A path planning method in a continuous fiber composite material manufacturing process
By using a discrete orientation-guided fiber path planning method, the problems of path continuity and buckling performance in continuous fiber reinforced composite materials were solved, and the high-efficiency load-bearing capacity and improved stability of perforated plates under compressive loads were achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- UNIV OF ELECTRONICS SCI & TECH OF CHINA
- Filing Date
- 2026-06-03
- Publication Date
- 2026-07-31
AI Technical Summary
Existing fiber path optimization methods for continuous fiber reinforced composites suffer from problems such as insufficient path continuity, lack of multi-objective coordination, lack of anti-overlap planning, and inaccurate buckling performance evaluation, resulting in low buckling strength of perforated plates under compressive loads.
A discrete orientation-guided continuous fiber path planning method is adopted. By defining a discrete orientation field, constructing a multi-objective cost function and chain-like fiber path growth, and combining collision detection and buckling performance analysis, a continuous and optimized fiber path is generated to improve buckling bearing capacity.
The buckling capacity of the perforated plate was improved, frequent fiber interruptions and overlaps were avoided, ensuring manufacturing feasibility, and the stability and safety margin of the structure were enhanced through precise buckling analysis.
Smart Images
Figure CN122490849A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of composite material structural design and additive manufacturing technology, specifically relating to a fiber path planning method for 3D printing continuous fiber reinforced composite perforated plates. In particular, this invention provides a discrete orientation-guided continuous fiber layup path planning method to improve the buckling load-bearing capacity of perforated laminates under compressive loads. Background Technology
[0002] Continuous fiber reinforced composites are widely used in aerospace and other fields due to their high specific strength and specific stiffness. However, composite plates with holes (such as bolt holes) often experience buckling instability under compressive loads, as the presence of holes significantly weakens structural stiffness and load-bearing capacity. Traditional laminate designs typically employ fixed layup angles (such as 0° or 90° symmetrical layups), which cannot adequately alleviate stress concentration at the hole edges, resulting in lower buckling strength in perforated plates. In recent years, with the development of continuous fiber 3D printing and other variable stiffness composite manufacturing technologies, researchers have begun to explore the improvement of structural mechanical properties through variable angle layups or curved fiber paths. For example, studies using principal stress trajectories to guide fiber layup have found that compared to traditional straight layups, it can significantly reduce peak stress at the hole edges, improving structural strength and stiffness.
[0003] However, existing fiber path optimization methods still have many shortcomings: (1) Insufficient path continuity: Many optimizations only focus on local fiber orientation optimization and do not ensure the continuous printing of fiber paths globally, which may lead to frequent fiber interruptions, which is not conducive to manufacturing and structural performance; (2) Lack of multi-objective coordination: Existing methods often take a single index (such as maximizing stress or stiffness) as the objective and ignore manufacturing feasibility, such as excessive curvature leading to printing difficulties, excessive fiber concentration or sparseness leading to unbalanced performance, etc.; (3) Lack of anti-overlap planning: Fiber paths are prone to intersecting or overlapping in complex areas. Existing path planning lacks an efficient collision detection mechanism, which may cause printed fiber stacking and affect structural quality; (4) Inaccurate buckling performance evaluation: For plates with non-uniform fiber path distribution, traditional buckling calculation usually assumes that the material is isotropic or the layup is uniform, without considering local stiffness differences, thus failing to accurately predict the actual buckling load.
[0004] In summary, there is an urgent need for a novel continuous fiber path planning method that can comprehensively consider multiple factors, ensuring global path continuity, fiber orientation consistency, and curvature manufacturability, while also optimizing fiber distribution in the board to avoid overlap, thereby improving the buckling resistance of perforated boards. Simultaneously, an improved buckling analysis method is needed to adapt to performance prediction under non-uniform fiber distributions. Summary of the Invention
[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a discrete orientation-guided continuous fiber path planning method, which can be used to 3D print perforated composite laminates to improve their buckling resistance. The method of this invention balances fiber placement continuity and multi-objective optimization, maximizing the buckling resistance of the structure while ensuring manufacturing feasibility.
[0006] To achieve the above objectives, the present invention adopts the following technical solution: a path planning method in the preparation process of continuous fiber composite materials, comprising the following steps:
[0007] S1. Define the discrete orientation field;
[0008] Based on design requirements and structural load conditions, a discrete set of preferred fiber orientation angles is pre-determined. The ideal fiber orientation distribution in each region within the plate is obtained through finite element analysis or topology optimization methods; for example, the local optimal orientation is determined based on the principal stress direction field. Then, a discretization strategy is employed to map the continuous orientation field to an approximate field on the aforementioned finite discrete orientation set, ensuring that the entire design domain is dominated by several discrete fiber orientations; for example, discretized into finite element centroid orientations. This discrete orientation guidance strategy provides directional guidance for chain path growth, reduces the amount of known information required, and has higher generalization ability.
