the Y direction from concentrated to dispersed, achieving a gradual change in the density of the filaments at the start of printing.

[0110] (4) Path connection based on the inter-fiber interconnection path network Net1 a. Sequential join: following the List SortedThe order in which each pair of adjacent curves (Crv) is processed sequentially. i (Current curve) and Crv i+1 (Next curve).

[0111] b. Generate the transition segment: On Surf1, using Crv i The endpoint and Crv i+1 The starting points are the two ends. In the network of interconnected fiber paths Net1, find the shortest path connecting these two points, consisting of curve segments within Net. Fit this shortest path with a smooth curve (such as a B-spline) to obtain the transition curve Bridge. i .

[0112] c. Connecting the beginning and end: Connecting CRV i Bridge transition curve i ,Crv i+1 Connect them sequentially, end to end, and merge them into a longer continuous curve.

[0113] d. Iterative integration: For List... Sorted Repeat steps b and c above for connecting all adjacent curve pairs. This process connects all independent fiber bundles sequentially end-to-end along the transition path of network Net1, ultimately generating a path named Path on the spreading plane Surf1. Final A single row of cabling that covers all original paths and is continuous and uninterrupted. Figure 4 i). Because List Sorted The path itself is obtained through interleaving and rearranging; its initial curve distribution is sparse, gradually becoming denser later, thus the final path... Final The initial segment exhibits a low path density, which gradually increases as the path extends, forming an overall distribution pattern from sparse to dense.

[0114] To improve structural performance, the fiber density distribution can be optimized based on the stress conditions. A step of optimizing the fiber arrangement based on a specific field is added between S43 and S44. This step is optional; for other implementations, it can be selected whether to perform this step based on actual needs. The specific operation is as follows: Obtaining the density field: In this embodiment, a finite element analysis (FEA) is performed on the three-dimensional matrix model. The load scenario is set with fixed boundary conditions at both ends of the main body of the matrix model. The load on the edge of the support rod is set to be directed towards the main body. The strain energy density distribution is calculated, and the energy field formed by the point cloud on its outer surface and its corresponding field values ​​is normalized and used as the density field. Figure 5 a).

[0115] Constructing a non-uniform mesh transformation pair Ma-Mb: In Grasshopper, using the RemeshByColour component, with the density field as the color input, transforms the original uniform mesh M... base The mesh is converted into a variable-density mesh Mb where vertex density is proportional to energy. Then, a length-based homogenization re-partitioning is performed on mesh Mb (e.g., via the Kangroo plugin) to obtain a mesh Ma with uniform vertex distribution, but sharing the same surface geometry and parameterized domain as Mb. At this point, a one-to-one mapping relationship is established between Ma (uniform mesh) and Mb (variable-density mesh), and they can be converted to each other using node indices or parametric coordinates.

[0116] Iterative mapping optimization: (1) Path point mapping: The fiber routing generated in step S43 is automatically decomposed into multiple smooth curve sequences Path based on Grasshopper's Explode component. i Each curve is divided into an equidistant point sequence P with a density of d=2mm. ij Using the MeshMap component of Kangaroo2 in Grasshopper, points are mapped from Ma to Mb to obtain a new point sequence.

[0117] (2) Mapping iteration: The new sequence points are further mapped from Ma to Mb. The entire mapping process is iterated N times (N=2 in this embodiment) to obtain the final point sequence.

[0118] (3) Path reconstruction: The final point sequence of each curve is reconnected with a smooth curve to form Path'. i By sequentially combining curve sequences, new paths are obtained. The final fiber paths will be denser in regions of high strain energy density, resulting in a path layout with superior mechanical properties. Figure 5 b).

[0119] S45. Path point and attitude calculation On the fiber busbars generated by S44, according to the dot density de Select a point. de From the formula de = de 0 ( a | n · m | +1) Confirm, where de 0 is the base point density (e.g., 2 mm in this embodiment). n Let be the surface unit normal vector of the sampling point on the three-dimensional mesh model (referred to as the surface unit normal vector of that point). m The preset fixed reference unit normal vector (in this embodiment, the direction of the axis of the main robotic arm end flange) is used. aThe adjustment coefficient is 10.0 in this embodiment. This formula adaptively increases the path point sampling density in areas where the surface of the 3D mesh model has a large tilt relative to the fixed reference direction (m), thereby reducing the amplitude of attitude change between adjacent sampling points, thus reducing the kinematic complexity of the collaborative robotic arm and ensuring the attitude control accuracy and printing pressure stability of the print head on complex curved surfaces.

