A fixture generation method and system based on parameterized templates and intelligent reasoning

Abstract

The application discloses a kind of based on parameterization template and intelligent reasoning clamp generation method and system, it is related to jig design technical field, including: obtaining workpiece digital model and processing process file, extracting workpiece geometric feature, datum, process and tool path information;Build typed constraint graph, with node and edge represent various tooling constraint relationship, and cutting load spectrum is encoded as the dynamic constraint condition of constraint edge;Search jig template skeleton in parameterization template library;Constraint solving is carried out to template skeleton, real-time call order-reduction physical simulation model is completed mechanical evaluation, rely on feedback signal to guide search direction;After solving convergence, according to feasible solution instantiation template skeleton, output jig model and solving certificate.The application improves solving efficiency and scheme reliability by simulation embedded closed-loop optimization, relies on parameterization template to consider design reusability, dual output can realize design whole-process traceability, meet the high-precision and compliance requirements of high-end manufacturing tooling design.

Description

Technical Field

[0001] This invention relates to the field of fixture design technology, specifically to a fixture generation method and system based on parametric templates and intelligent reasoning. Background Technology

[0002] As a core piece of process equipment in mechanical manufacturing systems, the design quality of fixtures directly determines the machining accuracy, surface quality, and production efficiency of workpieces. With the continuous advancement of information technology in manufacturing, computer-aided fixture design technology has evolved from early two-dimensional drawing assistance and 3D standard parts library retrieval to a semi-automated design paradigm based on feature recognition, knowledge engineering, and parametric driving. Simultaneously, the widespread adoption of product model definition technology allows workpiece digital models to carry complete design intent and manufacturing semantics, while CNC programming systems can provide detailed tool movement trajectories and cutting parameter information. This lays the data foundation for automatically extracting the multi-dimensional constraints required for fixture design from workpiece models and process data. At the solution technology level, the continuous maturation of constraint satisfaction problem-solving frameworks, finite element analysis, and reduced-order model methods makes efficient searching and physical verification in complex constraint spaces possible. The integration of these technological elements has propelled the evolution of fixture design from experience-dependent to data-driven, and from trial-and-error iteration to simulation-based verification.

[0003] However, existing intelligent fixture design methods still have systemic shortcomings when dealing with complex manufacturing tasks. First, at the information fusion level, existing methods can usually only use the geometric topology information of the workpiece to match fixture schemes. It is difficult to simultaneously analyze the hierarchical relationship between design datum, process datum, and inspection datum from the workpiece model, nor can it spatially map the tool path data in the CNC program with the geometric features of the workpiece. This results in a lack of process semantic support for the identification of locatable and clampable areas. When facing multi-process and multi-clamping machining tasks, the integrity and consistency of the design input information are difficult to guarantee. Second, at the constraint expression level, existing methods mostly treat positioning constraints, clamping constraints, and assembly constraints as static geometric rules. They have not established an explicit interference constraint model between the tool movement space and the fixture component arrangement space, nor have they transformed the cutting load spectrum that changes with the tool path position during machining into calculable dynamic mechanical constraints. This results in a lack of effective means to verify the rigid body stability and local deformation suppression capability of the design scheme under dynamic machining conditions.

[0004] Furthermore, at the solution mechanism level, existing methods generally adopt a serial processing mode of "first generating a complete fixture scheme, and then calling finite element software for independent mechanical property verification." The mechanical evaluation results only serve as the final basis for the scheme judgment and cannot actively guide the parameter search direction in the form of heuristic signals during the solution search process. When the verification fails, manual intervention is often required to readjust the parameters and iterate repeatedly. For high-dimensional fixture design spaces with multiple parameters and constraints, it is difficult to optimize the matching relationship between search efficiency and scheme feasibility. The above shortcomings together result in the existing methods failing to simultaneously meet the engineering reliability and design efficiency of the generated schemes when dealing with thin-walled structural parts, complex curved surface parts, and other workpieces with stringent requirements for clamping stiffness and machining accuracy. Summary of the Invention

[0005] In view of the above-mentioned problems, the present invention is proposed.

[0006] Therefore, the technical problem solved by this invention is that existing fixture generation methods have problems such as incomplete fusion of design input information, separation between static constraint expression and dynamic load modeling, separation of the solution process and mechanical performance verification leading to low search efficiency, and how to encode the cutting load spectrum into the dynamic constraint conditions of a unified constraint model, embed real-time physical simulation feedback in the constraint solution process, and realize the synchronous output of fixture parameterized model and traceable verification certificate.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, embodiments of the present invention provide a fixture generation method based on parameterized templates and intelligent reasoning, comprising the following steps: S1: Obtain the digital model of the workpiece and the machining process file, and extract the geometric features, datum information, process content and tool path information of the workpiece; S2: Construct a typed constraint graph, in which nodes represent workpiece features and fixture element placeholders, and edges represent positioning relationships, clamping relationships, support relationships, assembly relationships and machining accessibility relationships. The cutting load spectrum parsed from the toolpath information is encoded as the dynamic constraint conditions of the edges. S3: Retrieve a template skeleton that matches the workpiece digital model from a preset parametric template library. The template skeleton defines the type, layout topology, and adjustable parameter range of the fixture. S4: Perform constraint satisfaction solution on the template skeleton. During the solution process, for each candidate parameter combination, call the pre-built reduced-order physical simulation model in real time to evaluate the mechanical performance of the candidate parameter combination, and use the evaluation result as a heuristic feedback signal to input the constraint solver to guide the subsequent search direction. S5: When the constraint solver converges to a feasible solution that satisfies all static and dynamic constraints, the template skeleton is instantiated based on the feasible solution, and a parameterized model of the fixture and the corresponding solution certificate are generated and output.

[0008] As a preferred embodiment of the fixture generation method based on parameterized templates and intelligent reasoning described in this invention, step S1 specifically includes: Obtain the digital model of the workpiece and the machining process documents, and perform unified preprocessing on the data from cross-system sources in terms of format, coordinate system, naming rules and unit system; The workpiece digital model is subjected to topological analysis and manufacturing feature semantic recognition to extract workpiece geometric features directly related to fixture generation. The workpiece geometric features include reference plane, reference hole, surface to be machined, support sensitive surface and clamping action surface. Workpiece reference information is extracted synchronously. The reference information includes the correspondence between design reference, process reference and inspection reference, as well as the hierarchical relationship between primary reference, secondary reference and tertiary reference. Explicitly marked references are read directly, and references not explicitly marked are inferred. Design reference and process reference are distinguished and their priority order is recorded. Extract the process content from the processing technology document. The process content includes process number, processing location, processing sequence, processing type, number of clamping operations, processing allowance, and processing accuracy requirements. Tool path information is extracted from the machining process document, and the transformation relationship between the workpiece design coordinate system, the process clamping coordinate system and the machine tool machining coordinate system is established. The tool path information is mapped to the geometric surface corresponding to the workpiece digital model. The tool path information includes tool position coordinates, interpolation type, feed rate, spindle speed and tool posture data. The tool path information is segmented according to the machining area, cutting state and the degree of tool posture change. After extraction, a full data consistency check is performed. If there is a conflict between the toolpath coverage area and the candidate positioning surface, the corresponding candidate surface is remarked. If there is interference between the candidate clamping surface and the main approach direction of the tool, the corresponding candidate surface is remarked. Output a structured workpiece machining description dataset, including a feature list, a baseline relationship list, a process task list, a path segment list, and a coordinate mapping relationship table.

[0009] As a preferred embodiment of the fixture generation method based on parameterized templates and intelligent reasoning described in this invention, step S2 includes: Establish a node set, which includes workpiece feature nodes and fixture element placeholder nodes. The workpiece feature nodes are generated based on the workpiece geometric features extracted in step S1 and carry feature type, spatial position, normal direction, size range, surface condition, material properties, process association information and datum association information. The fixture element placeholder nodes include positioning element placeholders, clamping element placeholders, support element placeholders, connecting element placeholders and auxiliary element placeholders. An edge set is established, comprising positioning relationship edges, clamping relationship edges, support relationship edges, assembly relationship edges, and machining accessibility relationship edges. The positioning relationship edges characterize the datum correspondence and degree-of-freedom constraint relationship between workpiece feature nodes and positioning element placeholder nodes. The clamping relationship edges characterize the force relationship between clamping element placeholder nodes and workpiece feature nodes. The support relationship edges characterize the load-bearing contact relationship and deformation suppression relationship between support element placeholder nodes and workpiece feature nodes. The assembly relationship edges characterize the assembly connection constraints between fixture element placeholder nodes. The machining accessibility relationship edges characterize the spatial compatibility relationship between the tool movement space and the potential arrangement area of ​​fixture elements.

[0010] As a preferred embodiment of the fixture generation method based on parameterized templates and intelligent reasoning described in this invention, step S2 further includes: The set of nodes and the set of edges are given constraint attributes, which are divided into static constraint attributes and dynamic constraint attributes. The static constraint attributes represent fixed constraint requirements that are independent of the machining sequence, while the dynamic constraint attributes represent constraint requirements that are dynamically adjusted as the tool path segment changes. The constraints are set to a priority order, from highest to lowest: positioning relationship edge, processing accessibility relationship edge, clamping relationship edge, support relationship edge, and assembly relationship edge. The cutting load spectrum parsed from the tool path information in step S1 is encoded into the dynamic constraint conditions of the corresponding edge. During the encoding process, differentiated influence weights are set for path segments of different machining types. The deformation sensitivity weight of the finishing path segment is higher than that of the roughing path segment, and the load allowance weight of the roughing path segment is higher than that of the finishing path segment. The staged allowable deformation constraints are set according to the machining type.

[0011] As a preferred embodiment of the fixture generation method based on parameterized templates and intelligent reasoning described in this invention, step S3 includes: The system invokes a pre-defined parameterized template library, which includes multiple template entries. Each template entry is stored as a combination of functional description information, applicable condition information, structural topology information, parameter interface information, and constraint compatibility information, and adopts a three-level hierarchical organizational structure. Primary classification is based on workpiece reference pattern; Secondary classification is based on the main structural features of the workpiece; The three-level classification is based on the type of work process; Based on the structured workpiece machining description dataset output in step S1 and the typed constraint graph constructed in step S2, a template retrieval feature description is generated. The feature description includes the workpiece reference combination type, the number of main positioning features, the distribution of candidate clamping areas, the distribution of candidate support areas, the location of the area to be machined, the set of main tool approach directions, and the distribution of dynamic load sensitive areas. Template skeleton retrieval is performed using a combination of hard rule-based initial screening and multi-dimensional similarity comparison: The initial screening of rigid rules includes the consistency judgment of template reference mode and workpiece reference combination type, the coverage of template positioning unit quantity and workpiece degree of freedom restriction requirements, and the avoidance of the template base layout area from the main processing area and the main approach direction of the tool. Multidimensional similarity comparison is evaluated from two dimensions: node layer and edge layer. The comprehensive matching scores are sorted from high to low, and the top few template skeletons are selected to form a candidate template skeleton set.

[0012] As a preferred embodiment of the fixture generation method based on parameterized templates and intelligent reasoning described in this invention, step S3 further includes: Perform a fit check on each template skeleton in the candidate template skeleton set. The check includes: The compatibility between the number of template positioning units and the number of positioning degrees of freedom required by the workpiece; The compatibility between the working area of ​​the template clamping element and the allowable clamping area of ​​the workpiece; The coverage of the template support unit arrangement area with the dynamically load-sensitive area; Interference check between the template element contour envelope and the main approach direction of the tool; The template skeleton that passes the adaptability check is then mapped to the parameter interface with the typed constraint graph constructed in step S2: Map the positioning element placeholder nodes in the typed constraint diagram to the positioning unit interfaces of the template skeleton; Map the clamping element placeholder node to the clamping unit interface; Map the support element placeholder nodes to the support unit interfaces, and attach the constraint attributes corresponding to each relation edge to the corresponding interface; Based on the specific dimensions and process requirements of the current workpiece, the initial value range of the parameters in the template skeleton is narrowed and limited. If the placeholder node in the constraint diagram has no corresponding interface in the template skeleton, an additional interface is added according to the preset expansion rules. If there is a redundant interface in the template skeleton and there is no corresponding placeholder node in the constraint diagram, the interface is temporarily frozen and not included in the subsequent solution variable set.

[0013] As a preferred embodiment of the fixture generation method based on parameterized templates and intelligent reasoning described in this invention, step S4 includes: Using the typed constraint graph output in step S2 and the template skeleton output in step S3 as input, the adjustable parameters in the template skeleton are defined as solution variables, the range of values ​​of the adjustable parameters is defined as the search domain, and the constraint conditions carried by each edge in the typed constraint graph are mapped to the constraint conditions of the constraint solver. The solution variables include continuous variables and discrete variables. Candidate parameter combinations are generated. Before each candidate parameter combination enters the mechanical performance evaluation, static constraint pre-verification is performed. The verification includes whether the positioning unit completely constrains the six spatial degrees of freedom of the workpiece, whether the clamping action direction is compatible with the normal of the workpiece surface, whether the support unit falls into the allowable support area, whether there is spatial interference between the fixture elements, and whether the fixture elements intrude into the main approach direction area of ​​the tool.

