Additive manufacturing path generation method based on design intent analysis and non-grid expression

Abstract

The present application relates to the technical field of additive manufacturing, and discloses a method for generating an additive manufacturing path based on design intention analysis and non-grid expression, comprising the following steps: collecting a user input design intention, analyzing and processing the design intention to obtain non-grid expression data of a target object; constructing a manufacturing path generation domain based on the non-grid expression data, the construction of the manufacturing path generation domain being free of slice processing of triangular facet grids throughout; generating a manufacturing path for material deposition within the manufacturing path generation domain in combination with preset additive manufacturing process constraints; outputting control instructions of an additive manufacturing device according to the generated manufacturing path, while performing printability review on the control instructions, and outputting the control instructions to the additive manufacturing device after the review is passed. The present application eliminates the intermediate link of triangular facet grid construction and general slice processing, directly generates an executable manufacturing path based on a design intention, and avoids interference of grid topology problems on a manufacturing process.

Description

Technical Field

[0001] This invention relates to the field of additive manufacturing technology, specifically to an additive manufacturing path generation method based on design intent parsing and non-mesh representation. Background Technology

[0002] As additive manufacturing technology continues to be applied in fields such as industrial customization, cultural and creative exhibitions, and large-scale component manufacturing, the industry's requirements for the efficiency of the manufacturing chain, the controllability of the path, and the reliability of printing are constantly increasing. Current mainstream additive manufacturing path generation technologies generally use triangular mesh models as the core intermediate carrier from design intent to manufacturing execution. General-purpose slicing software processes the 3D mesh model into layers, ultimately generating G-code instructions that can drive the printing equipment. The triangular mesh model is a 3D representation of the target object's surface morphology, achieved by stitching together a large number of discrete triangular patches. General-purpose slicing software completes the conversion from the 3D model to the 2D printing path through planar sectioning, contour extraction, and path filling operations.

[0003] However, this path generation mode, which relies on triangular meshes as a necessary intermediate representation, is a core issue restricting the efficiency and flexibility of the additive manufacturing process. On the one hand, the mesh model itself is prone to inherent defects such as topological damage, non-manifold structures, and thin-walled feature failures, requiring significant manual repair before slicing can be completed. This directly leads to long pre-manufacturing preparation cycles and high operational barriers, making it difficult for non-professional users to independently complete the entire process. On the other hand, the complete manufacturing process requires switching between modeling software, model repair tools, slicing software, and printing control systems. The link from design intent to physical output is fragmented by multiple independent stages, making it impossible to achieve a seamless flow from design input to manufacturing execution. Furthermore, the path generation logic of general slicing algorithms is fixed and uniform, unable to flexibly adapt to customized process strategies and path design requirements such as continuous spirals and variable-layer high textures, making it difficult to meet differentiated structural optimization and path aesthetic design requirements. Therefore, the following solutions are proposed to address these issues. Summary of the Invention

[0004] Traditional additive manufacturing path generation methods rely on triangular meshes as an essential intermediate carrier, which has three inherent drawbacks: the mesh model is prone to topological damage and non-manifold structural problems, requiring a large amount of manual repair, resulting in long preparation cycles and high operational barriers; the complete process requires multiple switching between modeling, repair, slicing, and control systems, and the link from design intent to the physical object is fragmented by multiple independent links, resulting in low flow efficiency and high operating costs; the path generation logic of general slicing algorithms is fixed and uniform, which cannot flexibly adapt to customized process strategies such as continuous spirals and variable layer heights, making it difficult to meet differentiated design and optimization needs; To address the aforementioned technical problems, this invention provides an additive manufacturing path generation method based on design intent parsing and non-mesh representation, comprising the following steps: Step S1: Collect the design intent input by the user, parse and process the design intent to obtain the non-mesh representation data of the target object; the design intent information includes at least one of the following: the geometric dimensions, contour shape, and process requirements of the target object; Step S2: Construct a manufacturing path generation domain based on non-mesh representation data. The manufacturing path generation domain is different from the traditional slicing domain. In essence, it is a computational space that carries all the geometric information required for path generation. Each coordinate point of the nozzle trajectory can be directly solved by the function or field value in this domain without the need for intermediate mesh discretization steps. The construction of the manufacturing path generation domain does not depend on the slicing process of the triangular facet mesh. Step S3: Based on the preset additive manufacturing process constraints, generate a manufacturing path for material deposition within the manufacturing path generation domain; Step S4: Output control instructions for the additive manufacturing equipment according to the generated manufacturing path, and perform printability verification on the control instructions. After the verification is passed, output the control instructions to the additive manufacturing equipment.

[0005] Preferably, step S1 employs a semantic parameter parsing approach to process the design intent, corresponding to natural language, speech, or structured parameter class inputs, and includes the following steps: Step S11: Collect user input in natural language text, voice commands, or structured design parameters; Step S12: Use an artificial intelligence model to perform semantic recognition and parameter mapping on the input content, extract the geometric dimensions, texture type, and process parameters of the target object, and generate a design parameter vector; The AI ​​model employs an encoder-decoder network based on the Transformer architecture. The input layer receives a sequence of natural language text after word segmentation and encoding, or directly receives key-value pair embeddings of structured design parameters. The output layer generates a fixed-dimensional design parameter vector, which contains geometric dimension parameters, texture type encoding, and process parameters in a predetermined order. Parameter mapping rules are implemented through a pre-defined JSON dictionary, which maps the indexes of each dimension of the output vector to the names of geometric variables and process constraint parameters required in subsequent steps. During parsing, values ​​are retrieved by index and assigned to the corresponding radius-height function variables and process constraints. During training, manually labeled historical input-output samples are used as supervisory data. The label files are stored in JSON format, and the label structure is completely consistent with the dimensions of the output vector. The mean squared error loss function is used for training, and the Adam optimizer is selected. After training until the validation set loss converges, the model parameters are fixed for inference. Step S13: Output the non-mesh representation data, with the design parameter vector as the target object, to the downstream path generation stage; The semantic parameter parsing route also includes a model closed-loop optimization step: collecting the actual printing results of the additive manufacturing equipment, recording the sample data of successful and failed printing, constructing new training sample pairs by combining the design intent, non-mesh expression data and printing result labels of the corresponding samples, and iteratively optimizing the parameter mapping accuracy of the artificial intelligence model by adding new training samples.