[0120] For each path point, calculate the surface unit normal vector of that sampling point on the 3D mesh model. n and the surface unit normal vector of the sampling point on the three-dimensional mesh model t The normal (N) of the target attitude plane of the printhead is given by the formula N = (1-β). n - β t It is determined that β is the tilt parameter (β = 0.10 in this invention) to ensure that the printhead applies a certain clamping force to the fiber in the forward direction.

[0121] S46. Robotic Arm Pose Derivation and Velocity Planning: Based on the kinematic model of the dual robotic arms and the calibrated base coordinate system, the position and orientation plane of each path point are calculated to determine the end pose of the master robotic arm and the slave robotic arm, generating motion command lines robot_a_path and robot_b_path.

[0122] Meanwhile, to maintain a stable relative speed between the print head and the 3D substrate surface, the robotic arm's operating speed... The dynamic adjustment is based on the attitude change rate between adjacent path points, and the formula is: , in: For the print head from the first q -1 sampling point moved to the first q The velocity of movement at each sampling point; The preset base motion speed; For the first q The angle between the surface unit normal vector at the nth sampling point and the fixed reference unit normal vector, minus the angle between the nth and nth sampling points. q The absolute value of the difference between the angles corresponding to -1 sampling points (i.e., the change in the overall tilt angle): ; For the first q The sampling point and the first q The angle between the projected components of the surface unit normal vector at -1 sampling point on a plane perpendicular to the fixed reference unit normal vector (approximately the torsion angle of the main robotic arm 03 flange): ; No. q The angle between the surface unit normal vector at each sampling point and the fixed reference unit normal vector (i.e., the current absolute tilt): ; , , The preset weighting coefficients (in this invention, the weighting coefficients are taken as follows) =100, =100, =2), used to adjust the influence of various factors on speed. The fixed reference unit normal vector is the axis of the end flange of the main robotic arm. It is a fixed reference direction and serves as the absolute benchmark for measuring local attitude changes on the substrate surface.

[0123] S5: Wrap Printing This step is completed collaboratively between the host computer and dual UR robotic arms (such as UR7e). Using the dual robotic arm system, the main robotic arm controls the spatial pose of the substrate, carries the print head from the robotic arm, and through synchronous control and force feedback adjustment, winds and prints thermoplastic continuous carbon fiber composite material (PA12-CCF: Nylon 12 matrix continuous carbon fiber reinforced composite material) onto the substrate surface according to the planned path to form a wound body.

[0124] The synchronous control and force feedback adjustment operations are as follows: Synchronous control: The host computer establishes high-speed communication with the controllers of the two robotic arms through the real-time data exchange (RTDE) protocol, and synchronously sends the target pose according to the robot_a_path and robot_b_path instruction sequences to achieve precise coordinated motion.

[0125] Force feedback adjustment: During printing, force data (Fx, Fy, Fz) is collected in real time from the force sensor at the end of the robotic arm and uploaded to the host computer via the RTDE protocol. The force monitoring thread in the host computer continuously analyzes the magnitude of the resultant force. After a 10-second delay following program startup (waiting for system stabilization), the adjustment mechanism is activated. When the resultant force is greater than 5.0N, the pressure is determined to be too high. The host computer sends a stopL command to the slave robot arm controller to interrupt the current motion, and increases the Z-axis offset robot_b_z_offset (upper limit + 0.01m) according to the step size of 0.001m to raise the print head.

[0126] When the resultant force is less than 2.0N, the pressure is deemed insufficient, and the movement is interrupted. The Z-axis offset is reduced (lower limit -0.01m) to lower the print head.

[0127] After the offset is adjusted, the host computer recalculates the pose of subsequent path points based on the new robot_b_z_offset and resumes motion. This process achieves real-time, closed-loop constant-pressure winding control.

[0128] S6: Post-processing Once printing is complete, remove the winding body from the robotic arm and disassemble the fixing frame. Use a cutting tool to cut off the fiber layer along with the extension at the point where it connects to the substrate, and then finely polish the ends.

