A metal product design research and development collaborative management system and method

Abstract

The application discloses a metal product design research and development collaborative management system and method, and relates to the field; the system comprises a multi-source heterogeneous data access module, which is used for receiving and standardizing CAD geometric model data, material attribute parameters, manufacturing process constraint rules and historical project knowledge base data from multiple design terminals; a semantic feature extraction module, which is in communication connection with the multi-source heterogeneous data access module and is used for carrying out word segmentation and word vector coding on unstructured design requirement texts in the engineering data set; a collaborative matching degree calculation module; a conflict detection and resolution module; and an iterative optimization feedback module.The application realizes unified access and standardization of CAD model data, material performance parameters, process constraint rules and historical project knowledge bases through the multi-source heterogeneous data access module, breaks tool chain barriers and eliminates information silos.

Description

Technical Field

[0001] This invention relates to the field of metal products design and development collaborative management system and method. Background Technology

[0002] The metal products industry encompasses multiple sub-sectors, including fasteners, springs, metal tools, containers, and metal packaging containers. Its product design and development process typically involves the collaborative participation of multiple professional roles, such as structural engineers, materials engineers, process engineers, and manufacturing engineers. In the traditional R&D model, each professional often uses independent CAD / CAE tools for design, with data transmitted via local files or email, resulting in the following prominent problems: First, there is a serious problem of information silos. Design data from different disciplines are stored in their respective toolchains, lacking a unified data access and standardization mechanism. This results in key information such as material property parameters and process constraints not being shared and referenced in a timely manner during the design phase.

[0003] Second, cross-disciplinary collaboration is inefficient. Materials engineers and process engineers can only begin their work sequentially after the structural design is finalized; this serial workflow leads to lengthy R&D cycles. When a particular discipline raises modification requests, it often requires tracing back to upstream processes, resulting in significant rework costs.

[0004] Third, the matching degree of design, process, and materials is difficult to quantify and assess. During the scheme review stage, the material applicability, process feasibility, and geometric compatibility between different design schemes usually rely on expert experience for qualitative judgment, lacking a systematic quantitative evaluation model, which easily leads to subjectivity and inconsistency in review conclusions.

[0005] Fourth, parameter conflict detection is delayed. Due to the lack of real-time constraint verification mechanisms during parallel design across different disciplines, parameter conflicts are often only discovered during the manufacturing stage, by which time the cost of modification has increased exponentially.

[0006] In recent years, collaborative design platforms and product lifecycle management (PLM) systems have alleviated the aforementioned problems to some extent. However, existing systems mostly focus on document management and process approval, lacking in-depth modeling capabilities for the multi-dimensional matching relationships of materials, processes, and structures in the design process of metal products, and also lacking automated conflict detection and resolution mechanisms. Therefore, there is an urgent need for a collaborative management system and methodology specifically designed for the design and development of metal products to achieve efficient collaboration among multiple disciplines and intelligent decision support.

[0007] Therefore, this invention proposes a collaborative management system and method for the design and development of metal products. Summary of the Invention

[0008] The purpose of this invention is to address the shortcomings of existing technologies by proposing a collaborative management system and method for the design and development of metal products.

[0009] To achieve the above objectives, the present invention adopts the following technical solution: A collaborative management system for the design and development of metal products includes: The multi-source heterogeneous data access module is used to receive and standardize CAD geometric model data, material property parameters, manufacturing process constraint rules, and historical project knowledge base data from multiple design terminals, and output engineering datasets in a unified format. The semantic feature extraction module communicates with the multi-source heterogeneous data access module. It is used to perform word segmentation and word vector encoding on the unstructured design requirement text in the engineering dataset, normalize and align the structured engineering parameters, and map the two to a unified semantic vector space to generate design semantic feature vectors. The collaborative matching degree calculation module communicates with the semantic feature extraction module and has a built-in weighted cosine similarity model. It is used to receive the design semantic feature vectors corresponding to two or more design schemes and calculate the weighted matching degree score between the schemes in three dimensions: material selection, process route and geometric features. The conflict detection and resolution module is connected to the collaborative matching degree calculation module and the multi-source heterogeneous data access module, respectively. It is used to encode the parameter constraints in each design scheme into variable-domain-constraint triples based on the constraint satisfaction problem model, automatically detect conflicts between cross-disciplinary design parameters, and output a set of resolution strategies. The iterative optimization feedback module communicates with the conflict detection and resolution module. It encapsulates the matching score, conflict detection results and resolution strategy into a feedback data packet, which is then pushed to the corresponding design terminal in real time through the message middleware, driving each terminal to iteratively modify the design scheme.