[0006] Preferably, step S1 uses hand-drawn outline analysis to process the design intent, corresponding to the user's hand-drawn two-dimensional outline curve input, and includes the following steps: Step S14: Collect the two-dimensional side view contour point array input by the user through touch interaction; Step S15: Perform smoothing, noise reduction, boundary compliance verification, and geometric normalization on the hand-drawn outline points to complete the outline cleanup; Step S16: Establish a cylindrical coordinate system with the central rotation axis as the z-axis, and convert the cleaned contour into normalized contour data in the form of a radius-height function. The radius-height function is used to characterize the variation of the radius of the outer contour of the body of revolution with height. It can directly convert the 2D hand-drawn contour into a mathematical expression that can be used for path calculation without constructing a triangular mesh model. The formula is: In the formula, r is the radius of the outer contour of the cross section of the target rotating body at height z, and z is the height coordinate along the central rotation axis; Step S17: Using the normalized contour data as the non-mesh representation data of the target object, output it to the downstream path generation stage.

[0007] Preferably, step S2 includes the following steps: Step S21: Match the corresponding construction logic according to the type of non-mesh representation data. Non-mesh representation data includes at least one of parametric surfaces, implicit functions, distance fields, curve networks, design parameter vectors, and normalized contour data. The parametric surface defines the surface shape of the target object using parametric surface equations. Two independent parameters in the parameter domain establish a mapping relationship with the three-dimensional spatial coordinates. The entire parameter domain is directly used as the path generation domain on the two-dimensional manifold. During path calculation, parameters are sampled in the parameter domain according to the preset path spacing and direction to obtain the parameter coordinate sequence, which is then converted into nozzle motion trajectory points in three-dimensional space through surface mapping. The implicit function expresses the geometric surface of the target object using implicit equations, and defines the entity region and the external region by the positive and negative values ​​of the function values. The path generation domain is constructed by sampling the implicit function within the bounding box space of the target object, constructing a directed distance field as an intermediate expression, and then generating a series of equidistant offset contours by extracting isosurfaces. During path calculation, the intersection of the implicit equation and the horizontal plane is solved at each height layer to obtain a closed contour curve, and the sequence of points on the curve is directly used as the nozzle motion coordinates. The distance field expresses the symbolic distance from the surface of the target object to any point in space using distance values ​​at discrete spatial points. The domain of the distance field directly constitutes the path generation domain. During path calculation, the gradient information of the distance field is used to determine the surface normal, and a continuous motion trajectory is constructed by tracing geodesics or equidistant offset surfaces on the zero isosurface. The curve network is used to define the skeleton or contour features of the target object with a set of structured feature curves. The path generation domain uses these feature curves as guides to generate a path space covering the target surface by sweeping, lofting or constructing tubular neighborhoods. During path calculation, equal arc length or equal parameter discrete sampling is performed along the feature curve direction, and a continuous path network is generated by combining the topological connection relationship between the curves. The design parameter vector is matched with a preset geometric template according to the type identifier of the parameter vector. The corresponding parameterized surface or contour function is generated by parameter instantiation, and then the path generation domain is constructed according to the corresponding surface or contour logic. When calculating the path, the path generation template bound to the geometric template is called, and the values ​​in the design parameter vector are substituted into the template equation to directly calculate the nozzle motion coordinates. The standardized contour data expresses the radius variation law of the target object along a certain coordinate axis using a contour function or a cross-sectional point list. The contour is rotated around the central axis to generate a body of revolution, and the internal region of the body of revolution constitutes the path generation domain. During path calculation, path parameterization is performed within the domain of the contour function to directly calculate the spatial motion trajectory of the nozzle and map the two-dimensional contour into a three-dimensional path. Step S22: Construct a manufacturing path generation domain in the form of a path field, orientation field, or layer function space based on non-mesh representation data. The manufacturing path generation domain is used to directly calculate the nozzle motion trajectory without first generating a triangular 3D mesh and then performing slicing operations.

[0008] Preferably, step S3 includes a path generation step: Step S31: Match the path generation strategy according to the target object type and process requirements. The generated manufacturing path includes at least one of the following: continuous spiral path, layered contour path, non-planar path, multi-axis linkage path, and variable layer height path. Step S32: The proprietary process path compiler directly calculates the motion coordinate sequence of the nozzle based on the manufacturing path generation domain to complete the generation of the manufacturing path; For rotating objects, a continuous spiral path generation strategy is adopted. The nozzle's motion coordinates are calculated using cylindrical coordinate parametric equations, enabling continuous extrusion motion without layering or interruption. This improves path continuity and extrusion stability during the printing process. The formula is as follows: In the formula, Let be the rotation angle of the nozzle around its central axis. The pitch is the height increment corresponding to a single rotation. The path radius is the value corresponding to the rotation angle. This represents the path height at the corresponding rotation angle.