[0014] As a preferred embodiment of the fixture generation method based on parameterized templates and intelligent reasoning described in this invention, step S4 further includes: For candidate parameter combinations that have passed the static constraint pre-verification, a reduced-order physical simulation model is called in real time to evaluate mechanical performance. The reduced-order physical simulation model is a fast evaluation model obtained by reducing the order of a high-fidelity finite element analysis model in the offline stage. The mechanical performance evaluation combines the dynamic constraint conditions in the typified constraint diagram to perform segmented evaluation of the candidate parameter combinations at the path segment level throughout the entire processing cycle. The mechanical performance evaluation results are transformed into multi-dimensional heuristic feedback signals. These heuristic feedback signals include the comprehensive feasibility level of candidate parameter combinations, failure cause categories, and sensitive parameter direction information. These signals are input into the constraint solver to guide the subsequent search direction. After receiving the feedback signals, the constraint solver performs three types of search optimization actions: local parameter rearrangement, candidate branch pruning, and parameter boundary shrinkage. When the solver finds a feasible parameter combination that satisfies all static and dynamic constraints, and no better combination with an improvement in the overall feasibility score exceeding a preset threshold appears in several consecutive iterations, the solution is deemed to have converged, and the feasible parameter combination, mechanical performance evaluation results, constraint satisfaction status records, and heuristic search trajectory information are output.

[0015] As a preferred embodiment of the fixture generation method based on parameterized templates and intelligent reasoning described in this invention, step S5 specifically includes: Perform an eventual consistency check on the feasible solution output in step S4. The check includes whether the parameter values ​​are within the preset range of the template skeleton, whether all static and dynamic constraints in the typed constraint diagram are satisfied, and whether there are any contradictions between the parameter values ​​and the mechanical performance evaluation results. The feasible solution that has passed the verification is instantiated into a template skeleton based on its parameter values. The continuous parameters in the parameter values ​​directly drive the geometric update of the model. The discrete parameters in the parameter values ​​are instantiated by retrieving the standard part model of the corresponding specification from the preset standard fixture component library to generate the fixture parameterized model. A solution certificate is generated synchronously. The solution certificate includes scheme identification information, constraint satisfaction status list, mechanical performance verification report, solution process traceability information and final verification conclusion. Before output, the fixture parameterized model is associated and bound with the solution certificate. The association and binding includes embedding the unique identifier of the solution certificate in the fixture parameterized model file and recording the hash verification value of the fixture parameterized model file in the solution certificate.

[0016] Secondly, embodiments of the present invention provide a fixture generation system based on parameterized templates and intelligent reasoning, comprising: Data extraction module: acquires the digital model of the workpiece and the machining process document, and extracts the geometric features, datum information, process content and tool path information of the workpiece; Constraint graph construction module: Constructs a typed constraint graph, in which nodes represent workpiece features and fixture element placeholders, and edges represent positioning relationships, clamping relationships, support relationships, assembly relationships and machining accessibility relationships, and encodes the cutting load spectrum parsed from the toolpath information as the dynamic constraint conditions of the edges; Template retrieval module: Retrieves a template skeleton that matches the workpiece digital model from a preset parametric template library. The template skeleton defines the type, layout topology, and adjustable parameter value range of the fixture. Constraint Solving Module: Performs constraint satisfaction solving on the template skeleton. During the solving process, for each candidate parameter combination, a pre-built reduced-order physical simulation model is called in real time to evaluate the mechanical performance of the candidate parameter combination, and the evaluation result is used as a heuristic feedback signal to input the constraint solver to guide the subsequent search direction. Instantiation output module: When the constraint solver converges to a feasible solution that satisfies all static and dynamic constraints, the template skeleton is instantiated based on the feasible solution, and the fixture parameterized model and corresponding solution certificate are generated and output.

[0017] The beneficial effects of this invention are as follows: By constructing a typified constraint graph model that integrates geometric datum constraints, process clamping constraints, and machining dynamic load constraints, and encoding the cutting load spectrum as dynamic constraint conditions of the constraint edges, this invention enables the fixture scheme to simultaneously meet static assembly requirements and dynamic mechanical requirements throughout the entire machining cycle during the generation process. This fundamentally solves the iterative problem caused by the separation of geometric design and mechanical verification in traditional methods. By using a reduced-order physical simulation model as an embedded evaluation module of the constraint solver, and generating heuristic feedback signals from real-time mechanical evaluation results to guide the search direction, this invention achieves a "solve-evaluate-" process. The "feedback-resolve" closed-loop optimization mechanism significantly reduces the number of invalid candidate solutions evaluated, improves solution efficiency and the engineering feasibility of the final solution. By retrieving the parameterized template skeleton and generating a candidate set, the unbounded global structure search is transformed into local parameter optimization within the constrained template skeleton, taking into account both the reuse of mature design experience and solution flexibility. By outputting a dual-result structure of "fixture parameterized model" and "solution certificate", the fixture solution is provided with traceable constraint satisfaction proof and mechanical performance verification certificate, meeting the requirements of high-end manufacturing for design compliance and quality traceability. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the 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, wherein: Figure 1 The first embodiment of the present invention provides an overall flowchart of a fixture generation method based on parameterized templates and intelligent reasoning; Figure 2 This is a module connection diagram of a fixture generation system based on parameterized templates and intelligent reasoning, provided for the third embodiment of the present invention. Detailed Implementation

[0019] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0020] Example 1, referring to Figure 1 This is one embodiment of the present invention, which provides a fixture generation method based on parameterized templates and intelligent reasoning.

[0021] S1: Obtain the digital model of the workpiece and the machining process file, and extract the workpiece's geometric features, datum information, process content, and tool path information.

[0022] This step involves uniformly analyzing and standardizing multi-source data on the workpiece and process, transforming it into structured information required for subsequent fixture generation. This provides foundational data for subsequent positioning, clamping, support, and machining analysis. Without this standardized analysis, the nodes, edges, matching conditions, and load conditions of the subsequent constraint diagrams would not be effectively supported, leading to inconsistencies between the generated solution and actual machining requirements.

[0023] This step first involves acquiring the workpiece digital model and machining process documents. The workpiece digital model includes structured model files such as 3D CAD models, MBD (Model-Based Definition) models with dimensional tolerance annotations, and component models containing assembly relationships. The machining process documents include one or more combinations of process cards, operation cards, CNC programs, toolpath files, and MES (Manufacturing Execution System) manufacturing execution files containing machining sequences and parameters. Data acquisition can be achieved through local file reading, interface calls to the design and manufacturing system, or retrieval from the PDM (Product Data Management) system. Cross-system data requires unified format and coordinate system preprocessing. In case of automatic reading errors, manual verification and correction are performed through an interactive interface, and the correction results are considered valid input.

[0024] After data preprocessing, workpiece geometric features directly related to fixture generation are extracted, rather than the geometric information of the entire workpiece surface. The extracted geometric features include: datum planes, datum holes, outer contour surfaces, inner cavity surfaces, surfaces to be machined, machined surfaces, blank retention surfaces, support-sensitive surfaces, clamping surfaces, and typical manufacturing features such as holes, slots, cavities, and outer circles.

[0025] The specific extraction process is as follows: ① The topological analysis of the workpiece model is divided into basic geometric units such as faces, edges, vertices, and feature volumes; ② Geometric units are classified according to preset rules (area threshold, flatness, normal consistency, hole axis, curvature continuity, adjacency relationship, geometric closure); ③ Semantic recognition is completed based on the manufacturing feature rule library, and the geometric units are mapped to positioning, clamping, support, and processing features, avoiding template matching and constraint modeling from merely remaining at the mechanical correspondence at the geometric level.

[0026] Typical feature recognition rule examples: When the area of ​​a plane is greater than 15% of the maximum projected area of ​​the workpiece, and the plane does not belong to the area to be removed in the current process, it is marked as a candidate support surface; when the axial dimension of a cylindrical hole is stable and the positional tolerance requirement is no greater than 0.1mm, it is marked as a candidate positioning hole; when the surface roughness is no greater than Ra1.6μm and it is the datum for the main dimensional tolerance, its priority as a candidate datum surface is increased.

[0027] Workpiece datum information is extracted synchronously. Datum information is the reference for the entire process of workpiece design, inspection and processing. It includes the correspondence between design datum, process datum and inspection datum, as well as the hierarchical relationship between primary datum, secondary datum and tertiary datum.

[0028] The datum information is identified using multi-source data: explicitly marked datums are read directly; for unmarked datums, inferences are made based on the dimensional chain convergence position, process start surface, inspection reference surface, and clamping priority positioning object. During extraction, design datums and process datums are distinguished and their priorities are recorded. For multiple selectable datums, priority is determined according to process sequence, tolerance sensitivity, and process reuse requirements.

[0029] By extracting the reference information in a standardized manner, the subsequent fixture design can be synchronized with the design intent and process intent of the workpiece, avoiding the situation where the geometry is clampable but the reference transfer chain is unreasonable, thus ensuring the consistency between the fixture design and the workpiece machining quality requirements.

[0030] After completing the feature and datum extraction, the process content of the current process is extracted. The process content is used to clarify the processing object and technical requirements of the current processing stage of the workpiece. The extracted fields include: process number, process name, processing location, processing sequence, processing type, number of clamping operations, processing equipment information, processing allowance, processing accuracy requirements, status of previous process, and requirements of subsequent process. Among them, the processing accuracy requirements cover dimensional tolerance, geometric tolerance, and surface roughness requirements.

[0031] Machining types are categorized into three types based on machining accuracy and material removal: roughing, semi-finishing, and finishing. For vertical milling of steel parts, the definition is as follows: roughing refers to a single-pass depth of cut (ap) > 1 mm; semi-finishing refers to a single-pass depth of cut (ap) of 0.2 mm to 1 mm; and finishing refers to a single-pass depth of cut (ap) < 0.2 mm. This classification provides a clear process basis for subsequent dynamic load analysis and stiffness requirement matching.

[0032] When extracting process content, in addition to identifying the processing area and accuracy requirements of the current process, it is also necessary to identify the surface area that will serve as the reference surface, limiting surface, and avoidance surface for the next process after the current process is completed. This type of information will directly determine the layout logic of subsequent fixture positioning, clamping, and support elements.

[0033] Simultaneously extract tool path information for the current process, including data such as path position, feed direction, cutting parameters, and tool posture.

[0034] The specific extraction process is as follows: The CNC program is parsed segment by segment, reading the tool position coordinates, interpolation type, feed parameters, spindle parameters, and tool call information. A structured tool path sequence is formed according to the machining sequence or path order. The interpolation types include linear interpolation, circular interpolation, and rapid positioning. The stroke type is distinguished based on the feed rate: when the feed rate is >5000mm / min, it is determined to be a rapid positioning stroke; when the feed rate is ≤5000mm / min, it is determined to be an effective cutting stroke. This threshold can be adjusted according to the machine tool's rated parameters.

[0035] To facilitate the subsequent construction of the cutting load spectrum, the toolpath needs to be segmented. The segmentation is based on the machining area, cutting state, and degree of tool attitude change. A significant attitude change is defined as an angle greater than 5° between two adjacent tool positions, and a stable cutting path segment is defined as more than 20 consecutive tool positions corresponding to the same machining surface with a feed direction change of less than 10°. Through path segmentation, the original toolpath data can be organized into standardized path segment data suitable for subsequent dynamic constraint modeling.

[0036] While extracting toolpath information, it is necessary to establish the transformation relationship between the workpiece design coordinate system, the process clamping coordinate system, and the machine tool machining coordinate system to ensure that the toolpath information can be accurately mapped to the corresponding geometric surface of the workpiece digital model. In practice, the toolpath coordinate points are projected onto the workpiece geometric surface, and the surfaces to be machined, adjacent avoidance surfaces, and potential interference areas corresponding to each path segment are identified. This achieves a one-to-one correspondence between process content, machining location, and geometric features, providing a precise spatial basis for the subsequent association between the cutting load spectrum and constraint relationship edges.

[0037] After extracting the above four types of information, a full data consistency check must be performed to confirm that there are no logical conflicts between the region to which the geometric feature belongs, the processing object of the process, and the area covered by the tool path.

[0038] The criteria for defining obvious conflicts are: ① The toolpath coverage area overlaps with the candidate positioning surface, and the overlapping area accounts for more than 10% of the area of ​​the candidate positioning surface; ② The candidate clamping surface is located within 20mm of the main tool approach direction, posing a risk of tool interference. If the above conflicts exist, the corresponding candidate positioning surface will be remarked as a surface affected by machining, and the candidate clamping surface will be remarked as a surface restricted by accessibility, and their optional states in subsequent template matching and constraint modeling will be updated simultaneously.

[0039] This step ultimately outputs a structured workpiece machining description dataset for intelligent fixture generation, which includes at least: a feature list, a datum relationship list, a process task list, a path segment list, and a coordinate mapping relationship table. This dataset transforms multi-source, heterogeneous original design and manufacturing data into unified, standardized input, providing a complete data foundation for the subsequent construction of S2 typed constraint graphs.