[0009] S2. Geometric processing and selection of the starting seed point;
[0010] Obtain the geometric model of the perforated laminate to be designed, and utilize adaptive The algorithm extracts complex boundary contours of structures from a discrete set of points to generate accurate boundary representations required for path planning; for irregular shapes containing holes, The method can dynamically adjust the α value according to given parameters, filter out minor uneven features, and extract smooth outer boundaries and hole contours, providing a simplified and accurate design domain for subsequent path planning. After obtaining the outer boundary and each hole contour, regular sampling is performed along the boundary to generate the starting seed point of the chain fiber path. Let any boundary curve be denoted as... Using arc length parameter Represent its spatial coordinates ,in The total arc length of the boundary; given the sampling step size along the boundary. Taken as the fiber width If the seed point sequence on the boundary is multiplied by a factor of 1, then it can be represented as:
[0011] ;
[0012] Represents the first boundary curve The coordinate vector or position vector of each arc-length sampling point This indicates that the arc length parameter Substitute boundary parameterization mapping The point obtained later; No. The arc length parameter values corresponding to each sampling point (the cumulative arc length along the boundary from the starting point of the boundary) are discretized with a fixed step size: ; through external boundaries and the boundaries of each hole The above arc length sampling is performed sequentially to obtain a set of starting seed points covering the overall boundary, which are used for the chain fiber path growth in the subsequent step S4. , This indicates the number of holes, specifically the number of inner boundaries (hole boundaries). Therefore... Used to traverse the boundary of each hole; The subscript for "hole" is used to distinguish the outer boundary. With the boundary of the hole ;exist middle, This indicates that this is a set of hole boundaries. It is the first Index of the hole boundary;
[0013] S3. Construct a multi-objective cost function;
[0014] Before path planning, define a multi-objective cost function. The multi-objective function includes the following five components: alignment with the discrete orientation field, fiber orientation penalty, fiber spacing repulsion, volume fraction control, and guiding potential energy. These are the weighting coefficients corresponding to each objective component; To measure the deviation between the candidate direction and the local direction field, and to ensure that the fiber as a whole conforms to the local optimal force direction; To construct the path based on the minimum distance from the candidate segment to the hole edge and plate edge, and to avoid the path from touching the boundary; To utilize the deviation between the local fiber volume fraction and the target volume fraction near the candidate segment, local over-stacking or insufficient fiber density is suppressed. To encourage fibers to run parallel to existing paths within a suitable spacing by using the attraction distance parameter, which is beneficial for forming a force transmission belt; The cost function is constructed from the angle between the candidate direction and the previous direction to limit the single-step angle and avoid excessive curvature; multi-objective cost function. The format is:
[0015] ;
[0016] Indicates candidate directions for fiber path growth;
[0017] S4, chain fiber path growth;
[0018] After obtaining the discrete fiber orientation field and multi-objective cost function, a chain-like fiber path growth strategy is adopted. Starting from the initial seed point, it gradually extends to form a continuous set of fiber paths covering the entire structure. Let a certain growing fiber path be at the nth... The endpoint of the step is Step size is The chosen growth direction is Then the next endpoint is updated using the chained update formula:
[0019] ;
[0020] This update formula is iterated repeatedly, starting from the initial seed point. Departure order determined This generates a polygonal fiber centerline that runs through the design domain; step size They are on the same order of magnitude as the fiber width;
[0021] A global reference direction is introduced at each growth step. With local direction The weighted mixture is used to construct the reference direction in this step. :
[0022] ;
[0023] in For global-local mixed weights; when When the value is large, the path follows the local optimal direction more closely, which is beneficial for adaptive bending according to local forces at the hole edge and stress concentration area; when When the path is smaller, it is closer to the global direction, which helps improve overall straightness and fiber utilization; thus, by adjusting in space... A balance is struck between "local stress guidance" and "global deployment strategy";
[0024] A set of candidate directions is generated discretly within a finite angular range, centered on the mixed reference direction.