[0009] Preferably, step S3 further includes a process constraint and flow compensation step: Step S33: Load the preset additive manufacturing process constraints, which include at least one of the following: nozzle physical size, printing layer height, extrusion line width, minimum turning radius, maximum overhang angle, interlayer overlap, path self-intersection restriction, and nozzle collision boundary. Step S34: Calculate and match the corresponding material extrusion flow rate compensation parameters in real time based on the geometric offset between adjacent deposition paths or adjacent deposition layers, and synchronously bind the flow rate compensation parameters with the path point coordinates. Interlayer radius offset is used to quantify the radius variation between adjacent deposition layers, providing a geometric basis for dynamic adjustment of extrusion flow rate and ensuring uniform material deposition in variable cross-section regions. The interlayer radius offset is calculated using the following formula: In the formula, For height Interlayer radius offset at that location. The path radius of the current sedimentary layer. The path radius of the previous sedimentary layer. To print layer height; The extrusion flow rate calculation combines the current extrusion linewidth, printing layer height, and nozzle feed speed to dynamically adjust the material output according to changes in geometric contours, avoiding uneven wall thickness or over-extrusion defects. The extrusion flow rate satisfies the following: In the formula, For height Material extrusion flow rate at the location, This is the material extrusion compensation coefficient. This represents the extrusion linewidth of the current deposition layer. To print layer height, The nozzle feed rate is used to determine the stability of the deposition of variable cross-section structures by dynamically adjusting the extrusion flow rate, feed rate, and nozzle temperature based on the geometric offset in the large-scale particle extrusion additive manufacturing process.

[0010] Preferably, step S4 includes the following steps: Step S41: Generate control instructions for the additive manufacturing equipment based on the generated manufacturing path; Step S42: Use a reverse parsing mechanism to decompose the control commands in reverse, reconstruct the path trajectory and extrusion parameter sequence, and perform printability verification; The reverse analysis mechanism is based on the G-code instruction set specification, parsing motion and status commands in control instructions line by line. For linear and circular interpolation commands, it extracts the target coordinate points (X, Y, Z), extrusion amount E, and feed rate F, restoring continuous command segments into a sequence of path points and corresponding extrusion parameters. For coordinate setting commands, it extracts the current coordinate reset value and updates the coordinate system origin offset. For commands switching between absolute and relative coordinate modes, it records the current coordinate mode to ensure consistency in subsequent coordinate analysis. During the reverse analysis process, the length and turning angle of each path segment are recorded simultaneously, generating structured trajectory data that can be verified and calculated.

[0011] The judgment criteria and threshold setting logic for each item in the printability verification are as follows: Path continuity verification: Verify whether the distance between the endpoints of adjacent path segments is less than the preset tolerance threshold; verify whether the turning angle between path segments exceeds the maximum turning angle threshold allowed by the equipment. If it exceeds, it is judged as a sharp turn defect. The adjustment method is to insert an arc transition segment at the turning point. Extrusion continuity verification: Verify the abrupt change in the extrusion quantity E value between adjacent path segments, calculate the rate of change of extrusion quantity in a single segment, and if the rate of change exceeds the preset allowable abrupt change threshold, it is determined to be an extrusion quantity abrupt change defect, triggering the adjustment of the flow parameters of the corresponding segment; verify the non-negativity of extrusion quantity, and if a negative change occurs, it is determined to be a backflow anomaly. Dimensional compliance verification: Based on the path coordinate sequence obtained from reverse analysis, calculate the outer envelope dimension and minimum wall thickness of the target object, and compare them with the upper limit of the forming area dimension of the additive manufacturing equipment. If any dimension exceeds the area boundary, it is judged as an over-limit defect. The minimum wall thickness is compared with the minimum wall thickness threshold of the process. If it is less than the threshold, it is judged as a thin-wall failure defect. Printing time verification: Based on the length of each path segment and the feed speed F value obtained from reverse parsing, the estimated total printing time is calculated segment by segment. If the time exceeds the upper limit of the continuous running time of the device or the maximum allowable printing time set by the user, it is judged as a timeout defect. Inter-layer consistency verification: For path data containing floor height information, the path segments are grouped into layers according to height range, and the deviation between the actual cumulative height and the nominal floor height of each layer is calculated. If the deviation exceeds the preset floor height tolerance threshold, it is judged as a floor height inconsistency defect; the cross-sectional profile deviation between adjacent layers is calculated, and if the deviation exceeds the maximum allowable misalignment threshold between layers, it is judged as an inter-layer misalignment defect. Each verification item is configured with an independent threshold and a tiered handling strategy: if all verification items pass, the status is determined to be "passed" and a control command is directly output; if there are correctable local defects, the status is determined to be "corrected" and the verification is re-executed after correction according to the preset automatic adjustment logic; if there are uncorrectable structural defects, the status is determined to be "rejected" and a defect type description and design adjustment suggestions are returned; for on-site interactive manufacturing scenarios, the processing results of both the "corrected" and "rejected" statuses must be manually confirmed before the corresponding operation is executed, and the control command that passes the verification is accompanied by a unique traceability identifier and generated metadata information; Printability verification includes at least one of the following: path continuity verification, extrusion continuity verification, dimensional compliance verification, printing time verification, and interlayer consistency verification. The hierarchical processing mechanism is as follows: if all verification items are qualified, the status is determined to be passable, and control commands are directly output; if there are correctable local defects, the status is determined to be correctable, and the path parameters or extrusion process parameters are automatically adjusted before re-verification; if there are uncorrectable structural defects, the status is determined to be rejectable, and a defect type description and design adjustment suggestions are returned. For on-site interactive manufacturing scenarios, the processing results of corrective and rejectable statuses must be manually confirmed before the corresponding operation is executed. Control commands that pass verification are accompanied by a unique traceability identifier and generated metadata information. Step S43: Based on the verification results, the control commands are processed in a hierarchical manner, and the control commands that pass the verification are output to the additive manufacturing equipment.

[0012] Preferably, in step S3, while generating the manufacturing path, a visual preview interface is generated simultaneously. The visual preview interface includes a preview of the rotating appearance of the target object and a preview of the nozzle movement path. Both types of preview data originate from the same structured path data, ensuring that the preview effect is completely identical to the actual printing path. The entire process is set up with a three-level homogeneous data structure, namely, contour parameter layer data storing the parsed design intent and non-mesh expression data, path layer data storing the deposition path coordinates and supporting process parameters, and instruction metadata storing control instruction generation information and traceability identifiers. Corresponding mapping relationships and hash verification mechanisms are established between the three levels of data.