[0010] Preferably, the semantic feature extraction module includes: The text encoding subunit is configured to use a pre-trained language model to perform contextual encoding on the design requirement text and output a text semantic vector. The parameter encoding subunit is configured to perform Min-Max normalization on the structured engineering parameters and map them to the same dimensional space as the text semantic vector through a fully connected layer; The vector concatenation subunit is configured to concatenate the text semantic vector and the parameter encoding vector according to the feature dimension to form a unified design semantic feature vector.

[0011] Preferably: the weighted cosine similarity model in the collaborative matching degree calculation module has the following expression for calculating the matching degree score: ,in This represents the weighted matching score between design scheme A and design scheme B, where k is the dimension index, with values ​​of 1, 2, and 3, corresponding to the material selection dimension, process route dimension, and geometric feature dimension, respectively. Preset weight coefficients for the k-th dimension Let A and B represent the feature vectors of scheme A and scheme B in the k-th dimension, respectively.

[0012] Preferably: the constraints in the conflict detection and resolution module satisfy the problem model, and the conflict detection process is formalized as a triple CSP=(X,D,C), where: This is a set of design parameter variables, where each variable corresponds to a specific engineering parameter; Let D be the set of values ​​that each variable can take. i For variable x i The range of valid values; Let c be a set of constraints. j Defined as a Boolean function over a subset of variables; When constraint c exists j This causes the variable assignment combination to not satisfy c. j When a conflict is identified, a set of resolution strategies is generated using a backtracking search algorithm, prioritizing the adjustment of variables with the widest range of values.

[0013] Preferably: the conflict detection and resolution module calculates the conflict severity index I. conflict Its expression is: ,in The number of constraints that are violated. The preset priority weight for the j-th constraint that is violated. The attenuation coefficient is a positive value. The actual deviation of the j-th constraint is defined as the minimum distance between the current value of the variable and the boundary of the allowable range of the constraint.

[0014] Preferably, the iterative optimization feedback module is further configured to calculate the feedback priority weight P. feedback Its expression is: , This is a balancing factor used to adjust the relative weight of the matching score and the severity of conflict in the feedback priority. A weighted score is given to the matching degree between the current design scheme and the optimal reference scheme. The feedback data packet is calculated according to P, which represents the severity index of the conflict in the current design scheme. feedback The values ​​are sorted from highest to lowest and pushed out sequentially.

[0015] Preferably, the multi-source heterogeneous data access module is also configured to interface with PLM system, ERP system and MES system, obtain historical data of the entire product life cycle from PLM system, obtain material inventory and cost data from ERP system, obtain process execution data in the actual manufacturing process from MES system, and uniformly convert the above external system data into JSONSchema format for storage and distribution.

[0016] A collaborative management method for the design and development of metal products includes the following steps: S1 receives engineering data from multiple design terminals through the multi-source heterogeneous data access module, performs data cleaning, format standardization and metadata annotation, and generates a unified format engineering dataset. S2, the semantic feature extraction module performs semantic feature extraction on the engineering dataset, mapping the unstructured design requirement text and structured engineering parameters to a unified semantic vector space, and generating design semantic feature vectors corresponding to each design scheme; S3, through the collaborative matching degree calculation module, uses the weighted cosine similarity model to calculate the weighted matching degree score between different design schemes, and identifies the degree of matching in three dimensions: material selection, process route and geometric features. S4, through the conflict detection and resolution module, encodes and detects the parameter constraints in multi-disciplinary parallel design based on the constraint satisfaction problem model, identifies conflict parameters and generates a set of resolution strategies; S5, through iterative optimization of the feedback module, encapsulates the matching score, conflict detection results and resolution strategy into a feedback data package, sorts them according to feedback priority weights and pushes them to the corresponding design terminals in real time, driving each terminal to iteratively modify the design scheme until all constraints are met and the matching score reaches the preset threshold. In step S2, the unstructured design requirements text and structured engineering parameters are mapped to a unified semantic vector space, specifically including: The design requirement text is segmented into words, stop words are removed, and then it is input into a pre-trained language model to extract context-related word vector representations. After global average pooling, the text semantic vector is obtained. The structured engineering parameters are Min-Max normalized so that the values ​​of each dimension are mapped to the [0,1] interval, and then the parameter encoding vector is obtained by linear transformation through a fully connected layer. The text semantic vector and the parameter encoding vector are concatenated according to the feature dimension to form the design semantic feature vector; In step S5, each terminal is driven to iteratively modify the design scheme, specifically including: Set the maximum number of iterations T max Matching threshold S threshol ; In the t-th iteration (t=1, 2, ..., T) max Each design terminal receives feedback data packets, adjusts the corresponding design parameters according to the set of resolution strategies, and generates an updated design scheme. Repeat steps S2 to S4 to calculate the updated weighted matching score. ; like ≥S threshold ,and If the collaborative design is successful, the final design scheme set is output; otherwise, proceed to the (t+1)th iteration until T is reached. max Then output the current optimal solution and trigger a prompt for manual intervention.