[0025] For each candidate direction Using the multi-objective cost function from step S3 in the second... During the growth process, a geometric feasibility check is first performed on each candidate direction, and candidates that do not meet the geometric constraints are removed from the set, resulting in a feasible candidate set. Subsequently, a greedy selection is performed within the feasible set according to the cost function:
[0026] ;
[0027] Substitute this optimal direction into the chain update formula to generate new endpoints; if at some step If the path is empty, meaning all candidate directions are excluded due to exceeding the limit or overlap, then the path is considered to have reached a termination state and will no longer continue to grow. By iteratively executing the process of "candidate generation - geometric filtering - cost evaluation - optimal selection - chain update - termination judgment" on all seed points, a set of continuous fiber paths that do not illegally overlap, are continuously connected as a whole, macroscopically conform to the global direction, and locally conform to the stress distribution at the hole edge is finally obtained. Under the guidance of the multi-objective cost function, this chain fiber path growth mechanism realizes automatic optimization and continuous coverage in complex opening structures.
[0028] S5. Collision detection and correction;
[0029] During path growth, anti-overlap collision detection is performed in real time. For each new fiber segment generated, the separating axis theorem algorithm is used to detect its geometric collision with existing fiber paths. The design domain is divided into a uniform spatial grid, and a Python dictionary is used to record the already laid fiber segments. Collision testing is only performed on fiber segments within adjacent grids to avoid global pairwise checks and reduce computational complexity. Once a new segment is detected to be less than the safe distance from an existing path or to intersect with it, the algorithm will adjust the direction of the new segment: either by increasing the overlap penalty weight in the cost function to reselect a suboptimal direction, or by shortening the step size and retrying, to ensure that each path generated in the end is collision-free and meets the minimum fiber spacing requirement.
[0030] S6. Buckling performance analysis;
[0031] After fiber path planning is completed, the buckling performance of the structure is evaluated based on the obtained non-uniform fiber distribution, and the critical buckling load is calculated. and normalized form .
[0032] Furthermore, the sampling step size in step S2 For fiber width times.
[0033] Furthermore, the specific method for discretizing and generating a set of candidate directions within a finite angular range, centered on the mixed reference direction, in step S4 is as follows: Centered on the angle step size exist Take a set of discrete angles within the range :
[0034] ;
[0035] This represents the upper limit of the number of discrete candidate directions in the current reference direction. Take from both sides A discrete angle; for each candidate angle The corresponding candidate endpoints are obtained by updating according to the above formula. And candidate fiber segments; subsequent cost evaluation and geometric constraint checks are performed on this finite candidate set, avoiding a global search in a continuous orientation space.
[0036] Furthermore, in step S6, the critical buckling load The specific calculation method is as follows:
[0037] The plate structure is divided into several shell elements according to the mapped quadrilateral grid, based on the fiber laying direction at the centroid of each element. And calculate the fiber volume fraction of the unit by considering the proportion of the area covered by fibers. Then utilize based on The mixing rule: longitudinal modulus along the fiber direction Compared with the main Poisson The transverse modulus is calculated using the classical hybridization principle. In-plane shear modulus Then adopt the introduction ROM model of correction factor; resin shear modulus Depend on Give, This represents the elastic modulus of the matrix material. Indicates the Poisson's ratio of the matrix material;
[0038] After obtaining the equivalent orthogonal anisotropic engineering constants in the principal direction coordinate system of the element's local material, the element stiffness matrix is constructed through coordinate transformation. The linear elastic stiffness matrix of the overall structure is obtained by integrating along the thickness direction according to the laminated plate structure. Under given boundary conditions and pre-compression conditions, the corresponding geometric stiffness matrix is further formed. Thus, a generalized eigenvalue buckling is established;
[0039] ;
[0040] in, Let be the buckling load factor, and let the minimum eigenvalue be denoted as . , This represents the displacement (displacement control parameter) of the symmetrical edge displacement preload applied to the plate boundary under pre-buckling conditions. The vertical compressive reaction force under pre-buckling conditions is obtained by applying the symmetrical edge displacement preload to the vertical edge of the plate. The in-plane compressive resultant force is calculated using the following formula. ;
[0041] ;
[0042] in The length of the board side This represents the resultant compressive force within the critical plane (critical edge compressive reaction force), which is the total edge compressive force when the structure reaches critical buckling.