[0013] The present invention has the following beneficial effects: (1) This invention eliminates the intermediate steps of triangular mesh construction and general slicing processing, directly generating executable manufacturing paths based on design intent, thus fundamentally avoiding the interference of mesh topology issues on the manufacturing process. The method eliminates the need for pre-processing such as mesh repair, reducing the workload before path generation, lowering the requirements for professional model processing skills in the manufacturing process, and ensuring the smooth progress of the path generation stage. This mode can improve the first-time success rate of path generation, support lightweight and high-efficiency manufacturing needs, adapt to rapid manufacturing applications in different scenarios, and improve the ease of implementation of the path generation stage.

[0014] (2) The present invention constructs a complete execution link from design intent input to manufacturing instruction output, which can complete the entire process of input parsing, path generation, process compensation and instruction verification within the same system without switching between multiple independent software. The method can reduce the operation cost and time loss caused by switching between multiple links, improve the conversion efficiency from design input to physical manufacturing, and lower the operation threshold of the entire manufacturing process. This mode ensures the integrity and smoothness of the manufacturing link, supports the needs of rapid and real-time manufacturing, and realizes the continuous flow from design intent to physical output.

[0015] (3) The present invention uses proprietary path compilation logic to generate manufacturing paths, which is not limited by the logic of fixed slicing algorithms and can be adapted to various customized path strategies such as continuous spiral, variable layer height, and non-planar. The method embeds additive manufacturing process constraints into the path generation process, and can match the corresponding extrusion compensation parameters with geometric features to ensure the uniformity of material deposition and molding quality. This mode can implement differentiated structural optimization and path design requirements, and improve the controllability and process adaptability of manufacturing paths.

[0016] (4) This invention sets up a reverse verification mechanism for manufacturing instructions to build a complete closed loop for manufacturing safety verification. After generating manufacturing control instructions, the printability is verified from multiple dimensions by reverse parsing the path and extrusion parameters, and the verification results are graded to identify potential process risks in advance. This mechanism can reduce printing failures and material losses, improve the controllability of the manufacturing process, ensure the safe and stable operation of printing equipment and manufacturing process, and reduce the operational risks of the manufacturing process. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1This is a flowchart illustrating the additive manufacturing path generation method based on design intent parsing and non-mesh representation of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Please see Figure 1 This invention is an additive manufacturing path generation method based on design intent parsing and non-mesh representation, comprising the following steps: Step S1: Collect the user's input design intent, parse and process the design intent to obtain the non-mesh representation data of the target object; the design intent information includes at least one of the following: the geometric dimensions, contour shape, and process requirements of the target object; Step S2: Construct the manufacturing path generation domain based on non-mesh representation data. The manufacturing path generation domain is different from the traditional slicing domain. In essence, it is a computational space that carries all the geometric information required for path generation. Each coordinate point of the nozzle trajectory can be directly solved by the function or field value in this domain without the need for intermediate mesh discretization steps. The construction of the manufacturing path generation domain does not depend on the slicing process of the triangular facet mesh. Step S3: Based on the preset additive manufacturing process constraints, generate a manufacturing path for material deposition within the manufacturing path generation domain; Step S4: Output control instructions for the additive manufacturing equipment according to the generated manufacturing path, and perform printability verification on the control instructions. After the verification is passed, output the control instructions to the additive manufacturing equipment.

[0021] Step S1 uses a semantic parameter parsing approach to process the design intent, corresponding to natural language, speech, or structured parameter class inputs, and includes the following steps: Step S11: Collect user input in natural language text, voice commands, or structured design parameters; Step S12: Use an artificial intelligence model to perform semantic recognition and parameter mapping on the input content, extract the geometric dimensions, texture type, and process parameters of the target object, and generate a design parameter vector; The AI ​​model employs an encoder-decoder network based on the Transformer architecture. The input layer receives a sequence of natural language text after word segmentation and encoding, or directly receives key-value pair embeddings of structured design parameters. The output layer generates a fixed-dimensional design parameter vector, which contains geometric dimension parameters, texture type encoding, and process parameters in a predetermined order. Parameter mapping rules are implemented using a pre-defined JSON dictionary, which maps the indices of each dimension of the output vector to the names of geometric variables and process constraint parameters required in subsequent steps. During parsing, values ​​are retrieved by index and assigned to the corresponding radius-height function variables and process constraints. During training, manually labeled historical input-output samples are used as supervision data. The label files are stored in JSON format, and the label structure is completely consistent with the dimensions of the output vector. The mean squared error loss function is used for training, and the Adam optimizer is selected. After training until the validation set loss converges, the model parameters are fixed for inference. Step S13: Output the non-mesh representation data, with the design parameter vector as the target object, to the downstream path generation stage; The semantic parameter parsing route also includes a model closed-loop optimization step: collecting the actual printing results of the additive manufacturing equipment, recording the sample data of successful and failed printing, constructing new training sample pairs by combining the design intent, non-mesh expression data and printing result labels of the corresponding samples, and iteratively optimizing the parameter mapping accuracy of the artificial intelligence model by adding new training samples.