[0040] Compared to the conventional simple data preparation that only reads the model and process card, the core value of this step lies in the simultaneous completion of semantic parsing and standardization of the original data. This provides a unified data basis for the subsequent generation of constraint graph nodes, construction of relation edges, and setting of dynamic load conditions. It is the core foundation for ensuring the accuracy of subsequent template retrieval, the stability of the solution process, and the consistency between the final fixture scheme and the actual working conditions.

[0041] S2: Construct a typed constraint graph, in which nodes represent workpiece features and fixture element placeholders, and edges represent positioning relationships, clamping relationships, support relationships, assembly relationships and machining accessibility relationships. The cutting load spectrum parsed from the tool path information is encoded as the dynamic constraint conditions of the edges.

[0042] The core action of this step is to construct a typified constraint graph based on the structured workpiece machining description dataset output by S1. This constraint graph uses nodes to represent workpiece features and fixture element placeholders, and directed edges to represent five types of constraint relationships: positioning, clamping, support, assembly, and machining accessibility. At the same time, the cutting load spectrum obtained by parsing the tool path information is encoded as the dynamic constraint conditions of the corresponding edges.

[0043] It should be noted that this step is the core link connecting front-end data parsing and back-end constraint solving. Its core function is not simply to establish a graphical representation structure, but to uniformly map the multi-source heterogeneous constraint information scattered in the workpiece model, process documents, and toolpaths in S1 into a structured constraint carrier that can directly participate in template retrieval, constraint satisfaction solving, and mechanical performance evaluation. The typed constraint graph constructed in this step, unlike conventional adjacency graphs or assembly connection graphs, is a composite relationship model that simultaneously carries geometric datum constraints, process clamping constraints, component combination constraints, tool motion constraints, and dynamic cutting load constraints. It can incorporate all the positioning accuracy, clamping stability, support stiffness, assembly feasibility, and machining accessibility requirements that the fixture design needs to meet into the same solution framework, completely solving the industry pain point of the separation of geometric analysis, process analysis, and mechanical verification in traditional fixture design, and the repeated iterative backtracking.

[0044] This step is implemented sequentially in five sub-steps, S21 to S25, as detailed below: S21: Establishing the Node Set The node set of the constraint diagram is divided into two main categories: workpiece feature nodes and fixture element placeholder nodes. All nodes carry a unique identifier (ID) and a complete set of attributes, which provides the foundation for the subsequent establishment of constraint edge relationships.

[0045] Based on the feature list and datum relationship list output by S1, a corresponding workpiece feature node is generated for each valid feature. The node attribute set includes: unique feature ID, feature type, spatial position and normal vector, size range, surface condition, material properties, process association information, datum association information, and optional status flags. The feature type corresponds one-to-one with the S1 extraction results, and is subdivided into six categories: datum plane features, datum hole features, support candidate features, clamping action features, machining-affected features, and avoidance-sensitive features. It should be noted that different feature types correspond to different functional roles in fixture design: datum plane / datum hole features are used for datum transfer, support candidate features are used for load bearing, clamping action features are used for applying clamping force, and machining-affected / avoidance-sensitive features are used for filtering restricted areas in the fixture component placement area. Through attribute predefinition, a calling basis can be directly provided for subsequent edge relationship establishment, reducing repetitive judgment logic and improving constraint graph construction efficiency.

[0046] Placeholder nodes for fixture components refer to intermediate representations of fixture functional units whose functional roles, layout categories, and parameter interface information are currently defined, but whose specific standard part models, geometric dimensions, and assembly postures have not yet been determined. It should be noted that by setting placeholder nodes, subsequent template retrieval and constraint solving can first complete the constraint relationship construction at the functional layer, and then complete the component instantiation at the parameter layer, significantly improving the flexibility and scalability of fixture generation. Placeholder nodes are generated based on preset fixture function classification rules, and can also be pre-configured by combining the structural definition of the S3 template skeleton to be retrieved. The preset fixture function classification rules divide fixture functional units into five categories, corresponding to five types of placeholder nodes: positioning component placeholders, clamping component placeholders, support component placeholders, connecting component placeholders, and auxiliary component placeholders.

[0047] Furthermore, the specific pre-configuration logic for placeholder nodes is as follows: based on the number of workpiece features, process datum level, clamping times, processing area distribution, and processing type, the minimum number of positioning units, clamping units, and support units required for the current process is initially determined, and corresponding placeholder nodes are generated for each functional unit. For example, when the workpiece needs to constrain all degrees of freedom using the six-point positioning principle, a positioning element placeholder node matching the six-point positioning logic is generated; for thin-walled steel structural parts, thin-walled stress-sensitive areas with a wall thickness of no more than 3mm and a length-to-thickness ratio greater than 10, if they are subjected to periodic cutting loads during processing, the number of support element placeholder nodes needs to be increased accordingly, reserving interfaces for subsequent additional support unit configurations. Through pre-configuration, the pre-matching of fixture functional objects and workpiece process requirements in the constraint diagram at the quantity and role levels can be achieved.

[0048] S22: Establishment of the constraint edge set After the node set is constructed, a corresponding set of directed edges is established to represent the constraint relationships between nodes. The complete edge set includes five categories: positioning relationship edges, clamping relationship edges, support relationship edges, assembly relationship edges, and processing accessibility relationship edges. It should be noted that the five types of edges are not independent of each other, but together constitute the complete constraint network of the fixture design, and all constraint relationships must participate in the subsequent solution process simultaneously.

[0049] The positioning relationship edge is used to characterize the datum correspondence and degree-of-freedom constraint relationship between the workpiece feature nodes and the positioning element placeholder nodes. The construction logic is as follows: based on the datum attributes, normal direction, spatial distribution, and degree-of-freedom constraint requirements of the workpiece feature nodes, the association between the feature nodes and the corresponding placeholder nodes is completed. If the workpiece feature node corresponds to the main datum plane, a main positioning relationship edge is established between it and the positioning element placeholder node that performs the first positioning function. The edge attributes record the number of contact points, contact method, direction of degree-of-freedom constraint, and positioning allowable deviation range. The positioning allowable deviation range is defined as the offset of the positioning contact point relative to the theoretical contact position not exceeding 0.05mm. If the workpiece feature node corresponds to the datum hole / datum shaft, an auxiliary positioning relationship edge is established between it and the positioning element placeholder node that performs the lateral limiting function. The edge attributes record the limiting direction and repeatability accuracy requirements. The repeatability accuracy requirements are defined as the positioning deviation after repeated clamping of the same batch of workpieces not exceeding 0.02mm. It should be noted that the combination of freedom constraints formed by all positioning relationships must meet the six-point positioning principle to ensure that the six spatial degrees of freedom of the workpiece are fully constrained, avoiding workpiece position uncertainty caused by under-constraint, or clamping interference and assembly difficulties caused by over-constraint.

[0050] Clamping relationship edges characterize the force relationship between clamping element placeholder nodes and workpiece feature nodes. The construction logic is as follows: based on the clamping actionable attributes, local stiffness attributes, and surface condition of the workpiece feature nodes, combined with the cutting direction and cutting intensity of the current process, node association is completed. The core attribute set of clamping relationship edges includes clamping action direction, clamping action area, clamping function category, clamping sequence, and limiting conditions. The clamping function category is divided into three types: main clamping, auxiliary clamping, and anti-runaway clamping. Limiting conditions are used to mark whether the clamping action is restricted by the machining area, thin-walled area, heat-sensitive area, or avoidance area. Furthermore, risk marking rules are set simultaneously: if the clamping actionable surface is within 20mm of the boundary of the area to be machined, the clamping relationship edge is marked as a locally limited edge; if the local wall thickness of the area corresponding to the clamping action surface is no greater than 2mm, the clamping relationship edge is marked as a deformation-sensitive edge. Through risk classification, the safety and stability of the clamping area can be prioritized in subsequent solutions, avoiding the selection of clamping positions solely from a geometric contact perspective.

[0051] The support relationship edge is used to characterize the load-bearing contact relationship, local stiffness enhancement relationship, and deformation suppression relationship between the placeholder nodes of the support element and the feature nodes of the workpiece. The construction logic is as follows: based on the workpiece's self-weight distribution, clamping posture, cutting load application area, and local stiffness distribution, node association is prioritized in feature areas with high local stiffness within the gravity projection range, near the cutting load transmission path, and within the area of ​​high local stiffness. It should be noted that, in addition to the static support contact relationship, the support relationship edge also has pre-set dynamic requirement attributes for vibration suppression and deformation compensation. For example, if the cutting load direction of a certain machining area changes more than 10 times per second in the corresponding path segment, and the distance between this area and the adjacent support area is greater than 50mm, an additional local support relationship edge needs to be established to match the dynamic stiffness enhancement requirement of this area.

[0052] Assembly relationship edges are used to characterize the assembly connection constraints between fixture component placeholder nodes and between placeholder nodes and the basic installation structure. They primarily address the installability, adjustability, detachability, and resettlement of the fixture itself. Their core attribute set includes component connection method, installation sequence, relative position restrictions, adjustable direction, adjustment stroke, and reserved space requirements. Connection methods include bolt connections, keyway connections, T-slot connections, and pin connections. Typical constraint examples: When the installation area of ​​the positioning component placeholder node overlaps with the clamping component placeholder node in terms of assembly space, the assembly relationship edge must record the sequence constraint of "installing the positioning component first, then the clamping component"; when the support component placeholder needs to have height adjustable functionality, the assembly relationship edge must record the adjustment attribute of "adjustable in the vertical direction, with an adjustment stroke of 5mm~20mm".

[0053] The machinability accessibility relationship edge is used to characterize the spatial compatibility relationship between the tool movement space and the potential arrangement area of ​​fixture elements. Its core function is to explicitly incorporate the machinability requirements into the constraint model, avoiding problems such as tool interference and insufficient cutting space after the fixture scheme is generated. The specific construction logic is as follows: Based on the path segment list and coordinate mapping relationship output by S1, a corresponding tool influence region is established for each path segment. The tool influence region is defined as the spatial swept body formed by the tool envelope (the maximum contour rotation of the tool cutting part + the tool holder part) moving along the tool path, superimposed with a 5mm safety margin to form a closed area; the tool influence region is spatially compared with the potential fixture arrangement area around the workpiece feature nodes, and risk classification is completed according to the minimum spacing: minimum spacing < 20mm is marked as a high interference risk edge, 20mm ≤ minimum spacing ≤ 50mm is marked as a medium interference risk edge, and minimum spacing > 50mm is marked as a low interference risk edge. Through risk classification, the subsequent solver can be guided to prioritize avoiding high interference risk areas, improving search efficiency.

[0054] S23: Constraint Attribute Assignment and Priority Setting After the edge set is established, complete constraint attributes are attached to all nodes and edges, which are divided into two main categories: static constraint attributes and dynamic constraint attributes. Static constraint attributes represent fixed constraint requirements that are independent of the machining time sequence, including geometric dimension constraints, assembly relationship constraints, positioning accuracy constraints, initial clamping requirements, and no-go zone avoidance constraints, which remain unchanged during the constraint solution process. Dynamic constraint attributes represent changing constraint requirements that are related to the machining process time sequence, including cutting load constraints, vibration suppression constraints, thermal deformation constraints, and path time sequence constraints, which are dynamically adjusted as the tool path segment changes.

[0055] Furthermore, a pre-defined constraint priority rule is implemented to address the problem of prioritizing solutions when multiple types of constraints conflict simultaneously. The constraint priorities, from highest to lowest, are: positioning relationship edges, machining reachability relationship edges, clamping relationship edges, support relationship edges, and assembly relationship edges. By setting the priority, it is ensured that when local conflicts occur during the solution process, the core requirements of positioning accuracy and machining non-interference are prioritized, thus ensuring the basic usability of the final solution.

[0056] S24: Dynamic Constraint Connection and Cutting Load Spectrum Encoding It should be noted that the core innovation of this step lies in encoding the cutting load spectrum as the dynamic constraint conditions of the corresponding edge, so that the dynamic machining load is no longer additional information for subsequent separate verification, but a core component of the constraint diagram itself. This is also one of the key features that distinguishes this solution from traditional fixture design methods.

[0057] The cutting load spectrum is not simply a list of force values, but a collection of information on load intensity, triaxial load direction, location of application, load change rate, duration, and corresponding process stage along the toolpath segment. The cutting load is calculated using a pre-defined cutting force prediction model, with two feasible implementation methods: The first is an analytical model based on material cutting energy, employing the classic Kienzle cutting force formula. Inputs include the depth of cut, cutting width, feed rate, spindle speed, tool geometry, and workpiece material properties, calculating the triaxial components and load range of the cutting force. The second is a mapping model based on experimental data, pre-building a database of cutting force experiments for different workpiece material-tool type-cutting parameter combinations, and querying the load range for the corresponding toolpath segment through parameter matching.