[0129] S7: Core Removal The polished workpiece is completely immersed in a water tank filled with deionized water, with the water temperature controlled between 60℃ and 90℃. The dissolution of the PVA matrix is ​​accelerated by water circulation or stirring. After the internal PVA has completely dissolved, the workpiece is removed and the inner cavity is rinsed with clean water. Finally, it is dried to obtain a complex hollow structure part with a smooth internal continuous carbon fiber structure.

[0130] The above embodiments are only for illustrating the technical concept and features of the present invention, and are intended to enable those skilled in the art to understand the content of the present invention and implement it accordingly. They should not be construed as limiting the scope of protection of the present invention. All equivalent changes or modifications made in accordance with the spirit and essence of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A method for generating a 3D printing path for a multi-branched tubular carbon fiber part, characterized in that, It includes the following steps: (1) The three-dimensional mesh model of the multi-branched tubular carbon fiber part is split into multi-branch surfaces, and layer lines are generated for each independent branch surface; (2) Generate fiber routing lines for each branch based on the layered lines of each branch; (3) Connect the fiber routing lines of each branch at the branch joint to form a fiber routing line covering the entire surface of the part; (4) Path integration based on surface unfolding and phase adjustment, which includes: Surface spreading mapping steps: Construct a mapping relationship between the scroll surface and the two-dimensional spreading surface, wherein the U-parameter direction of the scroll surface corresponds to the rotation angle of the end flange of the main robotic arm that controls the attitude of the three-dimensional substrate, and the outer surface contour of the three-dimensional substrate is adapted to the inner cavity contour of the multi-branched tubular carbon fiber part; the V-parameter direction of the scroll surface corresponds to the geodesic distance field from the near edge to the far edge on the three-dimensional mesh model, the near edge is the opening edge on the three-dimensional mesh model that docks with the end flange of the main robotic arm, and the far edge is the opening edge that is farthest from the end flange of the main robotic arm; the mapping relationship maps the U-parameter direction of the scroll surface to the width direction of the two-dimensional spreading plane, and maps the V-parameter direction of the scroll surface to the length direction of the two-dimensional spreading plane; Generation of inter-fiber connection path network: Select at least one cutting plane, and map the intersection line of the cutting plane and the 3D mesh model and all edge lines of the 3D mesh model from the roll surface to the spreading surface according to the mapping relationship to obtain the inter-fiber connection path network. The cutting plane is determined through the following steps: Obtain the geometric center points of each opening edge of the three-dimensional mesh model; A set of candidate planes is generated based on the geometric center point; Close all openings in the three-dimensional mesh model to form a closed mesh model; From the candidate plane set, select a plane that intersects with the closed mesh model and forms a line with a value of 1. Spreading surface sorting and connection steps: Map the fiber routing line covering the entire part surface in step (3) onto the two-dimensional spreading surface to obtain a set of planar curves; based on the coordinate values ​​of the starting point of each planar curve in the width direction of the two-dimensional spreading surface, alternately interweave and reorder; according to the sorting order, using the connecting path network between the spreading fibers as the connecting carrier, connect the beginning and end endpoints of adjacent planar curves to form a single, uninterrupted planar routing line on the two-dimensional spreading surface; Inverse mapping step: The planar wiring is inversely mapped back to the scroll surface to obtain a continuous three-dimensional fiber wiring on the three-dimensional substrate surface.

2. The path generation method according to claim 1, characterized in that, The multi-branched tubular carbon fiber component includes a main pipe extending along a first axis, the main pipe having a first opening and a second opening disposed opposite to each other along the first axis; at least one branch pipe communicating with the main pipe, the branch pipe having a third opening away from the main pipe; wherein the at least one branch pipe is disposed at the same axial position as the main pipe; the first opening, the second opening and the third opening are all open ends, together forming at least three openings.

3. The path generation method according to claim 1, characterized in that, In step (4), the step of generating a set of candidate planes in the cutting plane determination step includes: A reference plane is determined based on any three of the geometric center points; For each of the reference planes, three associated planes are generated, wherein each associated plane is perpendicular to the reference plane and intersects the line determined by any two of the three points; The candidate plane set is composed of the reference plane and all the associated planes.