[0017] A collaborative management system for the design and development of metal products is integrated into a management terminal. The management terminal specifically includes a housing and a front cover fixed to the front side of the housing. A main control board is fixed inside the housing. The main control board directly integrates a multi-source heterogeneous data access module, a semantic feature extraction module, a collaborative matching degree calculation module, a conflict detection and resolution module, and an iterative optimization feedback module. Alternatively, the main control board can be connected to the multi-source heterogeneous data access module, semantic feature extraction module, collaborative matching degree calculation module, conflict detection and resolution module, and iterative optimization feedback module via a plug module. A touch screen display electrically connected to the housing is provided on the inner wall of the front cover.

[0018] Preferably, the back of the housing is provided with two sets of heat dissipation grooves, and the housing is provided with two sets of oppositely arranged cooling fans at the heat dissipation grooves. The inner wall of the housing located at the heat dissipation slot is also rotatably connected to a baffle for controlling the opening of the heat dissipation slot. A bidirectional telescopic device is fixed to the inner wall of the housing. The telescopic end of the bidirectional telescopic device is rotatably connected to a connecting rod, and the other end of the connecting rod is rotatably connected to the eccentric side wall of the baffle. The bidirectional telescopic device includes an outer shell fixed to the inner wall of the housing and two piston plates symmetrically slidably connected to the inner wall of the outer shell. A piston rod is fixed to the side wall of the piston plate, and the piston rod is rotatably engaged with the connecting rod. The inner wall of the outer casing is fastened with a spring, the other end of which is fastened to the side wall of the piston plate, and the inner wall of the outer casing is filled with a thermally expanding fluid.

[0019] The beneficial effects of this invention are as follows: 1. This invention achieves unified access and standardization of CAD model data, material performance parameters, process constraint rules, and historical project knowledge base through a multi-source heterogeneous data access module, breaking down toolchain barriers and eliminating information silos.

[0020] 2. This invention uses a semantic feature extraction module to uniformly map unstructured design requirement text and structured engineering parameters to a semantic vector space, enabling text descriptions and numerical parameters to be correlated and matched on the same dimension.

[0021] 3. This invention, through the weighted cosine similarity model built into the collaborative matching degree calculation module, realizes the quantitative matching degree evaluation of different design schemes in three key dimensions: material selection, process route, and geometric features, providing objective data support for scheme review.

[0022] 4. This invention achieves automatic identification of cross-disciplinary parameter conflicts through a conflict detection and resolution module based on a constraint satisfaction problem model. By combining a conflict severity index and a set of resolution strategies, it significantly reduces the time cost of manually investigating conflicts.

[0023] 5. This invention, based on conventional heat dissipation, by setting baffles, can, on the one hand, block the heat dissipation slots when not in use or when the shell temperature is low, preventing external impurities from entering and allowing the shell interior to heat up quickly to the optimal operating temperature. On the other hand, by controlling the thermal expansion fluid, it can also adaptively control the opening of the heat dissipation slots according to the temperature inside the shell, achieving an adaptive match between temperature and heat dissipation efficiency. At the same time, the thermal expansion fluid is set as a mixture of nitrogen dioxide and dinitrogen tetroxide, utilizing its reversible reaction energy to increase the thermal expansion coefficient of the thermal expansion fluid, thereby ensuring the reliability of its driving and sensing. Attached Figure Description

[0024] Figure 1 This is an architecture diagram of a collaborative management system for the design and development of metal products proposed in this invention; Figure 2 This is a flowchart of a collaborative management method for the design and development of metal products proposed in this invention; Figure 3 This is a structural diagram of the management terminal of a collaborative management system for the design and development of metal products proposed in this invention. Figure 4 This is an internal diagram of the management terminal of a collaborative management system for the design and development of metal products proposed in this invention. Figure 5 This is a partial cross-sectional view of the management terminal of a collaborative management system for the design and development of metal products proposed in this invention.