[0043] Furthermore, in step S6, the critical buckling load Normalized form The calculation method is as follows:
[0044] The Wu-style normalization method is used to perform dimensionless processing on the critical load; firstly, a model based on the longitudinal modulus is constructed. , plate thickness and board side length Dimensionless buckling load :
[0045] ;
[0046] Then, a critical load is introduced with reference to a quasi-isotropic reference laminate. Its parsing expression is:
[0047] ;
[0048] in, and Laminate invariants , Given:
[0049] ;
[0050] The boundary condition correlation coefficient is taken as the free horizontal side. Constrain the horizontal edge ,in The in-plane principal effect parameter is the equivalent in-plane tensile stiffness, which determines... The order of magnitude, and This represents the in-plane lateral coupling parameter, a coupling term related to the Poisson effect. It does not directly give the magnitude of the modulus, but rather through its relationship with... The ratio determines the equivalent Poisson's ratio. The Wu-style normalized critical buckling load parameter is defined. :
[0051] ;
[0052] Through the above process, the influence of local fiber orientation and volume fraction on plate stiffness was accurately considered at the finite element level, and the calculated critical buckling load was uniformly mapped to a Wu-style normalized index that can be directly compared with a quasi-isotropic reference plate. This allows for an objective evaluation of the buckling performance of different fiber path schemes.
[0053] Beneficial Effects: Compared with existing technologies, this invention has significant advantages and positive effects: First, the chain-growth algorithm ensures global continuity of the fiber path, avoiding the adverse effects of frequent fiber shearing on mechanical properties in traditional printing. Second, the multi-objective cost function coordinates the balance between mechanical properties and manufacturing constraints—while improving structural load-bearing capacity, it ensures that the fiber turning radius and laying spacing meet process requirements, avoiding manufacturing defects such as excessively small curvature or overlapping laying. Third, the introduced fiber volume fraction control makes the fiber distribution in the structure more uniform, fully utilizing the load-bearing capacity of each region and avoiding material waste and weak performance areas. In addition, the buckling analysis method based on non-uniform constitutive model can accurately predict the buckling load of the structure, providing a reliable basis for design optimization. Simulation results based on the Abaqus platform show that, compared with the traditional straight-layout reference plate, the fiber path scheme planned by this invention significantly improves the critical buckling load, and enhances the overall stability and safety margin of the structure. Attached Figure Description
[0054] Figure 1 Flowchart for discrete-guided continuous fiber path planning.
[0055] Figure 2 Based on adaptive The workflow for boundary edge extraction and loop construction.
[0056] Figure 3 It is the final manufacturable fiber path generated by the fiber path laying algorithm for different aperture ratios.
[0057] Figure 4 This is a diagram showing the influence of the global directional intervention coefficient on the normalized critical buckling load.
[0058] Figure 5 It is the effect of fiber orientation weight on the normalized critical buckling load.
[0059] Figure 6 It is the effect of fiber attraction weight on the normalized critical buckling load.
[0060] Figure 7 It is the effect of fiber curvature penalty weight on normalized critical buckling load.
[0061] Figure 8 It is the effect of the fiber distance coefficient on the normalized critical buckling load. Detailed Implementation
[0062] The general flow of the specific implementation scheme of the technical solution of the present invention is as follows: Figure 1 This includes the following steps S1 to S4:
[0063] S1: Read the plate geometry and finite element mesh to obtain the discrete fiber orientation field;
[0064] S2: Utilizing Adaptive Technology Extract the outer boundary of the plate and the boundaries of each hole;
[0065] S3: Chain fiber path growth is performed based on a multi-objective cost function under the guidance of a discrete direction field;
[0066] S4: Perform topology merging and connectivity optimization on multiple chain paths.
[0067] The specific implementation methods of each step are explained below, with a focus on steps S2 and S3.
[0068] In step S1, the geometric shape of the target plate (including the outer contour and hole contour) and the corresponding finite element mesh are first input. The finite element mesh can be a two-dimensional element with arbitrary topology (such as a triangle or quadrilateral), and the centroid coordinates of the element are denoted as... , This represents the total number of two-dimensional elements in the finite element mesh.
[0069] By using topology optimization, strength optimization, or other existing methods, a discrete fiber reference direction angle is given at the centroid of each element. This forms a discrete fiber orientation field. For the location required during the path growth process (which may not necessarily fall exactly at the element centroid), the local reference orientation can be calculated using methods such as neighborhood weighted averaging and nearest-neighbor interpolation. .
[0070] S2 is based on adaptive The overall process of boundary extraction and boundary sampling in the algorithm is as follows: Figure 2 This invention employs adaptive... The outer boundary and hole boundary are automatically extracted from the centroid set of the finite element method to adapt to local mesh density changes and avoid the traditional "global" approach. "The method requires manual parameter adjustment. First, all finite element centroids are assembled into a point set." Constructing a Delaunay triangulation on this point set yields a set of triangles:
[0071] ;
[0072] For any triangle Let the lengths of the three sides be:
[0073] ;
[0074] The area of the triangle is Its circumcircle radius It can be calculated using the following formula:
[0075] ;
[0076] Indicates three points The triangle formed The area (strictly positive) is used to characterize the local mesh density at each point. Define a local length scale. Preferably, it can be achieved through its The median distance to the nearest neighbors is given:
[0077] ;
[0078] in Point of nearest neighbor set This is a preset integer used to balance local stability and sensitivity.