[0022] Step S1 uses a hand-drawn outline analysis route to process the design intent, corresponding to the user's hand-drawn two-dimensional outline curve input, including the following steps: Step S14: Collect the two-dimensional side view contour point array input by the user through touch interaction; Step S15: Perform smoothing, noise reduction, boundary compliance verification, and geometric normalization on the hand-drawn outline points to complete the outline cleanup; Step S16: Establish a cylindrical coordinate system with the central rotation axis as the z-axis, and convert the cleaned contour into normalized contour data in the form of a radius-height function. The radius-height function is used to characterize the variation of the radius of the outer contour of the body of revolution with height. It can directly convert the 2D hand-drawn contour into a mathematical expression that can be used for path calculation without constructing a triangular mesh model. The formula is: In the formula, r is the radius of the outer contour of the cross section of the target rotating body at height z, and z is the height coordinate along the central rotation axis; Step S17: Using the normalized contour data as the non-mesh representation data of the target object, output it to the downstream path generation stage.

[0023] Step S2 includes the following steps: Step S21: Match the corresponding construction logic according to the type of non-mesh representation data. Non-mesh representation data includes at least one of parametric surfaces, implicit functions, distance fields, curve networks, design parameter vectors, and normalized contour data. Parametric surfaces define the surface shape of the target object using parametric surface equations. Two independent parameters within the parameter domain are mapped to three-dimensional spatial coordinates. The entire parameter domain is directly used as the path generation domain on the two-dimensional manifold. During path calculation, parameters are sampled within the parameter domain according to preset path spacing and direction to obtain a sequence of parameter coordinates, which is then converted into nozzle motion trajectory points in three-dimensional space via surface mapping. Implicit functions express the geometric surface of the target object using implicit equations, and define the entity region and the external region by the positive and negative values ​​of the function values. The path generation domain is constructed by sampling the implicit functions within the bounding box space of the target object, constructing a directed distance field as an intermediate expression, and then generating a series of equidistant offset contours by extracting isosurfaces. During path calculation, the intersection of the implicit equation and the horizontal plane is solved at each height layer to obtain a closed contour curve, and the sequence of points on the curve is directly used as the nozzle motion coordinates. The distance field expresses the symbolic distance from the surface of the target object to any point in space using distance values ​​at discrete points in space. The domain of the distance field directly constitutes the path generation domain. During path calculation, the gradient information of the distance field is used to determine the surface normal, and a continuous motion trajectory is constructed by tracing the geodesic or equidistant offset surface on the zero isosurface. Curve networks are used to define the skeleton or contour features of a target object with a set of structured feature curves. The path generation domain uses these feature curves as guides to generate a path space covering the target surface through sweeping, lofting, or constructing tubular neighborhoods. During path calculation, discrete sampling with equal arc length or equal parameters is performed along the direction of the feature curves, and a continuous path network is generated by combining the topological connection relationship between the curves. The design parameter vector matches a preset geometric template based on the parameter vector's type identifier. The corresponding parameterized surface or contour function is generated through parameter instantiation, and then the path generation domain is constructed according to the corresponding surface or contour logic. During path calculation, the path generation template bound to the geometric template is called, and the values ​​in the design parameter vector are substituted into the template equation to directly calculate the nozzle's motion coordinates. The standardized contour data expresses the radius variation law of the target object along a certain coordinate axis using a contour function or a list of cross-section points. The contour is rotated around the central axis to generate a body of revolution, and the internal region of the body of revolution constitutes the path generation domain. During path calculation, path parameterization is performed within the domain of the contour function to directly calculate the spatial motion trajectory of the nozzle and map the two-dimensional contour into a three-dimensional path. Step S22: Construct a manufacturing path generation domain in the form of a path field, orientation field, or layer function space based on non-mesh representation data. The manufacturing path generation domain is used to directly calculate the nozzle motion trajectory without first generating a triangular 3D mesh and then performing slicing operations.

[0024] Step S3 includes the path generation step: Step S31: Match the path generation strategy according to the target object type and process requirements. The generated manufacturing path includes at least one of the following: continuous spiral path, layered contour path, non-planar path, multi-axis linkage path, and variable layer height path. Step S32: The proprietary process path compiler directly calculates the motion coordinate sequence of the nozzle based on the manufacturing path generation domain to complete the generation of the manufacturing path; For rotating objects, a continuous spiral path generation strategy is adopted. The nozzle's motion coordinates are calculated using cylindrical coordinate parametric equations, enabling continuous extrusion motion without layering or interruption. This improves path continuity and extrusion stability during the printing process. The formula is as follows: In the formula, Let be the rotation angle of the nozzle around its central axis. The pitch is the height increment corresponding to a single rotation. The path radius is the value corresponding to the rotation angle. This represents the path height at the corresponding rotation angle.

[0025] Step S3 also includes process constraint and flow compensation steps: Step S33: Load the preset additive manufacturing process constraints, which include at least one of the following: nozzle physical size, printing layer height, extrusion line width, minimum turning radius, maximum overhang angle, interlayer overlap, path self-intersection restriction, and nozzle collision boundary. Step S34: Calculate and match the corresponding material extrusion flow rate compensation parameters in real time based on the geometric offset between adjacent deposition paths or adjacent deposition layers, and synchronously bind the flow rate compensation parameters with the path point coordinates. Interlayer radius offset is used to quantify the radius variation between adjacent deposition layers, providing a geometric basis for dynamic adjustment of extrusion flow rate and ensuring uniform material deposition in variable cross-section regions. The interlayer radius offset is calculated using the following formula: In the formula, For height Interlayer radius offset at that location. The path radius of the current sedimentary layer. The path radius of the previous sedimentary layer. To print layer height; The extrusion flow rate calculation combines the current extrusion linewidth, printing layer height, and nozzle feed speed to dynamically adjust the material output according to changes in geometric contours, avoiding uneven wall thickness or over-extrusion defects. The extrusion flow rate satisfies the following: In the formula, For height Material extrusion flow rate at the location, This is the material extrusion compensation coefficient. This represents the extrusion linewidth of the current deposition layer. To print layer height, The nozzle feed rate is used to determine the stability of the deposition of variable cross-section structures by dynamically adjusting the extrusion flow rate, feed rate, and nozzle temperature based on the geometric offset in the large-scale particle extrusion additive manufacturing process.