[0025] In the picture: 1. Housing; 2. Plug module; 3. Main control board; 4. Front cover; 5. Touch screen; 6. Baffle; 7. Heat dissipation slot; 8. Cooling fan; 9. Linkage rod; 10. Two-way telescopic device; 11. Outer shell; 12. Piston plate; 13. Piston rod; 14. Spring. Detailed Implementation

[0026] The technical solution of the present invention will be further described in detail below with reference to specific embodiments.

[0027] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "setting" should be interpreted broadly. For example, they can refer to a fixed connection or setting, a detachable connection or setting, or an integral connection or setting. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0028] Example 1: like Figure 1 As shown, the metal product design and development collaborative management system provided by the present invention adopts a layered architecture design, which includes, from bottom to top, a data access layer, a feature processing layer, a collaborative computing layer, a conflict management layer, and an application feedback layer.

[0029] The data access layer corresponds to the multi-source heterogeneous data access module, responsible for collecting raw data from various data sources. The feature processing layer corresponds to the semantic feature extraction module, responsible for performing feature engineering on the raw data. The collaborative computing layer corresponds to the collaborative matching degree calculation module, responsible for performing quantitative calculations of matching metrics between schemes. The conflict management layer corresponds to the conflict detection and resolution module, responsible for constraint encoding, conflict identification, and generation of resolution strategies. The application feedback layer corresponds to the iterative optimization feedback module, responsible for encapsulating the calculation results and pushing them to various design terminals.

[0030] Multi-source heterogeneous data access module The multi-source heterogeneous data access module is the data entry point for the entire system. Its main function is to uniformly access, clean, standardize, and store data scattered across different professional toolchains.

[0031] Module composition: This module further includes: a CAD data parsing subunit, a material parameter access subunit, a process rule access subunit, a historical knowledge base access subunit, and a data standardization subunit.

[0032] Module functions: The CAD data parsing subunit is configured to parse CAD geometric model files in formats such as STEP, IGES, and Parasolid from the structural engineer terminal, extract geometric feature parameters (such as wall thickness, hole diameter, chamfer radius, draft angle, etc.) and topological information (such as the number of entities, number of facets, Boolean operation relationships, etc.), and convert them into an intermediate representation format within the system.

[0033] The material parameter access subunit is configured to read key performance parameters of metallic materials from a material database or ERP system, including but not limited to: yield strength σ. sTensile strength σ b Elongation δ, Hardness HB, Density ρ, Coefficient of Thermal Expansion α thermal Thermal conductivity λ and material unit price C material These parameters are organized in key-value pairs and are accompanied by metadata tags such as material grade and applicable standards.

[0034] The process rule access subunit is configured to receive manufacturing process constraint rules from the process engineer terminal, including minimum casting wall thickness, lower limit of forging fillet radius, range of welding bevel angle, and tolerance of heat treatment deformation. These rules are stored in the form of structured constraints, and each rule includes: applicable material range, applicable geometric feature type, constraint parameter name, constraint upper and lower bounds, and constraint priority.

[0035] The historical knowledge base access sub-unit is configured to extract design schemes, review records, trial production data, and mass production feedback from historical metal product projects in the enterprise PLM system to construct a historical project knowledge graph. Nodes in the knowledge graph include: product model, material grade, process route, key geometric parameters, failure mode, and quality rating; edges represent the relationships between nodes (such as "materials used", "process experienced", "leading to failure", etc.).

[0036] The data standardization sub-unit is configured to uniformly convert data from the aforementioned sub-units into JSON Schema format. For numerical parameters, units are standardized (e.g., length is standardized to mm, force to N, and temperature to ℃); for textual parameters, encoding standardization is performed (e.g., material grades are standardized to GB / T standard naming). The standardized data is tagged with metadata such as timestamps, source terminal IDs, and data version numbers, and then persistently stored in the system's internal data lake.

[0037] The core principle of this module is data pattern matching and format conversion. Because the data formats output by different professional tools vary significantly, the module employs the Adapter Pattern to implement a dedicated adapter for each data source. The adapter is responsible for parsing the source data format into a universal intermediate representation, and then the data standardization subunit performs pattern mapping. The pattern mapping process is based on a predefined mapping rule table, which records the source field name, target field name, data type, unit conversion formula, and missing value handling strategy. When a new type of data source is connected, only the corresponding adapter needs to be added and the mapping rule registered; no modification to the code of other modules in the system is required, ensuring the system's scalability.

[0038] Semantic feature extraction module The semantic feature extraction module is a bridge connecting unstructured text data and structured numerical data. Its core function is to map the two different modalities of data to the same semantic vector space, so that subsequent matching degree calculation and conflict detection can be carried out under a unified mathematical framework.