[0079] Set a dimensionless magnification factor For each triangle Define its adaptive threshold :
[0080] ;
[0081] Constructing Adaptive Triangle Sets :
[0082] ;
[0083] Represents a triangle The radius of the circumcircle, Represents a set of points After constructing the Delaunay triangulation, the set of triangles obtained is used to retain all triangles. We count the three edges to obtain the edge set. For any undirected edge Statistics on its Number of times it appears Edges that appear only once are considered boundary edges, forming a set of boundary edges. :
[0084] ;
[0085] Represents an undirected edge In preserving the triangle set The number of times the boundary edges appear as sides of a triangle is considered, treating the boundary edges as an undirected graph. The edge set, in which This is the set of indices of the points appearing on the boundary edge. For each vertex... Construct an adjacency list :
[0086] ;
[0087] From any unvisited edge Departure, order Iterative selection:
[0088] ;
[0089] Equation (21) means: in an undirected graph with boundary, from the current vertex The set of adjacent points Select the next vertex and exclude the previous vertex. This is to avoid backtracking and allow for continuous advancement along the boundary. Updates will follow. This continues until the starting point is reached, resulting in an ordered sequence of vertices. Defined closed polygonal circuit Repeat the above process until all boundary edges have been visited, resulting in several closed loops. These correspond to the outer boundary and the boundaries of each hole, respectively.
[0090] For any closed loop:
[0091] ;
[0092] Its symbolic area for:
[0093] ;
[0094] All loops are classified by their sign area. The loops are sorted from largest to smallest, and the loop with the largest area is designated as the candidate outer contour. For other loops, the odd-even rule is used to determine whether they are contained within the outer contour; those containing the outer contour are marked as hole contours. In this way, polygonal descriptions of the outer boundary of the plate and all hole boundaries can be automatically obtained based on the finite element centroid.
[0095] In step S3, the present invention applies a discrete fiber direction field. Under the guidance of [the relevant authority], a single continuous fiber is discretized into a series of interconnected line segments, and the optimal expansion direction is selected at each step using a multi-objective cost function to achieve automatic growth of the chain-like path. The specific form of the multi-objective cost function is explained below:
[0096] For each valid candidate direction Construct the following multi-objective cost function:
[0097] ;
[0098] The specific physical meaning of each item has been explained in detail above, and will not be repeated here. The detailed construction form of each part will be explained below.
[0099] S3.1 Cost of Directional Deviation :
[0100] The orientation deviation cost is used to penalize the degree of deviation between the candidate orientation and the reference orientation field. Let the difference between the candidate orientation and the local reference orientation angle be:
[0101] ;
[0102] Indicates the first One candidate growth direction angle, This represents the local reference orientation angle calculated from the reference orientation field (which is the orientation angle after global-local fusion) at the current path endpoint, serving as the benchmark for candidate orientation generation and orientation deviation evaluation. Definition: When the candidate direction is exactly the same as the reference direction. The cost is zero; as the deviation angle increases, the cost is approximated as... The number of ways to increase this allows for constraints on directional consistency.
[0103] S3.2 Repulsion Cost :
[0104] Repulsion cost is used to prevent new fiber segments from getting too close to existing fibers, thus preventing excessive local overlap or overly thick regions. Let the current candidate segment be from... arrive Let the set of existing fiber segments in the neighborhood of a rectangular fiber segment be denoted as . For each existing fiber segment Calculate the minimum distance from the candidate segment to the current segment. .
[0105] Set effective fiber width and safety clearance coefficient Define the minimum safety gap threshold. :
[0106] ;
[0107] Indicates the effective fiber width (the width occupied by the equivalent fiber). For all satisfying For neighboring segments, introduce a rejection penalty:
[0108] ;
[0109] in To prevent the use of tiny constants with a denominator of zero, this cost increases dramatically when candidate segments are too close to existing fibers, thus automatically rejecting these directions; if If the neighboring segment does not contribute to the repulsion cost, then the neighboring segment does not contribute to the repulsion cost.