[0058] Furthermore, after the calculation is completed, the load information of each path segment is mapped to the clamping relationship edge, support relationship edge, and machining accessibility relationship edge associated with that path segment, thus completing the edge-level dynamic constraint connection. For example, if a machining path segment acts on the side wall area of ​​the workpiece, and the load direction is mainly towards the inside of the workpiece, then the load information of this path segment is preferentially encoded to the clamping relationship edge and support relationship edge adjacent to that side wall, indicating that this area needs to be matched with higher clamping holding force and support stiffness.

[0059] During the assembly process, differentiated influence weights are assigned to path segments of different machining types to match the core requirements of different processes: The finishing path segment is assigned a high deformation sensitivity weight, 2-3 times that of the roughing path segment, primarily controlling the deformation of the machining area; the roughing path segment is assigned a high load tolerance weight, 1.5-2 times that of the finishing path segment, primarily controlling the overall stiffness and impact resistance of the fixture system. Furthermore, the machining type classification of S1 is synchronized, setting staged allowable deformation constraints: the allowable deformation in the finishing stage is no more than 1 / 3 of the corresponding machining feature dimension tolerance, in the semi-finishing stage no more than 1 / 2, and in the roughing stage no more than 2 / 3. Through this graded weighting, the constraint solution fully conforms to the actual machining process requirements, rather than applying a uniform evaluation standard to all machining states.

[0060] S25: Figure 1 Consistency Correction and Conflict Check After the initial construction of the constraint graph is completed, a full consistency correction and edge conflict check are performed to ensure that the constraint graph is logically consistent and to avoid subsequent solution failures caused by front-end construction conflicts.

[0061] First execute Figure 1 Consistency correction checks for issues such as missing nodes, omitted edges, conflicting edge attributes, incorrect edge directions, and duplicate relationships between nodes, and cleans up and supplements invalid nodes and edges. Then, edge conflict checks are performed, focusing on two types of core conflicts: first, the same workpiece feature node is simultaneously marked as a high-priority positioning feature and a high-risk feature that affects processing; second, the same fixture element placeholder node simultaneously undertakes contradictory main positioning and avoidance functions.

[0062] If the above conflicts exist, the edge attributes will be rearranged or deleted according to the preset baseline priority, process priority, processing accessibility priority, and dynamic load sensitivity priority. If the conflict cannot be automatically corrected, the operator will be prompted through the interactive interface to adjust the input conditions and rebuild the constraint diagram.

[0063] This step ultimately outputs a complete typed constraint graph dataset, containing at least a list of nodes, an edge list, node attribute sets, static constraint attribute sets, and dynamic constraint attribute sets. Each edge has a clearly defined relationship type and constraint definition, forming a standardized constraint network that can be directly used for subsequent solutions. It should be noted that this step completes the core transformation from "structured data description" to a "computable solution model." Unlike the discretized model of geometric matching, rule filtering, and experience-based point selection in traditional fixture design, this step, for the first time, integrates five major categories of constraints—positioning, clamping, support, assembly, and machining accessibility—along with dynamic cutting load constraints into the same graph model. This provides a unified solution framework for subsequent S3 template retrieval and S4 constraint satisfaction solutions, and is a crucial step in ensuring the completeness, interpretability, and engineering feasibility of the final fixture solution.

[0064] S3: Retrieve a template skeleton that matches the digital model of the workpiece from a preset parametric template library. The template skeleton defines the type, layout topology, and adjustable parameter range of the fixture.

[0065] The core action of this step is: taking the structured workpiece machining description dataset output by S1 and the typed constraint graph constructed by S2 as input, a matching template skeleton is retrieved from the preset parametric template library; the template skeleton is used to define the functional type, layout topology and adjustable parameter value range of the fixture, providing a controlled search starting point and parameter boundary for the subsequent constraint satisfaction solution in S4.

[0066] It should be noted that the core function of this step is not to select a fixture solution with a similar shape from the template library, but to generate a pre-selected and limited basic structure for the fixture from three dimensions: functional layer, relational layer, and parameter layer. This transforms the original unbounded global structure search problem into a local parameter optimization problem within a constrained template skeleton. This not only reuses mature fixture design experience that has been verified in engineering, but also retains the flexibility of the solution. It can significantly reduce the solution dimension, improve convergence stability, and reduce invalid candidate solutions caused by unreasonable initial structures.

[0067] This step is implemented sequentially in five sub-steps, S31 to S35, as detailed below: S31: Preset and Hierarchical Organization of Parametric Template Libraries The parametric template library consists of multiple template entries, each corresponding to a reusable fixture structure prototype. Template entries are not stored solely as geometric models, but rather as a combination of functional descriptions, applicable conditions, structural topology information, parameter interface information, and constraint compatibility information. They also include pre-defined, engineering-verified fixed parameter values ​​and constraint relationships. The functional description information is used to characterize the fixture function category corresponding to the template, including positioning and clamping type, support enhancement type, hole system positioning type, hole combination positioning type, thin wall compensation type, and multi-station reuse type; the applicable condition information is used to characterize the workpiece category, datum type, process type, processing method, and structural feature range that the template is compatible with; the structural topology information is used to characterize the connection relationship and layout logic of the positioning, clamping, support, connection, and auxiliary units within the template; the parameter interface information is used to characterize the adjustable parameters such as position, direction, size, stroke, and installation angle of each component; the constraint compatibility information is used to characterize the template's compatibility with five types of constraints: positioning, clamping, support, assembly, and processing accessibility; the fixed parameter values ​​and fixed constraint relationships are fixed values ​​and rules determined according to mature design specifications, which do not need to be searched repeatedly in subsequent solutions, and can directly reduce the number of solution variables.

[0068] Furthermore, the parametric template library adopts a three-tiered hierarchical organizational structure to improve retrieval efficiency: The first-level classification is based on the workpiece's datum mode, divided into surface datum-dominated templates, hole datum-dominated templates, mixed surface datum templates, and complex curved surface support templates; the second-level classification is based on the workpiece's main structural features, divided into plate-type part templates, frame-type part templates, box-type part templates, bushing-type part templates, and thin-walled shell-type part templates; the third-level classification is based on the process type, divided into roughing templates, semi-finishing templates, finishing templates, and composite machining templates. The index can be supplemented and expanded according to the type of processing equipment and clamping method, with the core service being the rapid positioning of the current workpiece task, avoiding invalid comparisons in obviously incompatible templates.

[0069] S32: Template Retrieval Feature Description Generation Before performing template retrieval, a template retrieval feature description is generated based on the output results of S1 and S2. The retrieval features no longer rely on a single geometric similarity, but simultaneously cover the full-dimensional requirements of workpiece positioning, clamping, support, and machining avoidance.

[0070] The core content of the retrieval feature description includes: workpiece datum combination type, number of main positioning features, auxiliary positioning feature type, distribution of candidate clamping areas, distribution of candidate support areas, location of the area to be processed, set of main tool approach directions, area affected by the processing path, workpiece posture requirements, distribution of local stiffness-sensitive areas, and distribution of dynamic load-sensitive areas. Furthermore, workpiece dimensions, length-width-height ratio, main opening direction, maximum projected profile, proportion of thin-walled areas, number density of holes and slots, surface quality grade, number of clamping operations, and the position of the current process in the process route are added to form a complete retrieval feature set.

[0071] S33: Template initial screening and multidimensional similarity comparison Template skeleton retrieval is performed by combining hard rule initial screening with multidimensional similarity comparison. First, obviously incompatible template entries are quickly removed through initial screening, and then the candidate templates that pass the initial screening are subjected to refined similarity evaluation.

[0072] The core criteria for initial screening using rigid rules include: the template datum mode and workpiece datum combination type are consistent; the number of positioning units supported by the template covers the workpiece's required degrees of freedom constraints; the allowed clamping direction of the template is coordinated with the main cutting direction of the current process; the template base layout area avoids the main processing area and the main tool approach direction; and the number of support units supported by the template meets the support requirements under the current load conditions. Simultaneously, supplementary rules are added for workpiece feature type matching, datum layout matching, processing type matching, and clamping frequency matching. Typical examples: If the current workpiece uses the bottom surface as the main positioning surface and the two sides as auxiliary positioning surfaces, template frames using one-sided two-pin or one-sided one-pin-one-stop positioning methods are preferred; if the current process is the first clamping process, template frames suitable for blank clamping and with a large clamping stroke are preferred; if the current process is the final finishing process, template frames suitable for fine datum positioning and with matching positioning accuracy levels are preferred.

[0073] Furthermore, multi-dimensional similarity comparisons are performed on the candidate templates that pass the initial screening. The comparison dimensions cover datum feature similarity, process task similarity, constraint diagram substructure similarity, and dynamic load distribution similarity. Quantitative evaluation is completed from two dimensions: node layer and edge layer. The node layer mainly compares the type, number, spatial distribution, and datum association method of the workpiece feature nodes. The edge layer mainly compares the connection patterns, constraint priorities, and dynamic constraint distribution of positioning, clamping, support, and machining accessibility relationship edges. Example: If a candidate template and the target workpiece are consistent in the number of datum planes, hole positioning method, and direction of the main machining area, and the connection patterns of positioning and clamping relationship edges in the constraint diagram match, the matching priority of this template is increased. If a candidate template has similar geometric dimensions, but its clamping elements are mainly arranged in the main tool approach direction area, its matching priority is decreased.

[0074] After similarity comparison, the template skeletons are sorted from high to low according to the comprehensive matching score, and the top 3-5 template skeletons are selected to form a candidate template skeleton set. It should be noted that the comprehensive matching score is calculated with weights of 40% for rule compatibility, 30% for constraint diagram similarity, 20% for dynamic constraint adaptation, and 10% for geometric similarity. If multiple candidate templates have small differences in score and all meet the initial screening conditions, a secondary sort can be performed based on historical success rate, average solution time, and average fixture cost. This sorting is only for optimization reference and does not change the fundamental premise that the template must meet the core constraints of the workpiece. The core function of retaining multiple candidate templates is to avoid subsequent solution failures due to excessive initial structural deviations of a single template. When the first candidate template cannot obtain a feasible solution in S4 due to dynamic constraint conflicts, the solution can be automatically switched to the next candidate template to continue solving, reducing manual intervention.

[0075] S34: Template skeleton compatibility check and parameter interface mapping After screening candidate templates, a suitability check is performed on each template skeleton to confirm that there are no significant mismatches between the template and the current workpiece and process requirements. This improves the quality of templates entering the S4 solution stage and avoids wasting solution resources. The core content of the suitability check includes: the number of positioning units defined by the template matches the number of positioning degrees of freedom required by the workpiece; the action area of ​​the template clamping elements falls within the workpiece's allowable clamping range; the template support unit arrangement area covers the dynamically load-sensitive area; the template assembly connection direction is coordinated with the tooling installation space; and the template element contour envelope does not intrude into the main approach direction of the tool. Example: If the template skeleton defines only 2 clamping units, but the current workpiece has more than 2 high-load-sensitive areas, it is marked as an insufficiently supported template; if the angle between the arrangement direction of the template's main clamping elements and the main approach direction of the finishing tool is less than 15°, it is marked as a high-interference-risk template.

[0076] The template skeleton that passes the adaptability check needs to complete the parameter interface mapping with the S2 typed constraint diagram, transforming the template skeleton from an independent library structure into an initial framework to be solved, bound to the current workpiece task. Specifically, the positioning element placeholder nodes in the typed constraint diagram are mapped to the positioning unit interface of the template skeleton, the clamping element placeholder nodes are mapped to the clamping unit interface, and the support element placeholder nodes are mapped to the support unit interface. Simultaneously, the constraint attributes corresponding to each relational edge are attached to the corresponding interface.

[0077] Furthermore, based on the specific dimensions and process requirements of the current workpiece, the initial range of parameter values ​​for the template skeleton is narrowed and limited. For example, if the diameter of the positioning hole on the current workpiece is 12mm, the range of values ​​for the positioning pin diameter in the template skeleton is narrowed to the interval that matches the positioning hole, reducing the invalid search range in subsequent solutions. It should be noted that if the placeholder node in the constraint diagram has no corresponding interface in the template skeleton, an additional interface can be automatically added to the template skeleton according to the preset extension rules. The extension rules limit the function type, placement boundary, and adjustable parameter range of the newly added interface. If the template skeleton has redundant interfaces and there is no corresponding placeholder node in the constraint diagram, the interface is temporarily frozen and not included in the subsequent solution variable set.

[0078] S35: Candidate Template Set Output and Exception Handling This step ultimately outputs a set of candidate template skeletons that match the requirements of the current workpiece and process. It also outputs the layout topology description, adjustable parameter boundaries, interface mapping relationships, fixed constraint relationships, and initial compatibility evaluation results for each template.