[0026] Step S4 includes the following steps: Step S41: Generate control instructions for the additive manufacturing equipment based on the generated manufacturing path; Step S42: Use a reverse parsing mechanism to decompose the control commands in reverse, reconstruct the path trajectory and extrusion parameter sequence, and perform printability verification; The reverse analysis mechanism is based on the G-code instruction set specification, parsing motion and status commands line by line in the control instructions. For linear and circular interpolation commands, it extracts the target coordinate points (X, Y, Z), extrusion amount E, and feed rate F, restoring continuous command segments into a sequence of path points and corresponding extrusion parameters. For coordinate setting commands, it extracts the current coordinate reset value and updates the coordinate system origin offset. For commands switching between absolute and relative coordinate modes, it records the current coordinate mode to ensure consistency in subsequent coordinate analysis. During the reverse analysis process, the length and rotation angle of each path segment are recorded simultaneously, generating structured trajectory data that can be verified and calculated.

[0027] The judgment criteria and threshold setting logic for each item in the printability verification are as follows: Path continuity verification: Verify whether the distance between the endpoints of adjacent path segments is less than the preset tolerance threshold; verify whether the turning angle between path segments exceeds the maximum turning angle threshold allowed by the equipment. If it exceeds, it is judged as a sharp turn defect. The adjustment method is to insert an arc transition segment at the turning point. Extrusion continuity verification: Verify the abrupt change in the extrusion quantity E value between adjacent path segments, calculate the rate of change of extrusion quantity in a single segment, and if the rate of change exceeds the preset allowable abrupt change threshold, it is determined to be an extrusion quantity abrupt change defect, triggering the adjustment of the flow parameters of the corresponding segment; verify the non-negativity of extrusion quantity, and if a negative change occurs, it is determined to be a backflow anomaly. Dimensional compliance verification: Based on the path coordinate sequence obtained from reverse analysis, calculate the outer envelope dimension and minimum wall thickness of the target object, and compare them with the upper limit of the forming area dimension of the additive manufacturing equipment. If any dimension exceeds the area boundary, it is judged as an over-limit defect. The minimum wall thickness is compared with the minimum wall thickness threshold of the process. If it is less than the threshold, it is judged as a thin-wall failure defect. Printing time verification: Based on the length of each path segment and the feed speed F value obtained from reverse parsing, the estimated total printing time is calculated segment by segment. If the time exceeds the upper limit of the continuous running time of the device or the maximum allowable printing time set by the user, it is judged as a timeout defect. Inter-layer consistency verification: For path data containing floor height information, the path segments are grouped into layers according to height range, and the deviation between the actual cumulative height and the nominal floor height of each layer is calculated. If the deviation exceeds the preset floor height tolerance threshold, it is judged as a floor height inconsistency defect; the cross-sectional profile deviation between adjacent layers is calculated, and if the deviation exceeds the maximum allowable misalignment threshold between layers, it is judged as an inter-layer misalignment defect. Each verification item is configured with an independent threshold and a tiered handling strategy: if all verification items pass, the status is determined to be "passed" and a control command is directly output; if there are correctable local defects, the status is determined to be "corrected" and the verification is re-executed after correction according to the preset automatic adjustment logic; if there are uncorrectable structural defects, the status is determined to be "rejected" and a defect type description and design adjustment suggestions are returned; for on-site interactive manufacturing scenarios, the processing results of both the "corrected" and "rejected" statuses must be manually confirmed before the corresponding operation is executed, and the control command that passes the verification is accompanied by a unique traceability identifier and generated metadata information; Printability verification includes at least one of the following: path continuity verification, extrusion continuity verification, dimensional compliance verification, printing time verification, and interlayer consistency verification. The hierarchical processing mechanism is as follows: if all verification items are qualified, the status is determined to be passable, and control commands are directly output; if there are correctable local defects, the status is determined to be correctable, and the path parameters or extrusion process parameters are automatically adjusted before re-verification; if there are uncorrectable structural defects, the status is determined to be rejectable, and a defect type description and design adjustment suggestions are returned. For on-site interactive manufacturing scenarios, the processing results of corrective and rejectable statuses must be manually confirmed before the corresponding operation is executed. Control commands that pass verification are accompanied by a unique traceability identifier and generated metadata information. Step S43: Based on the verification results, the control commands are processed in a hierarchical manner, and the control commands that pass the verification are output to the additive manufacturing equipment.

[0028] While generating the manufacturing path in step S3, a visual preview interface is generated simultaneously. The visual preview interface includes a preview of the rotating appearance of the target object and a preview of the nozzle movement path. Both types of preview data come from the same structured path data, ensuring that the preview effect is completely identical to the actual printing path. The entire process is set up with a three-level homogeneous data structure, namely, contour parameter layer data storing the parsed design intent and non-mesh expression data, path layer data storing deposition path coordinates and supporting process parameters, and instruction metadata storing control command generation information and traceability identifiers. Corresponding mapping relationships and hash verification mechanisms are established between the three levels of data.

[0029] One specific application of this embodiment is: This embodiment describes a method for generating a path for large-scale particle extrusion additive manufacturing of a rotating component. The target component is a variable-diameter rotating body with a maximum outer diameter of 320mm and a total height of 450mm. The printing layer height is... The nozzle has an inner diameter of 4mm and is made of a mixture of sand and gravel particles.

[0030] S1: Collect and parse the user's input design intent to obtain non-grid representation data of the target object, including two optional input schemes.

[0031] Option 1: Use natural language parameter input mode.

[0032] The input is: a solid of revolution with a total height of 450mm, a bottom radius of 160mm, and a radius that shrinks to 90mm at a height of 225mm. A lightweight Transformer model is used to perform semantic parsing, extract geometric parameters, and generate parameter vectors. These vector data are directly used as non-mesh representation data. The model undergoes continuous closed-loop iteration: 350 sets of successful and defective printing samples are collected, and the design parameters, contour data, and printing result labels are used to form training samples. The model weights are repeatedly updated to improve parameter recognition accuracy.