[0110] S3.3 Attraction Cost
[0111] Attraction cost is used to avoid excessively large gaps in local areas, encouraging new fibers to move closer to existing fibers and improving coverage uniformity. The minimum distance from a candidate segment to all existing fiber segments in its neighborhood is denoted as:
[0112] ;
[0113] And ignore the extremely close-range cases that have already been handled by the exclusion cost, only consider In this case, a feature length is also set. The cost of attraction is defined as:
[0114] ;
[0115] When there are already fibers nearby in the vicinity of the candidate segment Smaller The value is relatively low; when the candidate fragment is in a local blank area, The larger the exponent, the more it increases dramatically, imposing an additional penalty on directions far from existing fibers, thus avoiding leaving large unfilled areas.
[0116] S3.4 Curvature Cost :
[0117] Curvature cost is used to limit the local turning angle of the path, ensuring a smooth fiber path and meeting the minimum allowable bending radius for continuous fibers. Let the actual fiber direction of the previous segment be... The candidate directions calculated in this study are: , Indicates the first Step endpoint At a given location, after filtering by geometric and technological constraints, the set of feasible candidate orientation angles is defined as the rotation angle. :
[0118] ;
[0119] and reduce it to an interval Minimum absolute difference within; curvature cost for:
[0120] ;
[0121] When the path extends approximately as a straight line locally The curvature cost is almost zero; however, the cost increases rapidly when sharp turns are required.
[0122] S3.5 Local Volume Fraction Cost ;
[0123] In order to control the local fiber volume fraction (area fraction) and avoid excessive local sparseness or over-packing, the present invention defines a local circular window near the candidate leading point and estimates the local volume fraction based on the length of existing fibers passing through the window.
[0124] Starting from the current segment Centered on, radius Let be the window radius and be the window area. .
[0125] Let the existing set of fiber segments be For each segment Calculate the length of the line segment where it intersects the window. Using nominal fiber width The approximate area covered by the fragment in the window is Local fiber volume fraction for:
[0126] ;
[0127] Set target local volume fraction To introduce an asymmetric quadratic penalty to apply a stronger penalty in fiber-deficient regions, we define:
[0128] ;
[0129] in This represents the cost of local fiber volume fraction deviation. This represents the target local fiber volume fraction, i.e., the fiber content benchmark value that is desired to be approximated in a local area, used to guide coverage uniformity. This represents the penalty relaxation factor, used to reduce the penalty intensity of this branch when the local volume fraction is slightly higher than the target value. Both are preset coefficients used to impose a strong penalty when the local volume fraction is insufficient, and to appropriately relax the constraints when it slightly exceeds the target value, so that the algorithm can prioritize filling sparsely distributed areas and improve the overall uniformity of fiber distribution.
[0130] For each valid candidate direction ,calculate To obtain the total cost Among all candidates that remain within the sheet geometry and do not violate basic process constraints, the direction with the minimum total cost is selected:
[0131] ;
[0132] And update the path accordingly:
[0133] ;
[0134] If all candidate directions are deemed invalid by overlap detection or process constraints, the current chain path is terminated and the entire path is deleted; if it is due to geometric out-of-bounds, the current path is terminated but the already generated path is retained. Repeating the above chain growth process for all boundary seed points yields multiple continuous fiber bundle paths covering the plate area.
[0135] Additional note: Before calculating the cost function, this invention preferably performs collision detection on the candidate rectangular fiber segments; if an unacceptable entity overlap occurs with an existing fiber segment (e.g., determined by rectangular-rectangular collision detection based on the separating axis theorem), the candidate direction is directly determined to be invalid and eliminated, and its cost is no longer calculated.
[0136] In step S4, the multiple chain fiber paths obtained in step S3 are subjected to topology analysis. For example, based on the distance and direction between the path endpoints, path endpoints that are close together and have matching directions are connected to form longer macroscopic paths, and very short isolated paths are deleted or merged to reduce the number of 3D printer starts and stops and improve continuity.
[0137] In a preferred embodiment, paths can be further merged by detecting bridging opportunities between paths (such as low-cost connections between two path endpoints). The aforementioned topology merging can be implemented using heuristic rules. This invention uses an adjacency list to determine the distance relationship between endpoints to achieve proximity connections, while setting a margin. If the distance exceeds the margin, the endpoints are considered too far apart, and no macrofiber connection is made.
[0138] Based on the above fiber laying algorithm, the fiber path results are as follows: Figure 3 Performance analysis under Abaqus, such as Figures 4-8 .