[0079] The template skeleton defines at least three core elements: First, the type of fixture function, including planar support with lateral limiting, hole positioning with pressure plate clamping, composite hole positioning, multi-point floating support, thin-walled compensation support, and modular combination fixture; second, the layout topology, i.e., the relative arrangement, connection sequence, and installation hierarchy of the functional units within the template, for example: the positioning unit is located at the bottom of the workpiece, the clamping unit is located above the workpiece, the support unit is located below the thin-walled area, and the connecting units are distributed along the edge of the base plate; third, the range of adjustable parameters, which are divided into continuous and discrete types. Continuous parameters include the adjustment range of the center position of the positioning element, the translation range of the action point of the clamping element, and the adjustment range of the height of the support element, etc. Discrete parameters include the model of the positioning element, the specification of the clamping element, and the form of the support head, etc. The range of all parameters is determined based on the template structure capability, standard component specifications, workpiece space allowance, and assembly and machining avoidance requirements.

[0080] If there is no template skeleton in the template library that directly matches the current workpiece task, a general modular template skeleton is called as the initial solution skeleton. The general template skeleton includes at least a base plate, configurable positioning unit interface, clamping unit interface, and support unit interface, and the parameter range covers the clamping requirements of conventional workpieces. If necessary, the operator can make local modifications to the general template skeleton through the interactive interface before entering the subsequent solution process, so as to ensure the adaptability of the solution to irregular parts and new process tasks.

[0081] It should be noted that this step plays a crucial role in transitioning from constraint modeling to the starting point of structure generation, serving as a core bridging link between S2 and S4. Unlike traditional fixture design that relies on subjective selection based on human experience, this step simultaneously incorporates workpiece characteristics, process requirements, and typified constraint diagrams into the retrieval process. This provides clear data and constraint basis for template selection, significantly improving the accuracy of initial structure selection, the convergence efficiency of subsequent solutions, and the engineering adaptability of the final fixture design.

[0082] S4: Perform constraint satisfaction solving on the template skeleton. During the solution process, for each candidate parameter combination, call the pre-built reduced-order physical simulation model in real time to evaluate the mechanical performance of the candidate parameter combination, and use the evaluation result as a heuristic feedback signal to input the constraint solver to guide the subsequent search direction.

[0083] This step is the core automated solution step implemented by the computer. The core inputs are the typed constraint graph output by S2, the candidate template skeleton set output by S3, and the pre-built and stored reduced-order physical simulation model library in a computer-readable medium. The core actions are: to perform constraint satisfaction solution on the template skeleton. During the solution process, for each set of candidate parameter combinations, the reduced-order physical simulation model is called in real time to complete the mechanical performance evaluation, and the evaluation results are converted into heuristic feedback signals to input into the constraint solver to guide the subsequent search direction.

[0084] It should be noted that the core innovation of this step differs from the existing technology in that it abandons the traditional serial mode of fixture design, which involves "first generating complete candidate solutions and then performing physical simulation verification separately". Instead, it uses a reduced-order physical simulation model as an embedded evaluation module of the constraint solver. Real-time mechanical evaluation is completed when each set of candidate parameter combinations is generated, forming a closed-loop mechanism of "solving-evaluation-feedback-resolving". This allows the solver to actively tend towards parameter regions with better mechanical performance, avoiding wasting computing power on a large number of mechanically infeasible solutions, and significantly improving the solution efficiency and the engineering feasibility of the final solution.

[0085] This step is implemented sequentially according to the five computer-executable steps S41 to S45, as detailed below: S41: Solver Input Initialization and Solver Framework Configuration This step is a preliminary preparation for the solution process, and it involves four core tasks: input data parsing, solution variable definition, constraint mapping, and solver framework configuration.

[0086] First, perform standardized parsing of the input data: Parse the typed constraint graph output by S2 and extract workpiece feature nodes, fixture element placeholder nodes, five types of constraint relationship edges, static constraint attribute set, and dynamic constraint conditions encoded into the edges; Parse the candidate template skeleton set output by S3 and extract the fixture type definition, layout topology description, adjustable parameter value range, continuous / discrete parameter classification identifier, fixed parameter value, fixed constraint relationship, and constraint diagram interface mapping relationship for each template; Load a pre-built library of reduced-order physical simulation models. Each template skeleton in the library corresponds to a pre-trained reduced-order simulation model, which can be called by the solver in real time.

[0087] Then complete the basic configuration of the solution framework: Variable definition: The adjustable parameters in the template skeleton are defined as solution variables. Continuous variables include the center position of the positioning element, the coordinates of the action point of the clamping element, and the height of the support element, while discrete variables include the model of the positioning element, the specification of the clamping element, and the form of the support head. Domain definition: Define the range of values ​​for the adjustable parameters output by S3 as the search domain of the corresponding solution variable; Constraint definition: The static and dynamic mechanical constraints carried by each edge of the typed constraint graph are fully mapped to the constraint conditions of the solver; Solver selection: Adopt a constraint satisfaction problem solving framework adapted to this scenario, including a CSP (Constraint Satisfaction Problem) solver based on backtracking search, a mixed integer programming solver based on branch and bound, and a metaheuristic solver based on genetic algorithm. Regardless of the framework used, the closed-loop solution logic of this step shall be followed. Constraint priority configuration: Strictly follow the priority rules set in S2, in descending order: positioning relationship edge constraint, processing accessibility relationship edge constraint, clamping relationship edge constraint, support relationship edge constraint, assembly relationship edge constraint, to ensure that the core availability of the solution is prioritized in the event of local conflicts.

[0088] S42: Candidate Parameter Combination Generation and Static Constraint Pre-verification This step uses the logic of "constraint pre-pruning - hierarchical variable assignment - static compliance verification" to generate candidate parameter combinations, eliminate obviously infeasible solutions in advance, and reduce the computing power consumption of subsequent simulation evaluation.

[0089] First, parameter domain pre-pruning is performed: based on the core constraints of positioning relationship edges, machining accessibility relationship edges, and assembly relationship edges, initial pruning is performed on the parameter interface and variable search domain of the template skeleton to eliminate parameter values ​​that obviously violate geometric occupancy, degree of freedom constraints, tool avoidance, and installation sequence requirements, thereby narrowing the solution search space.

[0090] Subsequently, candidate parameter combinations are generated according to a preset variable priority order: the variable assignment order follows the rule of "positioning parameters first → clamping parameters second → support parameters third → auxiliary parameters last," ensuring that the key parameters with the greatest impact on the feasibility of the solution are locked first in the early stage of the solution, alleviating the combinatorial explosion problem in the high-dimensional parameter space. After each set of full variable assignments is completed, a corresponding set of candidate parameter combinations is generated. This combination is the specific value scheme of all adjustable parameters under the current template skeleton, corresponding to a set of fixture configurations to be verified.

[0091] Each set of candidate parameter combinations must undergo static constraint pre-verification before entering the mechanical evaluation. The verification includes: Does the positioning unit completely constrain the workpiece's six spatial degrees of freedom, without under-constraint or over-constraint issues? The clamping direction is compatible with the normal direction of the workpiece surface, and the clamping area falls within the allowable clamping range; The support unit falls into the allowable support area, and there is no spatial interference between the fixture components; The fixture component outline does not intrude into the tool's main approach direction area, eliminating the risk of machining interference. The fixture mounting interface is compatible with the machine tool foundation mounting structure.

[0092] Combinations that fail pre-verification are directly deemed invalid. For example: if the spacing between positioning elements is less than 80% of the minimum allowable spacing of the template, it is judged as a positioning redundancy conflict combination; if the action point of the clamping element is less than 20mm from the boundary of the area to be processed, it is judged as a high-risk combination for processing interference. Only candidate parameter combinations that pass all static constraint pre-verifications can proceed to the next step of mechanical performance evaluation.

[0093] S43: Real-time evaluation of mechanical properties using embedded reduced-order simulation For candidate parameter combinations that have passed static pre-verification, the constraint solver calls the reduced-order physical simulation model of the corresponding template in real time to complete the mechanical performance evaluation of the entire processing cycle. It should be noted that "real-time calling" here means that the solver completes the simulation evaluation at each valid branch decision, rather than calling it in batches after generating a complete set of solutions, ensuring that each search step is supported by mechanical performance data.

[0094] The reduced-order physical simulation model is a fast approximation model obtained through pre-training using the intrinsic orthogonal decomposition (POD) method or response surface surrogate model method in the offline stage, based on the high-fidelity finite element analysis model. Its computation time is controlled to be on the order of 1 / 1000 to 1 / 100 of that of the full-order finite element model under the same working conditions, and the evaluation can be completed in a single search iteration without significantly increasing the solution time. The model inputs are: the coordinates of the positioning point / clamping point / support point determined by the candidate parameter combination, clamping direction, clamping sequence, support height, workpiece material properties, template structure stiffness properties, and cutting load vectors of each path segment output by S1; the model outputs are: the maximum deformation of the workpiece throughout the entire cycle, the deformation of the key machining area, the maximum displacement of the positioning contact point, the slippage trend of the clamping contact point, the natural frequency of the fixture system, the clamping safety margin, and the vibration sensitivity index.

[0095] Mechanical performance evaluation is not simply a static deformation calculation, but rather a segmented evaluation of the force response of candidate combinations throughout the entire processing cycle, combining dynamic constraints in a typified constraint diagram. Rough machining path section: Focus on evaluating overall rigidity, clamping stability, and impact resistance; Semi-finished machining path section: Focus on assessing support balance and local vibration trends; Finishing path segment: Focus on evaluating the deformation and positioning accuracy of key machining areas.

[0096] During the evaluation process, high-load and high-precision sensitive path segments are prioritized as representative working conditions to improve evaluation efficiency. It should be noted that the core difference between dynamic and static constraints is that static constraints use a fixed threshold, while dynamic constraint thresholds are dynamically adjusted according to changes in the toolpath segment. A candidate parameter combination can only be considered mechanically feasible if it meets the corresponding dynamic mechanical constraint requirements in all path segments throughout the entire machining cycle. For example, the dynamic constraint requirement for clamping edges is that the workpiece and positioning element remain in contact throughout the entire machining cycle, and the clamping point slippage does not exceed 0.02mm; the dynamic constraint requirement for supporting edges is that the maximum deformation of the supporting area does not exceed the allowable upper limit of the corresponding machining stage throughout the entire machining cycle.

[0097] S44: Heuristic Feedback Signal Generation and Search Direction Guidance After completing the mechanical performance evaluation, the evaluation results are transformed into multi-dimensional heuristic feedback signals, which are then input into the constraint solver to guide the subsequent search direction and achieve closed-loop optimization. It should be noted that the heuristic feedback signals are not simply "pass / fail" indicators, but rather structured data containing the merit levels of the solutions, failure causes, sensitive parameter directions, and correction suggestions, enabling the solver to shift from blind enumeration to directed search.

[0098] First, complete the classification and signal transformation of the evaluation results: The scheme satisfies all static and dynamic constraint thresholds: it is marked as a high-priority feasible combination, and its search priority for neighborhood parameter combinations is increased; The solution only shows slight deviations in local processing areas, while all other constraints are met: it is marked as a correctable combination, and the direction of the parameter with the highest sensitivity to deviation indicators is identified and fed back, such as prompting to adjust the support height, shift the clamping point, and optimize the clamping sequence; The solution has core defects such as clamping failure, positioning drift, and resonance risk: it is marked as an infeasible combination and the solver is prohibited from continuing to expand the search within this parameter range.

[0099] The core dimensions of the feedback signals include: comprehensive feasibility level, positioning stability level, clamping safety level, support adequacy level, machining avoidance level, critical area deformation level, and dynamic load adaptability level. The comprehensive feasibility level is divided into four categories: excellent, good, correctable, and infeasible. The critical area deformation level is divided into three categories: satisfactory, close to the upper limit, and exceeding the upper limit. The clamping safety level is divided into three categories: sufficient safety margin, insufficient margin, and potential failure risk. Example: If the predicted deformation in the critical area of ​​finishing exceeds 1 / 3 of the dimensional tolerance, it is marked as a high-precision risk combination; if the clamping safety margin in the roughing path segment is less than 80% of the preset benchmark value, it is marked as a dynamic clamping inadequacy combination.

[0100] Furthermore, the feedback signal is precisely adapted based on a predefined sensitivity mapping rule: the sensitivity mapping rule is a predefined "adjustable parameter-mechanical performance index" correlation strength matrix, which clarifies the influence weight of various parameters on different mechanical indicators. For example, the locating pin diameter mainly affects the locating contact stiffness, the clamping force value mainly affects the workpiece slippage trend, and the support pin height mainly affects the vibration amplitude in the thin-walled region. When a certain mechanical indicator does not meet the requirements, the solver automatically identifies the core parameter that has the greatest impact on that indicator, and prioritizes adjusting the value of that parameter or increasing its priority in subsequent searches.

[0101] After receiving the feedback signal, the solver performs three types of search optimization actions: Local parameter rearrangement: dynamically adjust the search order of variables. For example, when multiple candidate combinations fail to clamp consecutively, prioritize the search for clamping parameters and support parameters. Candidate branch pruning: When multiple combinations within a certain parameter range show a clear trend of failure, the branch is directly deleted to terminate the invalid search. Parameter boundary shrinkage: The search range of continuous parameters is dynamically reduced based on the feedback results. For example, if multiple combinations show that the clamping block action point is too close to the surface to be processed, the selectable range of the clamping block action point is shifted 2~5mm away from the surface to be processed.