4. The path generation method according to claim 1, characterized in that, The surface unfolding mapping step (4) specifically includes: Define the proximal edge and the distal edge: with the plane where the end flange of the main robotic arm is located as a reference, select the proximal edge and the distal edge on the three-dimensional mesh model; Generate a basic parametric surface: Calculate the geodesic distance field from the near edge to each point on the surface of the 3D mesh model. The geodesic distance field is 0 at the near edge and the maximum geodesic distance at the far edge. Based on the geodesic distance field, extract closed contour lines distributed along the model axis. Generate a NURBS surface as the basic surface based on the closed contour lines. The U parameter direction of the basic surface corresponds to the wrapping direction of the contour lines, and its parameter range is set to [0, 1] to linearly map the actual rotation phase interval of the main robotic arm from -180° to +180°. The V parameter direction corresponds to the axis from the near edge to the far edge, and its parameter range is set to [0, 1] to linearly map the normalized geodesic distance from the near edge to the far edge. Constructing a complete scroll surface: Periodically extend the range of U parameters of the base surface from [0, 1] to [-0.5, 1.5] to construct a complete scroll surface with a parameter domain of [-0.5, 1.5] × [0, 1], where the U parameters in the sub-interval of [0, 1] correspond to the actual rotation phase interval of the robotic arm; Generate a two-dimensional spreading plane and establish a mapping: Create a rectangular plane as the spreading surface, whose length is defined as mapping the maximum geodesic distance, and whose width direction is defined as mapping the virtual rotation range of the main robotic arm from -360° to +360° to provide the periodic margin required for phase adjustment; establish the mapping relationship from the scroll surface to the spreading plane; And / or, in the spreading surface sorting and connection step of step (4), the phase adjustment is: the phase adjustment is achieved by translating the planar curve along the width direction of the spreading surface by an integer multiple of its width, so as to adapt to the actual rotation working range of the main robotic arm, specifically including: The width direction of the spread surface is defined as mapping the virtual rotation range of the main robotic arm from -360° to +360°, with a total width of W, where the range [0.25W, 0.75W] corresponds to the actual rotation working range of the main robotic arm from -180° to +180°. For each planar curve, calculate its coordinate range [Y] along the width direction on the spread surface. min , Y max ]; If Y min < 0.25W or Y max If the value is greater than 0.75W, then calculate and determine an integer k with the smallest absolute value, such that after the curve is translated along the width direction by a distance of k×0.5W, the length of the translated curve in the interval [0.25W, 0.75W] is not less than a preset threshold. And / or, in the spreading and connecting steps of step (4), the alternating interlacing reordering specifically includes: Arrange all planar curves in ascending order according to the coordinates of their starting points on the unfolded surface to obtain the initial sequence; Perform n rounds of alternating interleaving operations on the initial sequence, where n > 1. In the k-th round, the sequence obtained in the previous round will be used to perform the operation on the 2-th sequence. k-1 Each ordered sublist is split into two new sublists; After completing n rounds of splitting, we get 2. n A final sublist; By using a round-robin extraction method, the first element of the current element is extracted from each of the final sublists in turn until all sublists are empty, thereby forming the reordered target sequence. And / or, in the spreading and connecting steps of step (4), the specific steps for connecting the beginning and end endpoints of adjacent planar curves using the interconnecting path network between the spreading fibers as the connecting carrier are as follows: Sequential connection: Process each pair of adjacent planar curves in the order determined by the sorting. Generating a transition segment: For a pair of adjacent planar curves after sorting, the end point of the first curve and the start point of the second curve are used as the start and end points of the transition segment. In the network of connecting paths between the spread fibers, the shortest path that can connect the start and end points is found, and the shortest path is fitted into a smooth transition curve. Connecting the first and last curves: Connect the first curve, the transition curve, and the second curve in sequence to form a longer continuous curve; Iterative integration: For all adjacent curve pairs after sorting, repeat the steps from generating transition segments to iterative integration until all planar curves are integrated into a single continuous and uninterrupted planar cabling line on the spread surface.