Claims
1. A path planning method in the preparation process of continuous fiber composite materials, comprising the following steps: S1. Define the discrete orientation field; Based on design requirements and structural load conditions, a set of discrete fiber orientation angles is pre-determined; the ideal fiber orientation distribution in each region of the plate is obtained through finite element analysis or topology optimization; then, a discretization strategy is adopted to map the continuous orientation field to an approximate field on the above finite discrete orientation set, so that the entire design domain is dominated by several discrete fiber orientations. S2. Geometric processing and selection of the starting seed point; Obtain the geometric model of the perforated laminate to be designed, and utilize adaptive The algorithm extracts the complex boundary contours of the structure from a discrete set of points to generate the accurate boundary representation required for path planning. After obtaining the outer boundary and the contours of each hole, it performs regular sampling along the boundary to generate the starting seed points for the chain fiber path. Let any boundary curve be denoted as... Using arc length parameter Represent its spatial coordinates ,in The total arc length of the boundary; given the sampling step size along the boundary. Then the seed point sequence on this boundary is represented as: ； Represents the first boundary curve The coordinate vector or position vector of each arc-length sampling point This indicates that the arc length parameter Substitute boundary parameterization mapping The point obtained later; No. The arc length parameter values corresponding to each sampling point are discretized at a fixed step size: ; through external boundaries and the boundaries of each hole The above arc length sampling is performed sequentially to obtain a set of starting seed points covering the overall boundary, which are used for the chain fiber path growth in the subsequent step S4. , This indicates the number of holes, i.e., the number of inner boundary lines; therefore Used to traverse the boundary of each hole; The subscript for "hole" is used to distinguish the outer boundary. With the boundary of the hole ;exist middle, This indicates that this is a set of hole boundaries. It is the first Index of the hole boundary; S3. Construct a multi-objective cost function; Before path planning, define a multi-objective cost function. The multi-objective function includes the following five components: alignment with the discrete orientation field, fiber orientation penalty, fiber spacing repulsion, volume fraction control, and guiding potential energy. These are the weighting coefficients corresponding to each objective component; To measure the deviation between the candidate direction and the local direction field, and to ensure that the fiber as a whole conforms to the local optimal force direction; To construct the path based on the minimum distance from the candidate segment to the hole edge and plate edge, and to avoid the path from touching the boundary; To utilize the deviation between the local fiber volume fraction and the target volume fraction near the candidate segment, local over-stacking or insufficient fiber density is suppressed. To encourage fibers to run parallel to existing paths within a suitable spacing by using the attraction distance parameter, which is beneficial for forming a force transmission belt; The cost function is constructed from the angle between the candidate direction and the previous direction to limit the single-step angle and avoid excessive curvature; multi-objective cost function. The format is: ； Indicates candidate directions for fiber path growth; S4, chain fiber path growth; After obtaining the discrete fiber orientation field and multi-objective cost function, a chain-like fiber path growth strategy is adopted. Starting from the initial seed point, it gradually extends to form a continuous set of fiber paths covering the entire structure. Let a certain growing fiber path be at the nth... The endpoint of the step is Step size is The chosen growth direction is Then the next endpoint is updated using the chained update formula: ； This update formula is iterated repeatedly, starting from the initial seed point. Departure order determined This will generate a polygonal fiber centerline that runs through the design domain; Step length They are on the same order of magnitude as the fiber width; A global reference direction is introduced at each growth step. With local direction The weighted mixture is used to construct the reference direction in this step. : ； in For global-local mixed weights; when When the value is large, the path follows the local optimal direction more closely, which is beneficial for adaptive bending according to local forces at the hole edge and stress concentration area; when When the path is smaller, it is closer to the global direction, which helps improve overall straightness and fiber utilization; thus, by adjusting in space... A balance is struck between "local stress guidance" and "global deployment strategy"; A set of candidate directions is generated discretly within a finite angular range, centered on the mixed reference direction. For each candidate direction Using the multi-objective cost function from step S3 in the second... During the growth process, a geometric feasibility check is first performed on each candidate direction, and candidates that do not meet the geometric constraints are removed from the set, resulting in a feasible candidate set. ; Subsequently, a greedy selection is performed within the feasible set according to the cost function: ； Then, this optimal direction is substituted into the chain update formula to generate new endpoints; if If the path is empty, meaning all candidate directions are excluded due to exceeding the limit or overlap, then the path is considered to have reached a termination state and will no longer grow. By iteratively executing the process of "candidate generation - geometric filtering - cost evaluation - optimal selection - chain update - termination judgment" on all seed points, a set of continuous fiber paths that do not illegally overlap, are continuously connected as a whole, macroscopically conform to the global direction, and locally conform to the stress distribution at the hole edge is finally obtained. Under the guidance of the multi-objective cost function, this chain fiber path growth mechanism realizes automatic optimization and continuous coverage in complex opening structures. S5. Collision detection and correction; During path growth, anti-overlap collision detection is performed in real time. For each new fiber segment generated, the separating axis theorem algorithm is used to detect its geometric collision with existing fiber paths. The design domain is divided into a uniform spatial grid, and a Python dictionary is used to record the already laid fiber segments. Collision testing is only performed on fiber segments within adjacent grids to avoid global pairwise checks and reduce computational complexity. Once a new segment is detected to be less than the safe distance from an existing path or to intersect with it, the algorithm will adjust the direction of the new segment: either by increasing the overlap penalty weight in the cost function to reselect a suboptimal direction, or by shortening the step size and retrying, to ensure that each path generated in the end is collision-free and meets the minimum fiber spacing requirement. S6. Buckling performance analysis; After fiber path planning is completed, the buckling performance of the structure is evaluated based on the obtained non-uniform fiber distribution, and the critical buckling load is calculated. and normalized form .