[0033] Option 2: Use the hand-drawn outline input mode.

[0034] The operator draws a two-dimensional side view contour on the touch interface, obtaining a discrete contour point sequence. First, the point sequence is smoothed, filtered, boundary checked, and normalized. Then, a cylindrical coordinate system is established with the rotation center as the z-axis, and a contour function whose radius varies with height is constructed. In the formula, The cross-sectional radius corresponding to the current height. This represents the axial height coordinate.

[0035] The component's contour function is set in segments: Altitude range : Altitude range : The piecewise function is used as the normalized contour data to complete the construction of the non-mesh model, without generating a three-dimensional triangular mesh model throughout the process.

[0036] S2: Construct a manufacturing path generation domain based on non-mesh representation data.

[0037] This embodiment uses a piecewise radius function as non-mesh data to directly construct a two-dimensional path field in cylindrical coordinates, without the need for STL mesh slicing. The rotational space formed by the contour curve rotating around the z-axis is used as the path generation domain, retaining only the internal region of the contour as the nozzle motion space and discarding invalid external regions.

[0038] S3: Generate a continuous spiral extrusion path under process constraints and simultaneously calculate the dynamic extrusion flow rate.

[0039] This embodiment uses a continuous spiral uninterrupted extrusion path, and the cylindrical coordinate equation of the spiral trajectory is as follows: In the formula, The rotation angle represents the height increment per revolution. The nozzle changes its radial radius synchronously with the rotation angle, strictly following the variable diameter profile to achieve continuous molding.

[0040] Loading process constraints: Extrusion line width Nozzle feed speed Extrusion compensation coefficient .

[0041] First, calculate the change in inter-story radius using the following formula: when At that time, the abrupt change value of the radius of adjacent layers can be calculated, and the extrusion amount can be adjusted in real time according to the radius offset.

[0042] The instantaneous extrusion flow rate of the material is calculated in real time using the following formula: Substituting the fixed process parameters into this embodiment: The basic extrusion flow rate is calculated; in the radius contraction and expansion sections, combined with Slightly increase or decrease the flow rate to avoid accumulation or material shortage at the diameter change position; the path point coordinates are bound to the extrusion flow rate corresponding to each position.

[0043] Simultaneously with path generation, a homogeneous preview is initiated: while generating the nozzle spiral trajectory coordinates, the component's solid shape is rendered concurrently. All preview data originates from the same set of contour functions and path coordinates, ensuring complete consistency between the simulation image and the actual extrusion trajectory. Three layers of homogeneous data are also constructed: a contour parameter layer, a path coordinate and process parameter layer, and an instruction metadata layer. A hash mapping is established between these three layers to prevent data inconsistencies.

[0044] S4: Generate equipment motion commands and perform reverse verification.

[0045] The spiral path is discretized into tens of thousands of coordinate points and converted into motion control commands that the extrusion equipment can recognize. A reverse analysis algorithm is used to reverse the G-code commands into trajectory coordinates and extrusion flow sequence, and four verifications are carried out in sequence: path continuity verification, extrusion flow continuity verification, external dimensions verification, and total printing time verification.

[0046] Verification and execution of the three-level handling strategy: 1. Once all projects pass the test, control instructions will be issued directly and a unique traceability number will be assigned. 2. If only localized fluctuations in extrusion volume exist, the program will automatically fine-tune the flow parameters of the corresponding section and then re-verify. 3. If structural defects such as path self-intersection or nozzle collision occur, terminate the command output and provide the operator with contour modification suggestions.

[0047] For on-site online prototyping scenarios, all corrections and rejections require manual confirmation before the process can continue.

[0048] Finally, the entire variable-diameter rotating body is extruded and molded along a continuous spiral path, resulting in a component with uniform wall thickness and no obvious defects in the transition area of ​​the variable diameter.

[0049] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. An additive manufacturing path generation method based on design intent parsing and non-mesh representation, comprising the following steps: Step S1: Collect the design intent information input by the user, parse and process the design intent to obtain the non-grid representation data of the target object; Step S2: Construct a manufacturing path generation domain based on non-mesh representation data. The construction of the manufacturing path generation domain does not depend on the slicing process of the triangular mesh. Step S3: Based on the preset additive manufacturing process constraints, generate a manufacturing path for material deposition within the manufacturing path generation domain; Step S4: Output control instructions for the additive manufacturing equipment according to the generated manufacturing path, and perform printability verification on the control instructions. After the verification is passed, output the control instructions to the additive manufacturing equipment.

2. The additive manufacturing path generation method based on design intent parsing and non-mesh representation according to claim 1, characterized in that: Step S1 employs a semantic parameter parsing approach to process design intent, corresponding to natural language, speech, or structured parameter class inputs, and includes the following steps: Step S11: Collect user input in natural language text, voice commands, or structured design parameters; Step S12: Use an artificial intelligence model to perform semantic recognition and parameter mapping on the input content, extract the geometric dimensions, texture type, and process parameters of the target object, and generate a design parameter vector; Step S13: Output the non-mesh representation data, with the design parameter vector as the target object, to the downstream path generation stage; The semantic parameter parsing route also includes an offline model optimization step: after multiple rounds of printing, the actual printing results of the additive manufacturing equipment are collected, the sample data of successful and failed printing are recorded, and the design intent, non-mesh expression data and printing result labels of the corresponding samples are constructed into new training sample pairs. The parameter mapping accuracy of the artificial intelligence model is iteratively optimized by adding new training samples.