[0102] S45: Solving for convergence control and outputting results This step is the final stage of the solution process, completing four core tasks: convergence determination, multi-template switching, exception backtracking, and result output.

[0103] First, a convergence criterion is set: the solver is considered convergent when it finds a feasible parameter combination that satisfies all core static and dynamic constraints, and no better combination with an improvement of more than 2% in the overall feasibility score appears in 10 consecutive iterations. If multiple feasible combinations satisfy all constraints, they are selected in the following priority order: fewer fixture components, simpler assembly steps, lower average clamping load, sufficient deformation margin in key areas, and minimal template modification. The optimal feasible solution is then determined.

[0104] Synchronous configuration of multi-template parallel / serial solution mechanism and exception backtracking rules: For the candidate template skeleton set output by S3, the solution can be performed sequentially from high to low according to the comprehensive matching score, or multiple templates can be solved in parallel. If the current template fails to converge to a feasible solution within the preset number of iterations (default 100, which can be adjusted according to the number of parameters and computing power), it will automatically switch to the next candidate template to continue solving; If all candidate templates fail to converge to a feasible solution, the template backtracking mechanism is triggered, a backtracking request is sent to S3, the extended candidate template skeleton set is retrieved again, and the solution process is executed again to avoid the failure of the overall generation process due to the limitations of the template structure.

[0105] During the solution process, all solution trajectory data is recorded simultaneously, including: the number of evaluated candidate combinations, the pass rate of static constraints, the pass rate of dynamic constraints, the mechanical performance indicators of each combination, and the branch records of the search tree. This data is used for traceability analysis of the solution process, early termination of invalid branches, and optimization of subsequent solutions.

[0106] The core output of this step includes: the optimal feasible parameter combination that satisfies all constraints, the corresponding mechanical performance evaluation report for the entire path segment, the constraint satisfaction status list, and heuristic search trajectory data; the output results are directly used as the core input for instantiating the S5 template skeleton.

[0107] It should be noted that this step completes the core transformation from the "initial framework to be solved" to the "verified engineering parameter scheme". Unlike the traditional fixture design mode of "separation of generation and verification", this step ensures that the mechanical performance of each candidate scheme is verified during the generation process through the closed-loop solution mechanism of embedded simulation feedback, thereby reducing rework of the scheme from the root and ensuring the mechanical reliability and processing stability of the final fixture scheme in actual processing.

[0108] S5: When the constraint solver converges to a feasible solution that satisfies all static and dynamic constraints, the template skeleton is instantiated based on the feasible solution, and a parameterized model of the fixture and the corresponding solution certificate are generated and output.

[0109] This step is an automated closing step implemented by the computer. The core inputs are the optimal feasible parameter combination output by S4, the corresponding template skeleton output by S3, and the typed constraint graph constructed by S2. The core actions are: to instantiate the template skeleton based on the feasible solution after the constraint solution converges, generate the fixture parameterized model and the corresponding solution certificate, and complete the structured output.

[0110] It should be noted that the core innovation of this step differs from the traditional fixture design model that only outputs geometric models. It adopts a dual output structure of "fixture parametric model + corresponding solution certificate". It outputs not only a digital model of the fixture that can be directly used for manufacturing and assembly, but also traceable and auditable constraint satisfaction and mechanical verification certificates. This transforms the original black-box design that relies on personal experience into a compliant design result that is interpretable throughout the entire process and traceable to all elements, meeting the quality traceability and compliance requirements of fixture design in the high-end manufacturing field.

[0111] This step is implemented sequentially according to the five computer-executable steps S51 to S55, and the specific details are as follows: S51: Eventual Consistency Verification of Feasible Solutions Before instantiating the template skeleton, an eventual consistency check is performed on the optimal feasible solution output by S4 to ensure the compliance and accuracy of the input data and avoid the instantiation result becoming invalid due to parameter passing errors. The check consists of three core dimensions: Parameter validity check: Check that all parameter values ​​are within the adjustable parameter range preset by the S3 template skeleton, and that discrete parameters are all within the preset optional specification list, with no out-of-range or non-standard values. Constraint integrity check: Confirm that the feasible solution satisfies all static and dynamic constraints in the typed constraint diagram, with no constraint omissions or logical conflicts; Consistency verification of results: Confirm that there are no contradictions between the parameter values ​​of the feasible solution, the mechanical performance evaluation results, and the constraint satisfaction state. If necessary, the full-order finite element model can be called to perform high-precision verification of the key working conditions of finishing to ensure the accuracy of the reduced-order simulation results.

[0112] Feasible solutions that pass the verification are entered into the instantiation process; feasible solutions that fail the verification are marked as invalid solutions and automatically return to S4 to re-execute the constraint satisfaction solution.

[0113] S52: Template skeleton layer instantiation Through the feasible solution of the final verification, the template skeleton is instantiated according to the preset hierarchical order, and the abstract template skeleton containing only functional roles, layout topology and parameter boundaries is transformed into a complete digital definition of fixture with definite geometric dimensions, spatial position, assembly relationship and component specifications.

[0114] Instantiation is strictly performed in the hierarchical order of "basic installation structure → positioning elements → support elements → clamping elements → connecting elements → auxiliary elements" to avoid assembly reference loss and circular dependency issues caused by improper instantiation order. First, instantiate the basic installation structure, including the fixture base plate and machine tool adapter interface, and determine the overall coordinate system and installation datum of the fixture; Then instantiate the positioning element, and generate the corresponding geometric model on the basic installation structure according to the position and model parameters of the positioning element in the feasible solution, and establish the assembly constraint relationship with the basic structure. Then, support elements are instantiated, and support pins, floating supports, and other element models are generated based on the support position and height parameters in the feasible solution. Next, the clamping elements are instantiated, and the clamping action point, action direction, and specification parameters in the feasible solution are used to generate component models such as clamping blocks and pressure plates, and establish corresponding assembly constraints. Then instantiate the connecting elements, including standard connecting parts such as bolts and pins, and automatically select and assemble them according to the assembly relationship; Finally, auxiliary components are instantiated, including tool setting blocks, chip removal grooves, error prevention structures, and other auxiliary structures.

[0115] During instantiation, continuous parameters directly drive model geometry updates based on the values ​​of feasible solutions; discrete parameters are instantiated by retrieving standard parts models of corresponding specifications from a pre-set library of standard fixture components. Each component in the library carries metadata such as specification parameters, material properties, and interface dimensions. For non-standard components such as irregularly shaped pressure blocks and customized transition connection plates, parameterized construction rules are automatically invoked based on feasible solution parameters to automatically generate them.

[0116] S53: Instantiated Model Interference Check and Consistency Confirmation After instantiation, two core checks are automatically performed to ensure that the generated fixture model is completely consistent with the solution objective, eliminating any risks to engineering implementation: Assembly interference check: Perform global static interference and assembly path detection on the complete fixture assembly to confirm that there is no volume overlap or path obstacle between components, and that the minimum distance between the component and the tool envelope meets the S2 preset safety threshold. Consistency confirmation of solution results: Check that the spatial coordinates, action direction, and specification selection of the instantiated model are completely consistent with the feasible solution output by S4, confirm that the component arrangement does not intrude into the main approach direction region of the tool, and that the layout of the positioning, clamping, and support units is consistent with the requirements of the constraint diagram.

[0117] If interference or consistency deviation is detected, local parameter correction, component replacement, or assembly relationship adjustment will be performed automatically; if automatic correction is not possible, the operator will be prompted to confirm the adjustment through the interactive interface until the model passes all verifications.

[0118] S54: Binary Output Generation and Association Binding After verification, two core outputs are generated simultaneously: the fixture parameterized model and the corresponding solution certificate, and the two are then linked and bound together.

[0119] Fixture parametric model generation: The system automatically generates a complete parametric model of the fixture, which includes at least the following: the overall 3D structural model of the fixture, the component assembly relationship model, the component parameter table, the standard parts list, the key installation dimensions table, and the workpiece-fixture clamping correspondence. The model output format supports common 3D formats such as STEP and IGES, as well as the native formats of mainstream CAD software, while retaining a complete parametric feature tree to support subsequent local modifications and adjustments.

[0120] Furthermore, supporting engineering auxiliary documents can be generated simultaneously, including fixture assembly engineering drawings, parts processing drawings, component procurement lists, assembly process cards, and on-site clamping operation guidelines, reducing repetitive work for engineering personnel.

[0121] Solving certificate generation: The certificate is a structured technical credential tied to the current fixture solution, rather than a simple result summary, and its core consists of five parts: Scheme identifiers: workpiece number, process number, fixture scheme number, template skeleton identifier, generation time, system version number; Constraint Satisfaction List: List the conditions under which each of the five types of constraint relationships is satisfied, mark the threshold and actual value, and specify the result as "passed / condition passed / not applicable"; Mechanical performance report: includes key indicators for the entire S4 path segment (maximum workpiece deformation, positioning point displacement, clamping point slippage trend, system natural frequency) and comparison results with thresholds; Process traceability information: number of candidate combinations, pass rate of static / dynamic constraints, number of convergence iterations, and summary of search path; Verification conclusion: The overall compliance of the plan is clarified by "pass / conditionally pass / fail".

[0122] Two-way association binding: Complete the two-way binding between the model and the certificate: embed the certificate's unique identifier and verification code into the parameterized model, and record the model's hash verification value in the certificate to ensure that the two correspond one-to-one and avoid mismatch during the process.

[0123] S55: Result Encapsulation and Template Library Iterative Feedback After completing the binary output, two final tasks are performed: result encapsulation and template library iteration. Result structured encapsulation: The parametric model, solution certificate, component list, parameter table, assembly tree and engineering auxiliary files are packaged into a fixture solution result set for the corresponding workpiece process. A unique identifier is generated according to the workpiece number, process number and timestamp, which supports retrieval, calling and version management. Template library iteration feedback: The current solution, solution success rate, mechanical indicators, and convergence iteration data are fed back to the S3 parameterized template library, the template skeleton usage statistics are updated, the retrieval matching and sorting logic is optimized, and the template library is continuously iterated.

[0124] It should be noted that this step, which transforms the validated abstract parameter combination into a deliverable engineering result, is a core element of value delivery. Unlike the traditional approach that only outputs a geometric model, this step, through a dual output and binding mechanism, ensures that the result possesses both geometric definition and verification basis, providing complete data support for compliance review, quality traceability, and continuous optimization throughout the fixture's entire lifecycle.

[0125] Example 2 is the second embodiment of the present invention, which differs from the previous embodiment in that: If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art or the current technical solution, can be embodied in the form of a software product. This current computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0126] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0127] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0128] Example 3, referring to Figure 2 As an embodiment of the present invention, a fixture generation system based on parametric templates and intelligent reasoning is provided, which includes a data extraction module, a constraint graph construction module, a template retrieval module, a constraint solving module and an instantiation output module; Data extraction module: acquires the digital model of the workpiece and the machining process document, and extracts the geometric features, datum information, process content and tool path information of the workpiece; Constraint graph construction module: Constructs a typed constraint graph, in which nodes represent workpiece features and fixture element placeholders, and edges represent positioning relationships, clamping relationships, support relationships, assembly relationships and machining accessibility relationships, and encodes the cutting load spectrum parsed from the toolpath information as the dynamic constraint conditions of the edges; Template retrieval module: Retrieves a template skeleton that matches the workpiece digital model from a preset parametric template library. The template skeleton defines the type, layout topology, and adjustable parameter value range of the fixture. Constraint Solving Module: Performs constraint satisfaction solving on the template skeleton. During the solving process, for each candidate parameter combination, a pre-built reduced-order physical simulation model is called in real time to evaluate the mechanical performance of the candidate parameter combination, and the evaluation result is used as a heuristic feedback signal to input the constraint solver to guide the subsequent search direction. Instantiation output module: When the constraint solver converges to a feasible solution that satisfies all static and dynamic constraints, the template skeleton is instantiated based on the feasible solution, and the fixture parameterized model and corresponding solution certificate are generated and output.

[0129] Example 4 is an embodiment of the present invention, which provides a fixture generation method based on parameterized templates and intelligent reasoning. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation / comparative experiments.

[0130] This embodiment compares the method of the present invention with two existing technical methods, namely the rule-based parametric template matching method and the traditional serial simulation verification method, through a specific multi-process machining task—the fixture generation process of titanium alloy thin-walled blades for aero-engines. This verifies the beneficial effects of the present invention in terms of scheme generation efficiency, first-pass yield of mechanical properties, and adaptability to dynamic working conditions throughout the entire process.