2. The path planning method in the preparation process of a continuous fiber composite material as described in claim 1, characterized in that, The sampling step size in step S2 For fiber width times.
3. The path planning method in the preparation process of a continuous fiber composite material as described in claim 1, characterized in that, The specific method for discretizing and generating a set of candidate directions within a finite angular range, centered on the mixed reference direction, is as follows: Centered on the angle step size exist Take a set of discrete angles within the range : ； This represents the upper limit of the number of discrete candidate directions in the current reference direction. Take from both sides A discrete angle; for each candidate angle The corresponding candidate endpoints are obtained by updating according to the above formula. And candidate fiber segments; subsequent cost evaluation and geometric constraint checks are performed on this finite candidate set, avoiding a global search in a continuous orientation space.
4. The path planning method in the preparation process of a continuous fiber composite material as described in claim 1, characterized in that, Critical buckling load in step S6 The specific calculation method is as follows: The plate structure is divided into several shell elements according to the mapped quadrilateral grid, based on the fiber laying direction at the centroid of each element. And calculate the fiber volume fraction of the unit by considering the proportion of the area covered by fibers. Then utilize based on The mixing rule: longitudinal modulus along the fiber direction Compared with the main Poisson The transverse modulus is calculated using the classical hybridization principle. In-plane shear modulus Then adopt the introduction ROM model of correction factor; resin shear modulus Depend on Give, This represents the elastic modulus of the matrix material. Indicates the Poisson's ratio of the matrix material; After obtaining the equivalent orthogonal anisotropic engineering constants in the principal direction coordinate system of the element's local material, the element stiffness matrix is constructed through coordinate transformation. The linear elastic stiffness matrix of the overall structure is obtained by integrating along the thickness direction according to the laminated plate structure. Under given boundary conditions and pre-compression conditions, the corresponding geometric stiffness matrix is further formed. Thus, a generalized eigenvalue buckling is established; ； in, Let be the buckling load factor, and let the minimum eigenvalue be denoted as . , This represents the displacement of the symmetrical edge displacement preload applied to the plate boundary under pre-buckling conditions. The vertical compressive reaction force under pre-buckling conditions is obtained by applying the symmetrical edge displacement preload to the vertical edge of the plate. The in-plane compressive resultant force is calculated using the following formula. ; ； in The length of the board side This represents the resultant compressive force within the critical plane, which is the total edge compressive force when the structure reaches critical buckling.
5. The path planning method in the preparation process of a continuous fiber composite material as described in claim 1, characterized in that, Critical buckling load in step S6 Normalized form The calculation method is as follows: The Wu-style normalization method is used to perform dimensionless processing on the critical load; firstly, a model based on the longitudinal modulus is constructed. , plate thickness and board side length Dimensionless buckling load : ； Then, a critical load is introduced with reference to a quasi-isotropic reference laminate. Its parsing expression is: ； in, and Laminate invariants , Given: ； The boundary condition correlation coefficient is taken as the free horizontal side. Constrain the horizontal edge ,in The in-plane principal effect parameter is the equivalent in-plane tensile stiffness, which determines... The order of magnitude, and This represents the in-plane lateral coupling parameter, a coupling term related to the Poisson effect. It does not directly give the magnitude of the modulus, but rather through its relationship with... The ratio determines the equivalent Poisson's ratio; the Wu-style normalized critical buckling load parameter is defined. : ； Through the above process, the influence of local fiber orientation and volume fraction on plate stiffness was accurately considered at the finite element level, and the calculated critical buckling load was uniformly mapped to a Wu-style normalized index that can be directly compared with a quasi-isotropic reference plate. This allows for an objective evaluation of the buckling performance of different fiber path schemes.