5. The path generation method according to claim 1, characterized in that, The multi-branch surface splitting in step (1) includes: Obtain the edge lines of the openings at the ends of each branch of the three-dimensional mesh model; A dynamic model is constructed, and collision constraints, adsorption forces that make it close to the surface of the three-dimensional mesh model, elastic constraints that maintain its length, bending constraints that maintain the angle of its adjacent line segments, and gravitational constraints that point to other edge lines are applied to each of the said end opening edge lines. Smooth branch cutting lines are generated through iterative solutions using physical simulations. The three-dimensional mesh model is divided along each of the branch cutting lines to obtain multiple independent branched tubular meshes. And / or, the step (1) of generating layered lines for each independent branch surface includes: For each branched tubular grid, calculate the geodesic distance field on its surface from the dividing line to the end edge line; Using a preset average layer height as the spacing, a series of closed contour lines are extracted from the geodesic distance field as initial layer lines; The initial layer lines are aligned by seam alignment: taking the first layer line as a reference, the seam points of the remaining layer lines are adjusted to the positions that are closest to the reference seam point along the curve arc length. The initial layer lines are then subjected to orientation unification: the wrapping direction of all layer lines is unified.

6. The path generation method according to claim 5, characterized in that, In step (2), generating the fiber optic cabling for each branch includes: Calculate the first x The formula for calculating the offset of the seam point of the layered line is as follows: ,in, Representing the i The layered line and the first i The average floor height between +1 floor lines i <x , This is the preset offset angle value; Move all the seam points of the layered lines along their own curve direction by the corresponding offset amount. Generate the same number of equal division points on each layer line after offset, and connect the equal division points with the same index on adjacent layer lines in sequence with a smooth curve to form the fiber busbar of that branch.

7. The path generation method according to claim 1, characterized in that, The step (3) of connecting the fiber routing lines of each branch at the branch joint includes: Filtering candidate connection pairs: For the target fiber busbar, among all the other branches of the fiber busbar, search for fiber busbars whose three-dimensional spatial distance between their starting point and the starting point of the target fiber busbar is less than a preset threshold, and form a candidate set; Evaluate connection quality: For each fiber busbar in the candidate set, construct a spatial bridging curve connecting the starting point of the target fiber busbar and the starting point of the candidate fiber busbar, and project the spatial bridging curve vertically onto the surface of the three-dimensional mesh model. Calculate the total bending energy of the projected curve as a measure of connection smoothness. Selection and Connection: Select the fiber busbar with the smallest smoothness metric value from the candidate set as the optimal connection pair, and construct a smooth geometric transition curve to connect the two fiber busbars in the optimal connection pair into a continuous curve.

8. The path generation method according to claim 1, characterized in that, It also includes fiber arrangement optimization steps based on specific fields: The strain energy density distribution field of the multi-branched tubular part under the target stress condition is obtained as the density field of the fiber arrangement. The fiber routing obtained in step (3) is iteratively mapped and optimized based on the density field to make the fiber paths more densely distributed in regions with high strain energy density and more sparsely distributed in regions with low strain energy density.

9. The path generation method according to claim 8, characterized in that, The iterative mapping optimization specifically refers to: Construct mesh mapping pairs: Using the density field as the input color, obtain a variable density mesh whose vertex density is proportional to the density field value; then re-homogenize the variable density mesh to obtain a new uniform mesh, thus forming a mesh pair; Perform iterative mapping: Map the set of points on the fiber cabling line between the uniform grid and the variable density grid. During mapping, the path points are directly mapped to the same parameter coordinates on the variable density grid according to the parameter coordinates of the path points on the uniform grid, so that they gather in the high density area. The mapping process is iterated multiple times until a preset number of iterations is reached.

10. The path generation method according to claim 1, characterized in that, It also includes path point and attitude calculation steps, which include: sampling path points on the three-dimensional fiber routing line according to adaptive density, and calculating the target attitude of the robotic arm that controls the attitude of the print head for each sampling point; Wherein, the adaptive density is given by the formula de = de 0 ( a | n · m | + 1) Confirm, de To adapt to density, de 0 is the base point density. n Let be the surface unit normal vector of the sampling point on the 3D mesh model. m The preset fixed reference unit normal vector is preferably located in the direction of the centerline of the end flange of the main robotic arm. a This is the preset adjustment coefficient; The target posture of the robotic arm is defined by the normal of the print head target posture plane, and the normal of the print head target posture plane is given by the formula N = (1- β ) n - βt Calculate, where N is the normal to the target orientation plane of the print head. n Let be the surface unit normal vector of the sampling point on the 3D mesh model. t Let be the unit tangent vector of the fiber cabling at the sampling point. β The preset tilt parameter, and 0 < β < 1.