[0131] Using a titanium alloy blade from the low-pressure compressor of a certain type of aero-engine as the test object, the workpiece has the following dimensions: length 315mm, chord width 128mm, maximum thickness 18.6mm, minimum blade wall thickness 1.8mm, and material TC4 titanium alloy. The machining task involves completing three operations—roughing, semi-finishing, and finishing—on a five-axis machining center. The roughing depth of cut is 1.5mm, the semi-finishing depth of cut is 0.5mm, and the finishing depth of cut is 0.18mm. The finishing surface tolerance requirement is ±0.025mm. The workpiece digital model was imported in STEP format with MBD annotations, containing complete design datum, dimensional tolerance, and geometric tolerance annotation information. The machining process file is a CNC program file generated by UGNX post-processing, containing complete tool movement trajectories and cutting parameters for the three operations. The experimental hardware environment was configured as follows: a graphics workstation with an Intel Xeon W-2295 processor, 128GB DDR4 memory, and an NVIDIA RTX A5000 graphics card, and equipped with the CATIA V5 3D CAD system, ABAQUS finite element analysis software, and the fixture intelligent generation prototype system developed by our research group.

[0132] This embodiment sets up three methods for comparison. The first is a rule-based parametric template matching method, which extracts the geometric features and reference information of the workpiece, matches the optimal template from a preset standard fixture template library according to coding rules, directly outputs the fixture scheme, and calls finite element software for independent verification. If it fails, the designer adjusts the template selection and parameter settings according to the verification report and re-executes the process. The second is the traditional serial simulation verification method, in which process designers complete the preliminary design of the fixture scheme based on the workpiece characteristics and experience, and then call finite element software for mechanical property verification. If the verification fails, the clamping position and clamping force parameters are adjusted based on experience, and it is iterated repeatedly until the acceptance conditions are met or the preset iteration limit is reached. The third is the method of this invention, which is executed according to steps S1 to S5.

[0133] Step S1: The system reads the MBD digital model of the blade through the CATIA secondary development interface, extracting one back arc datum surface, one blade basin datum surface, one tenon datum surface, two datum holes, one blade profile to be machined, six candidate support surfaces, and four clamping surfaces. The complete machining content for roughing, semi-finishing, and finishing processes is extracted from the CNC program file generated by UG post-processing: Roughing: 4156 tool points, spindle speed 8000 rpm, feed rate 1800 mm / min, depth of cut 1.5 mm, width of cut 12 mm; Semi-finishing: 6234 tool points, spindle speed 12000 rpm, feed rate 1200 mm / min, depth of cut 0.5 mm, width of cut 6 mm; Finishing: 11382 tool points, spindle speed 18000 rpm, feed rate 900 mm / min, depth of cut 0.18 mm, width of cut 2.5 mm. The toolpaths for the three machining operations are segmented according to the machining area, cutting state, and degree of tool posture change. Roughing yields 6 stable cutting path segments and 2 transitional path segments; semi-finishing yields 9 stable cutting path segments and 4 transitional path segments; and finishing yields 11 stable cutting path segments and 6 transitional path segments. Spatial mapping of the workpiece design coordinate system, process clamping coordinate system, and machine tool machining coordinate system is completed simultaneously.

[0134] Step S2: Construct a typed constraint diagram. The workpiece feature nodes include 1 back arc datum plane node, 1 blade basin datum plane node, 1 tenon datum plane node, 2 datum hole nodes, 6 support candidate surface nodes, 4 clamping action surface nodes, and 1 machining-affected surface node, totaling 16 nodes. The fixture element placeholder nodes include 3 main positioning element placeholders, 2 auxiliary positioning element placeholders, 4 clamping element placeholders, 2 support element placeholders, 4 connecting element placeholders, and 2 auxiliary element placeholders, totaling 17 nodes. The edge set includes 5 positioning relationship edges, 4 clamping relationship edges, 2 support relationship edges, 8 assembly relationship edges, and 6 machining accessibility relationship edges. Based on the Kienzle cutting force formula, the cutting load spectrum for each path segment of the three processes was calculated: the peak load for roughing occurred in path segment 3, with three-dimensional components Fx=287.6N, Fy=453.2N, and Fz=112.8N; the peak load for semi-finishing occurred in path segment 5, with three-dimensional components Fx=96.3N, Fy=158.7N, and Fz=38.5N; and the peak load for finishing occurred in path segment 7, with three-dimensional components Fx=42.8N, Fy=71.3N, and Fz=17.2N. The load spectrum of the three processes was encoded as dynamic constraint conditions for clamping and supporting edges, and differentiated influence weights were set according to the machining type in step S1: the deformation sensitivity weight of the finishing path segment was 2.8 times that of the roughing path segment, and the load tolerance weight of the roughing path segment was 1.8 times that of the finishing path segment. Static and dynamic constraint attributes are attached to the corresponding nodes and edges according to the requirements of step S23. The constraint priority is set in the order of positioning relationship edge, processing accessibility relationship edge, clamping relationship edge, support relationship edge, and assembly relationship edge.

[0135] Step S3: Parametric Template Library Retrieval. The pre-set parametric template library for aerospace blades is categorized at the first level by workpiece reference mode, including 34 sets of surface reference dominant templates, 18 sets of hole reference dominant templates, and 12 sets of curved surface reference dominant templates; at the second level by blade structural features, including 22 sets of thin-walled shell templates and 16 sets of irregular curved surface templates; and at the third level by process type, including 18 sets of roughing templates, 15 sets of semi-finishing templates, and 12 sets of finishing templates. For the current workpiece, the tenon bottom surface and back arc reference surface are used as the main positioning surfaces, and the double-sided lug holes of the blade are used as auxiliary positioning surfaces. After initial screening using rigid rules, 8 candidate templates are retained. Through multi-dimensional similarity comparison, evaluation is performed from two dimensions: node layer and edge layer. The top 4 template skeletons with the highest comprehensive matching scores form a candidate set, with scores of 0.892, 0.834, 0.781, and 0.743, respectively. Adaptability checks were performed on each candidate template. The finishing support unit of candidate template 2 did not fully cover the sensitive area in the middle section of the blade, and the clamping element of candidate template 4 had an angle of 11° with the main approach direction of the finishing tool; both failed the adaptability check. Ultimately, candidate templates 1 and 3 passed all checks. The placeholder nodes in the constraint diagram were bound to the corresponding interfaces of the template skeleton through interface mapping, and the range of values ​​for the locating pin diameter was narrowed to the adaptability range based on the measured diameter of the blade locating hole.

[0136] Step S4: Constraint Satisfaction Solution. The first candidate template skeleton, "tenon bottom surface positioning with lug hole limiting and multi-point floating support," is used as the initial solution skeleton. There are 23 adjustable parameters in total, including 16 continuous parameters (5 for the center position of the positioning element, 7 for the action point of the clamping element, and 4 for the height of the support element) and 7 discrete parameters (2 for the model of the positioning element, 3 for the specification of the clamping element, and 2 for the form of the support head). A hybrid integer programming solver based on branch and bound is used to solve the constraints. The adjustable parameters are categorized into continuous and discrete branches for hierarchical solution. During the solution process, after each set of candidate parameters is generated, a pre-built POD reduced-order physical simulation model is called for evaluation. The reduced-order model is pre-trained offline based on the full-order finite element model of this blade workpiece. A single evaluation takes approximately 0.18 seconds, only 1 / 760th of the time taken by the full-order model. According to the requirements of step S43, the mechanical performance evaluation focuses on the roughing, semi-finishing, and finishing paths: the roughing path emphasizes checking the overall stiffness and clamping stability; the semi-finishing path emphasizes evaluating the support balance and local vibration trend; and the finishing path emphasizes checking the deformation and positioning accuracy of the key blade profile. When the evaluation results of the reduced-order model show that the predicted deformation of a certain candidate combination in the blade profile of finishing path 7 exceeds the allowable threshold, the solver identifies the position of clamping point C2 and the height of support pin D3 as sensitive parameters according to the sensitivity mapping rule, and prioritizes adjusting the assignment order of these two sets of parameters in subsequent searches. The solution convergence criterion is set as follows: the improvement of the comprehensive feasibility score does not exceed 2.5% in 8 consecutive iterations, and the deformation of each processing stage does not exceed the allowable threshold of the corresponding process. When the first candidate template converges to a feasible solution within the preset number of iterations (100), the solution process is terminated and the optimal feasible parameter combination is output.

[0137] Step S5: Perform a final consistency check on the feasible solution output from S4. All parameter values ​​are within the preset range of the template skeleton, and the discrete parameter selections are all from the preset optional specification list. There are no contradictions between the parameter values ​​and the mechanical performance evaluation results. Instantiate the template skeleton layer by layer for the feasible solution that has passed the check, generating a parameterized model of the fixture containing 3 positioning elements, 4 clamping elements, 2 floating support elements, and 1 connecting base plate. Simultaneously generate a solution certificate, including a scheme identifier, a constraint satisfaction status list (all 25 constraint edges of five types pass), a mechanical performance verification report (containing complete data on the maximum deformation of the workpiece, displacement of the positioning point, slippage trend of the clamping point, and the natural frequency of the system for each path segment of the three processes), and solution process traceability information. Complete the two-way hash binding between the model file and the certificate before output.

[0138] The rule-based parametric template matching method and the traditional serial simulation verification method use the same workpiece data and initial constraints. The template library for the parametric template matching method is a standard fixture parts library used in the industry, and the matching rules select the optimal template based on the similarity of workpiece feature codes. The traditional serial simulation verification method was independently completed by an engineer with over 8 years of experience in aerospace blade fixture design, and every adjustment and time spent during the design process was recorded. Key recorded data are shown in Table 1. Table 1: Experimental Data Recording Table

[0139] The experimental data clearly show that the method of this invention is significantly superior to the two existing technical methods in both the efficiency of scheme generation and the quality of the final scheme. The following is a detailed analysis from three aspects.

[0140] Regarding the efficiency of scheme generation, the parametric template matching method took an average of 43.8 minutes for 15 sets of experiments, while the serial simulation verification method took an average of 56.2 minutes. Both methods adopt a serial mode of "independent verification after scheme generation." When the verification finds out-of-tolerance deformation or insufficient clamping force, designers need to adjust parameters based on experience and re-execute the entire process. This results in a significant positive correlation between the total time and the wall thickness and accuracy level of the workpiece—the smaller the wall thickness, the higher the accuracy requirement, the more iterations, and the longer the time. Taking BLADE-10 with a wall thickness of only 1.4mm as an example, the parametric template matching method took 62.3 minutes, and the serial simulation verification method took 84.8 minutes, reflecting the increasingly serious efficiency bottleneck of ultra-thin wall parts in the traditional serial mode. The method of this invention took an average of only 13.9 minutes for 15 sets of experiments, which is about 68.3% less than the parametric template matching method and about 75.3% less than the serial simulation verification method. The core source of this efficiency advantage lies in the fact that by embedding the reduced-order physical simulation model into each round of search iteration of the constraint solver, the mechanical evaluation results guide the parameter search direction in real time in the form of heuristic feedback signals, and branches that show mechanical failure trends are pruned in advance, avoiding large-scale blind searches in infeasible regions.

[0141] Regarding the quality of the solutions, the average maximum deformation of the workpiece during the entire machining cycle under the fixture solution generated by the parametric template matching method is 27.4 μm, while the average deformation under the serial simulation verification method is 32.6 μm. However, under the fixture solution generated by the method of this invention, the average maximum deformation of the workpiece is 11.4 μm, a reduction of approximately 58.4% compared to the parametric template matching method and approximately 65.0% compared to the serial simulation verification method. The parametric template matching method, relying solely on preset static geometric rules for template selection, cannot predict the impact of tool cutting forces on local areas of the workpiece during the solution generation stage, leading to deviations in the support layout of thin-walled areas such as blade profiles from the optimal position. Although the serial simulation verification method introduces finite element analysis for mechanical property verification, the verification result is only a binary judgment of "pass / fail." When faced with complex three-dimensional deformation fields, engineers find it difficult to accurately determine which parameters should be adjusted and the magnitude and direction of the adjustment. The method of this invention encodes the cutting load spectrum of each path segment of the three processes into dynamic constraint conditions of the constraint graph edges. This allows each candidate scheme to traverse the stress conditions of the entire machining cycle during the solution generation process. The position and height of the support element can be optimized in a targeted manner based on the deformation evaluation results at the path segment level. Finally, the maximum deformation of the scheme within the entire machining cycle is precisely controlled.

[0142] Regarding compliance and traceability, the solution certificate mechanism introduced in this embodiment demonstrates a unique advantage not found in the two existing methods. The outputs of the parametric template matching method and the serial simulation verification method are only three-dimensional fixture models. The constraint satisfaction states, mechanical performance verification data, and parameter adjustment history involved in the design process are not recorded in a structured form. When subsequent verification of the solution's compliance is required, the verification process must be re-executed. The solution certificate output by the method of this invention contains 26 verifiable items, covering complete information such as the satisfaction states of each of the five types of constraint edges, the comparison of mechanical performance indicators and thresholds for the entire path segment of the three processes, and the iterative trajectory of solution convergence. This provides a structured technical certificate independent of the three-dimensional model for the process review and quality archiving of the fixture solution. This feature has significant engineering application value in the tooling design of key components such as aero-engine blades, and can meet the stringent requirements of airworthiness regulations for manufacturing process quality traceability.