3. The additive manufacturing path generation method based on design intent parsing and non-mesh representation according to claim 1, characterized in that: Step S1 uses a hand-drawn outline analysis route to process the design intent, corresponding to the user's hand-drawn two-dimensional outline curve input, and includes the following steps: Step S14: Collect the two-dimensional side view contour point array input by the user through touch interaction; Step S15: Perform smoothing, noise reduction, boundary compliance verification, and geometric normalization on the hand-drawn outline points to complete the outline cleanup; Step S16: Establish a cylindrical coordinate system with the central rotation axis as the z-axis. Convert the cleaned contour into normalized contour data in the form of a radius-height function. The radius-height function characterizes the variation of the radius of the outer contour of the body of revolution with height. It can directly convert the two-dimensional hand-drawn contour into a mathematical expression that can be used for path calculation without constructing a triangular mesh model. The formula is: In the formula, r is the radius of the outer contour of the cross section of the target rotating body at height z, and z is the height coordinate along the central rotation axis; Step S17: Using the normalized contour data as the non-mesh representation data of the target object, output it to the downstream path generation stage.

4. The additive manufacturing path generation method based on design intent parsing and non-mesh representation according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: Match the corresponding construction logic according to the type of non-mesh representation data. Non-mesh representation data includes at least one of parametric surfaces, implicit functions, distance fields, curve networks, design parameter vectors, and normalized contour data. Step S22: Construct a manufacturing path generation domain in the form of a path field, orientation field, or layer function space based on non-mesh representation data. The manufacturing path generation domain is used to directly calculate the nozzle motion trajectory without first generating a triangular 3D mesh and then performing slicing operations.

5. The additive manufacturing path generation method based on design intent parsing and non-mesh representation according to claim 1, characterized in that: Step S3 includes a path generation step: Step S31: Match the path generation strategy according to the target object type and process requirements. The generated manufacturing path includes at least one of the following: continuous spiral path, layered contour path, non-planar path, multi-axis linkage path, and variable layer height path. Step S32: The process path compiler directly calculates the motion coordinate sequence of the nozzle based on the manufacturing path generation domain to complete the generation of the manufacturing path; For rotating objects, a continuous spiral path generation strategy is adopted. The nozzle's motion coordinates are calculated using cylindrical coordinate parametric equations, enabling continuous extrusion motion without layering or interruption. This improves path continuity and extrusion stability during the printing process. The formula is as follows: In the formula, Let be the rotation angle of the nozzle around its central axis. The pitch is the height increment corresponding to a single rotation. The path radius is the value corresponding to the rotation angle. This represents the path height at the corresponding rotation angle.

6. The additive manufacturing path generation method based on design intent parsing and non-mesh representation according to claim 5, characterized in that: Step S3 also includes process constraint and flow compensation steps: Step S33: Load the preset additive manufacturing process constraints, which include at least one of the following: nozzle physical size, printing layer height, extrusion line width, minimum turning radius, maximum overhang angle, interlayer overlap, path self-intersection restriction, and nozzle collision boundary. Step S34: Calculate and match the corresponding material extrusion flow rate compensation parameters in real time based on the geometric offset between adjacent deposition paths or adjacent deposition layers, and synchronously bind the flow rate compensation parameters with the path point coordinates. Interlayer radius offset is used to quantify the radius variation between adjacent deposition layers, providing a geometric basis for dynamic adjustment of extrusion flow rate and ensuring uniform material deposition in variable cross-section regions. The interlayer radius offset is calculated using the following formula: In the formula, For height Interlayer radius offset at that location The path radius of the current sedimentary layer. The path radius of the previous sedimentary layer. To print layer height; The extrusion flow rate calculation combines the current extrusion linewidth, printing layer height, and nozzle feed speed to dynamically adjust the material output according to changes in geometric contours, avoiding uneven wall thickness or over-extrusion defects. The extrusion flow rate satisfies the following: In the formula, For height Material extrusion flow rate at the location, This is the material extrusion compensation coefficient. This represents the extrusion linewidth of the current deposition layer. To print layer height, This refers to the nozzle feed speed.

7. The additive manufacturing path generation method based on design intent parsing and non-mesh representation according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: Generate control instructions for the additive manufacturing equipment based on the generated manufacturing path; Step S42: Use a reverse parsing mechanism to decompose the control commands in reverse, reconstruct the path trajectory and extrusion parameter sequence, and perform printability verification; Printability verification includes at least one of the following: path continuity verification, extrusion continuity verification, dimensional compliance verification, printing time verification, and interlayer consistency verification. The hierarchical processing mechanism is as follows: if all verification items are qualified, the status is determined to be passable, and control commands are directly output; if there are correctable local defects, the status is determined to be correctable, and the path parameters or extrusion process parameters are automatically adjusted before re-verification; if there are uncorrectable structural defects, the status is determined to be rejectable, and a defect type description and design adjustment suggestions are returned. For on-site interactive manufacturing scenarios, the processing results of corrective and rejectable statuses must be manually confirmed before the corresponding operation is executed. Control commands that pass verification are accompanied by a unique traceability identifier and generated metadata information. Step S43: Based on the verification results, the control commands are processed in a hierarchical manner, and the control commands that pass the verification are output to the additive manufacturing equipment.

8. The additive manufacturing path generation method based on design intent parsing and non-mesh representation according to claim 1, characterized in that: While generating the manufacturing path in step S3, a visual preview interface is generated simultaneously. The visual preview interface includes a preview of the rotating appearance of the target object and a preview of the nozzle movement path. Both types of preview data come from the same structured path data, ensuring that the preview effect is completely identical to the actual printing path. The entire process is set up with a three-level homogeneous data structure, namely, contour parameter layer data storing the parsed design intent and non-mesh expression data, path layer data storing deposition path coordinates and supporting process parameters, and instruction metadata storing control command generation information and traceability identifiers. Corresponding mapping relationships and hash verification mechanisms are established between the three levels of data.

Images