11. The path generation method according to claim 10, characterized in that, The path point and attitude calculation steps also include velocity planning: Based on the attitude change rate between adjacent sampling points, the motion speed of the print head relative to the three-dimensional substrate between each sampling point is dynamically planned, and the motion speed is calculated according to the following formula: ， in: For the print head from the first q -1 sampling point moved to the first q The velocity of movement at each sampling point; The preset base motion speed; For the first q The angle between the surface unit normal vector of the sampling point and the fixed reference unit normal vector, minus the angle between the sampling point and the fixed reference unit normal vector. q -1 is the included angle between the corresponding sampling points, and the absolute value of the difference obtained; For the first q The sampling point and the first q -1 The angle between the projected components of the surface unit normal vector at the sampling point on the plane perpendicular to the fixed reference unit normal vector; No. q The angle between the surface unit normal vector of each sampling point and the fixed reference unit normal vector; , , These are the preset weighting coefficients.

12. An additive manufacturing method for a multi-branched tubular carbon fiber part, characterized in that, Based on the three-dimensional model of the multi-branched tubular carbon fiber part, a continuous fiber printing path is generated using the path generation method as described in any one of claims 1 to 11. Based on the continuous fiber printing path, the printing equipment is controlled to perform continuous carbon fiber winding printing on a three-dimensional substrate surface that matches the shape of the multi-branched tubular carbon fiber part, so as to prepare the multi-branched tubular carbon fiber part.

13. The additive manufacturing method according to claim 12, characterized in that, Specifically as follows: Three-dimensional substrate preparation and installation steps: Prepare a three-dimensional substrate of water-soluble material, and install the three-dimensional substrate on the end of the main robotic arm through a fixed frame, so that one branch end face of the three-dimensional substrate corresponds to and is fixedly connected to the working end face of the flange at the end of the main robotic arm; Path planning and control command generation steps: Based on the three-dimensional model of the multi-branched tubular carbon fiber part, a continuous fiber printing path is generated using the path generation method as described in any one of claims 1 to 11. Based on the continuous fiber printing path and the model of the three-dimensional matrix, a coordinated motion command for controlling the main robotic arm and the slave robotic arm carrying the print head is generated. Winding and printing step: Control the main robotic arm and the slave robotic arm to move in coordination according to the coordinated motion command, and perform continuous winding and printing of carbon fibers on the three-dimensional substrate surface to form a composite material winding body; Post-processing and core removal steps: The wound body is post-processed and the three-dimensional matrix is ​​dissolved and removed to obtain the multi-branched tubular carbon fiber part.

14. The additive manufacturing method according to claim 13, characterized in that, The winding and printing step also includes synchronous control and force feedback adjustment: The host computer communicates with the two robotic arm controllers through a real-time data exchange protocol to achieve motion synchronization; The printing pressure is collected in real time by the force sensor at the end of the robotic arm; When the pressure is greater than the first threshold, the tool coordinate system of the robotic arm is controlled to shift by a preset step along its positive Z-axis. When the pressure is less than the second threshold, the tool coordinate system is controlled to shift the preset step size along the negative Z-axis. After the offset, the entire tool coordinate system of the robotic arm is updated, and the pose of all subsequent path points is recalculated based on this. After generating the updated motion command, printing is resumed to maintain the printing pressure within the preset range.

15. The additive manufacturing method according to claim 13, characterized in that, In the three-dimensional substrate preparation and installation steps, the water-soluble material is polyvinyl alcohol. The three-dimensional substrate and its edge extensions are integrally printed by fused deposition modeling and assembled into a rigid frame by connecting rods and flange fasteners. The three-dimensional model of the whole consisting of the three-dimensional substrate and the edge extensions is used as the path to generate the required three-dimensional model. And / or, in the post-processing and core removal steps, the post-processing includes removing the edge extension and the additive structure attached thereto. The process of dissolving and removing the three-dimensional matrix specifically involves immersing the entangled body in deionized water at 60°C to 90°C until the water-soluble material is completely dissolved. And / or, the master robotic arm and the slave robotic arm are UR series collaborative robots, and the collaborative motion commands are synchronously sent to the controllers of the two robotic arms through a real-time data exchange protocol.

Images