[0143] In summary, the 15 sets of comparative test data in this embodiment fully verify the beneficial effects of the present invention in improving fixture generation efficiency, enhancing the mechanical performance of the solution, strengthening the adaptability to dynamic working conditions of multiple processes, and providing traceable compliance credentials. Compared with the parametric template matching method and the serial simulation verification method, the present invention systematically improves the existing technical mode by constructing a typed constraint graph that integrates static and dynamic constraints, embedding reduced-order physical simulation as a solver evaluation module, supporting adaptive switching of the structural skeleton with a candidate template set, and outputting a binary result structure with verification credentials.

[0144] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A fixture generation method based on parameterized templates and intelligent reasoning, characterized in that, include: S1: Obtain the digital model of the workpiece and the machining process file, and extract the geometric features, datum information, process content and tool path information of the workpiece; S2: Construct a typed constraint graph, in which nodes represent workpiece features and fixture element placeholders, and edges represent positioning relationships, clamping relationships, support relationships, assembly relationships and machining accessibility relationships. The cutting load spectrum parsed from the toolpath information is encoded as the dynamic constraint conditions of the edges. S3: Retrieve a template skeleton that matches the workpiece digital model from a preset parametric template library. The template skeleton defines the type, layout topology, and adjustable parameter range of the fixture. S4: Perform constraint satisfaction solution on the template skeleton. During the solution process, for each candidate parameter combination, call the pre-built reduced-order physical simulation model in real time to evaluate the mechanical performance of the candidate parameter combination, and use the evaluation result as a heuristic feedback signal to input the constraint solver to guide the subsequent search direction. S5: When the constraint solver converges to a feasible solution that satisfies all static and dynamic constraints, the template skeleton is instantiated based on the feasible solution, and a parameterized model of the fixture and the corresponding solution certificate are generated and output.

2. The fixture generation method based on parameterized templates and intelligent reasoning as described in claim 1, characterized in that, Step S1 specifically includes: Obtain the digital model of the workpiece and the machining process documents, and perform unified preprocessing on the data from cross-system sources in terms of format, coordinate system, naming rules and unit system; The workpiece digital model is subjected to topological analysis and manufacturing feature semantic recognition to extract workpiece geometric features directly related to fixture generation. The workpiece geometric features include reference plane, reference hole, surface to be machined, support sensitive surface and clamping action surface. Workpiece reference information is extracted synchronously. The reference information includes the correspondence between design reference, process reference and inspection reference, as well as the hierarchical relationship between primary reference, secondary reference and tertiary reference. Explicitly marked references are read directly, and references not explicitly marked are inferred. Design reference and process reference are distinguished and their priority order is recorded. Extract the process content from the processing technology document. The process content includes process number, processing location, processing sequence, processing type, number of clamping operations, processing allowance, and processing accuracy requirements. Tool path information is extracted from the machining process document, and the transformation relationship between the workpiece design coordinate system, the process clamping coordinate system and the machine tool machining coordinate system is established. The tool path information is mapped to the geometric surface corresponding to the workpiece digital model. The tool path information includes tool position coordinates, interpolation type, feed rate, spindle speed and tool posture data. The tool path information is segmented according to the machining area, cutting state and the degree of tool posture change. After extraction, a full data consistency check is performed. If there is a conflict between the toolpath coverage area and the candidate positioning surface, the corresponding candidate surface is remarked. If there is interference between the candidate clamping surface and the main approach direction of the tool, the corresponding candidate surface is remarked. Output a structured workpiece machining description dataset, including a feature list, a baseline relationship list, a process task list, a path segment list, and a coordinate mapping relationship table.

3. The fixture generation method based on parameterized templates and intelligent reasoning as described in claim 2, characterized in that, Step S2 includes: Establish a node set, which includes workpiece feature nodes and fixture element placeholder nodes. The workpiece feature nodes are generated based on the workpiece geometric features extracted in step S1 and carry feature type, spatial position, normal direction, size range, surface condition, material properties, process association information and datum association information. The fixture element placeholder nodes include positioning element placeholders, clamping element placeholders, support element placeholders, connecting element placeholders and auxiliary element placeholders. An edge set is established, comprising positioning relationship edges, clamping relationship edges, support relationship edges, assembly relationship edges, and machining accessibility relationship edges. The positioning relationship edges characterize the datum correspondence and degree-of-freedom constraint relationship between workpiece feature nodes and positioning element placeholder nodes. The clamping relationship edges characterize the force relationship between clamping element placeholder nodes and workpiece feature nodes. The support relationship edges characterize the load-bearing contact relationship and deformation suppression relationship between support element placeholder nodes and workpiece feature nodes. The assembly relationship edges characterize the assembly connection constraints between fixture element placeholder nodes. The machining accessibility relationship edges characterize the spatial compatibility relationship between the tool movement space and the potential arrangement area of ​​fixture elements.

4. The fixture generation method based on parameterized templates and intelligent reasoning as described in claim 3, characterized in that, Step S2 also includes: The set of nodes and the set of edges are given constraint attributes, which are divided into static constraint attributes and dynamic constraint attributes. The static constraint attributes represent fixed constraint requirements that are independent of the machining sequence, while the dynamic constraint attributes represent constraint requirements that are dynamically adjusted as the tool path segment changes. The constraints are set to a priority order, from highest to lowest: positioning relationship edge, processing accessibility relationship edge, clamping relationship edge, support relationship edge, and assembly relationship edge. The cutting load spectrum parsed from the tool path information in step S1 is encoded into the dynamic constraint conditions of the corresponding edge. During the encoding process, differentiated influence weights are set for path segments of different machining types. The deformation sensitivity weight of the finishing path segment is higher than that of the roughing path segment, and the load allowance weight of the roughing path segment is higher than that of the finishing path segment. The staged allowable deformation constraints are set according to the machining type.

5. The fixture generation method based on parameterized templates and intelligent reasoning as described in claim 4, characterized in that, Step S3 includes: The system invokes a pre-defined parameterized template library, which includes multiple template entries. Each template entry is stored as a combination of functional description information, applicable condition information, structural topology information, parameter interface information, and constraint compatibility information, and adopts a three-level hierarchical organizational structure. Primary classification is based on workpiece reference pattern; Secondary classification is based on the main structural features of the workpiece; The three-level classification is based on the type of work process; Based on the structured workpiece machining description dataset output in step S1 and the typed constraint graph constructed in step S2, a template retrieval feature description is generated. The feature description includes the workpiece reference combination type, the number of main positioning features, the distribution of candidate clamping areas, the distribution of candidate support areas, the location of the area to be machined, the set of main tool approach directions, and the distribution of dynamic load sensitive areas. Template skeleton retrieval is performed using a combination of hard rule-based initial screening and multi-dimensional similarity comparison: The initial screening of rigid rules includes the consistency judgment of template reference mode and workpiece reference combination type, the coverage of template positioning unit quantity and workpiece degree of freedom restriction requirements, and the avoidance of the template base layout area from the main processing area and the main approach direction of the tool. Multidimensional similarity comparison is evaluated from two dimensions: node layer and edge layer. The comprehensive matching scores are sorted from high to low, and the top few template skeletons are selected to form a candidate template skeleton set.

6. The fixture generation method based on parameterized templates and intelligent reasoning as described in claim 5, characterized in that, Step S3 also includes: Perform a fit check on each template skeleton in the candidate template skeleton set. The check includes: The compatibility between the number of template positioning units and the number of positioning degrees of freedom required by the workpiece; The compatibility between the working area of ​​the template clamping element and the allowable clamping area of ​​the workpiece; The coverage of the template support unit arrangement area with the dynamically load-sensitive area; Interference check between the template element contour envelope and the main approach direction of the tool; The template skeleton that passes the adaptability check is then mapped to the parameter interface with the typed constraint graph constructed in step S2: Map the positioning element placeholder nodes in the typed constraint diagram to the positioning unit interfaces of the template skeleton; Map the clamping element placeholder node to the clamping unit interface; Map the support element placeholder nodes to the support unit interfaces, and attach the constraint attributes corresponding to each relation edge to the corresponding interface; Based on the specific dimensions and process requirements of the current workpiece, the initial value range of the parameters in the template skeleton is narrowed and limited. If the placeholder node in the constraint diagram has no corresponding interface in the template skeleton, an additional interface is added according to the preset expansion rules. If there is a redundant interface in the template skeleton and there is no corresponding placeholder node in the constraint diagram, the interface is temporarily frozen and not included in the subsequent solution variable set.

7. The fixture generation method based on parameterized templates and intelligent reasoning as described in claim 6, characterized in that, Step S4 includes: Using the typed constraint graph output in step S2 and the template skeleton output in step S3 as input, the adjustable parameters in the template skeleton are defined as solution variables, the range of values ​​of the adjustable parameters is defined as the search domain, and the constraint conditions carried by each edge in the typed constraint graph are mapped to the constraint conditions of the constraint solver. The solution variables include continuous variables and discrete variables. Candidate parameter combinations are generated. Before each candidate parameter combination enters the mechanical performance evaluation, static constraint pre-verification is performed. The verification includes whether the positioning unit completely constrains the six spatial degrees of freedom of the workpiece, whether the clamping action direction is compatible with the normal of the workpiece surface, whether the support unit falls into the allowable support area, whether there is spatial interference between the fixture elements, and whether the fixture elements intrude into the main approach direction area of ​​the tool.

8. The fixture generation method based on parameterized templates and intelligent reasoning as described in claim 7, characterized in that, Step S4 also includes: For candidate parameter combinations that have passed the static constraint pre-verification, a reduced-order physical simulation model is called in real time to evaluate mechanical performance. The reduced-order physical simulation model is a fast evaluation model obtained by reducing the order of a high-fidelity finite element analysis model in the offline stage. The mechanical performance evaluation combines the dynamic constraint conditions in the typified constraint diagram to perform segmented evaluation of the candidate parameter combinations at the path segment level throughout the entire processing cycle. The mechanical performance evaluation results are transformed into multi-dimensional heuristic feedback signals. These heuristic feedback signals include the comprehensive feasibility level of candidate parameter combinations, failure cause categories, and sensitive parameter direction information. These signals are input into the constraint solver to guide the subsequent search direction. After receiving the feedback signals, the constraint solver performs three types of search optimization actions: local parameter rearrangement, candidate branch pruning, and parameter boundary shrinkage. When the solver finds a feasible parameter combination that satisfies all static and dynamic constraints, and no better combination with an improvement in the overall feasibility score exceeding a preset threshold appears in several consecutive iterations, the solution is deemed to have converged, and the feasible parameter combination, mechanical performance evaluation results, constraint satisfaction status records, and heuristic search trajectory information are output.

9. The fixture generation method based on parameterized templates and intelligent reasoning as described in claim 8, characterized in that, Step S5 specifically includes: Perform an eventual consistency check on the feasible solution output in step S4. The check includes whether the parameter values ​​are within the preset range of the template skeleton, whether all static and dynamic constraints in the typed constraint diagram are satisfied, and whether there are any contradictions between the parameter values ​​and the mechanical performance evaluation results. The feasible solution that has passed the verification is instantiated into a template skeleton based on its parameter values. The continuous parameters in the parameter values ​​directly drive the geometric update of the model. The discrete parameters in the parameter values ​​are instantiated by retrieving the standard part model of the corresponding specification from the preset standard fixture component library to generate the fixture parameterized model. A solution certificate is generated synchronously. The solution certificate includes scheme identification information, constraint satisfaction status list, mechanical performance verification report, solution process traceability information and final verification conclusion. Before output, the fixture parameterized model is associated and bound with the solution certificate. The association and binding includes embedding the unique identifier of the solution certificate in the fixture parameterized model file and recording the hash verification value of the fixture parameterized model file in the solution certificate.

10. A fixture generation system based on parametric templates and intelligent reasoning, used to implement the fixture generation method based on parametric templates and intelligent reasoning as described in any one of claims 1 to 9, characterized in that, include: Data extraction module: acquires the digital model of the workpiece and the machining process document, and extracts the geometric features, datum information, process content and tool path information of the workpiece; Constraint graph construction module: Constructs a typed constraint graph, in which nodes represent workpiece features and fixture element placeholders, and edges represent positioning relationships, clamping relationships, support relationships, assembly relationships and machining accessibility relationships, and encodes the cutting load spectrum parsed from the toolpath information as the dynamic constraint conditions of the edges; Template retrieval module: Retrieves a template skeleton that matches the workpiece digital model from a preset parametric template library. The template skeleton defines the type, layout topology, and adjustable parameter value range of the fixture. Constraint Solving Module: Performs constraint satisfaction solving on the template skeleton. During the solving process, for each candidate parameter combination, a pre-built reduced-order physical simulation model is called in real time to evaluate the mechanical performance of the candidate parameter combination, and the evaluation result is used as a heuristic feedback signal to input the constraint solver to guide the subsequent search direction. Instantiation output module: When the constraint solver converges to a feasible solution that satisfies all static and dynamic constraints, the template skeleton is instantiated based on the feasible solution, and the fixture parameterized model and corresponding solution certificate are generated and output.

Images