[0039] Module composition: This module includes: a text encoding subunit, a parameter encoding subunit, and a vector concatenation subunit.

[0040] Module functions: The text encoding subunit is configured to process unstructured text data from design requirement documents, technical specifications, review comments, etc. First, the text is segmented using either jieba (for Chinese) or NLTK (for English), and stop words and low-frequency noise words are removed based on the TF-IDF or TextRank algorithms. Then, the cleaned word sequence is input into a pre-trained language model (such as BERT-base or RoBERTa) for context encoding, extracting the context-related word vector representation for each word. Finally, global average pooling is performed on the vector representations of all words to obtain a fixed-dimensional text semantic vector.

[0041] The parameter encoding subunit is configured to process structured engineering parameters from the data access layer. First, the numerical parameters are subjected to Min-Max normalization, which linearly maps the values ​​of each dimension to the [0,1] interval.

[0042] The core principle of this module is multimodal feature fusion. In the design scenario of metal products, the design requirement text describes the product's functional objectives, usage environment, and performance indicators (e.g., "corrosion-resistant fasteners for high-temperature and high-pressure environments"), while engineering parameters provide specific geometric dimensions, material grades, and manufacturing processes. Both are indispensable—text alone cannot provide precise quantification, and parameters alone cannot convey the design intent. By encoding text and parameters separately into the same dimensional space and then concatenating them, semantic-level fusion is achieved. The use of a pre-trained language model ensures that the text encoding can capture the contextual meaning of domain-specific terms (e.g., the different meanings of "heat treatment" in different contexts), while the learnable parameters of the fully connected layer allow the parameter encoding to adaptively adjust the contribution weights of each engineering parameter dimension to the final feature vector.

[0043] Collaborative matching degree calculation module Feature subvector splitting unit is configured to split the design semantic feature vector of each design scheme. Split into three sub-vectors according to predefined dimensional partitioning rules: (Material selection dimension sub-vector) (Process route dimension sub-vector) and (Geometric feature dimension sub-vectors). The partitioning rules are determined based on the original data source corresponding to each dimension of the vector—dimensions derived from material parameter access sub-units and material-related keywords in the text. This is derived from the dimension classification of process rule access sub-units and process-related keywords in the text. The dimensions are derived from the CAD data parsing sub-unit and the geometrically related keywords in the text. .

[0044] The dimension weight configuration unit is configured with preset weight coefficients for the three dimensions of storage and management. ,satisfy The weighting coefficients can be configured by the system administrator according to the company's R&D strategy. For example, for high-precision metal products, the weight of geometric feature dimensions can be appropriately increased. For low-cost products produced in large quantities, the weight of the process route dimension can be increased. The weighted matching score calculation unit is configured to calculate the cosine similarity of two input design schemes A and B in three dimensions, and then sum them according to their weights to obtain the final weighted matching score S. match (A, B), the score ranges from [-1, 1], and a larger value indicates a higher degree of overall matching between the two solutions in terms of materials, processes, and geometry. When S match When (A, B) > 0.8, the system determines that the two schemes are highly compatible and can be merged into one scheme for subsequent processing; when S match When (A, B) < 0.3, the system determines that there is a significant difference between the two schemes and a special review process needs to be triggered.

[0045] Module principle: The core principle of this module is multi-dimensional weighted similarity measurement. Cosine similarity measures the closeness of two vectors in a direction, unaffected by the vector magnitude, making it suitable for comparing the relative similarity of different design schemes in feature space. The weighted summation method allows the system to flexibly adjust the contribution ratio of each dimension according to the company's actual R&D priorities. For example, for metal products in the aerospace field, the material selection dimension (e.g., titanium alloy vs. aluminum alloy) has a much greater impact on the final product performance than the process route dimension; in this case, a weighted summation method can be used. For everyday hardware products, the process route (such as stamping vs. die casting) has a greater impact on cost and efficiency, and can be set... .

[0046] Conflict detection and resolution module The conflict detection and resolution module is responsible for automatically identifying parameter conflicts and generating feasible resolution strategies during multi-disciplinary parallel design processes. It is a key module for reducing rework and shortening the R&D cycle.

[0047] Module composition: This module includes: a constraint coding subunit, a conflict identification subunit, a conflict severity calculation subunit, and a conflict resolution strategy generation subunit.

[0048] Module functions: The constraint encoding subunit is configured to encode the parametric constraints in each design scheme into a standard triple CSP=(X,D,C) of the constraint satisfaction problem (CSP). Here, each variable x in the variable set X... i This corresponds to a specific engineering parameter (such as "wall thickness", "welding current", "heat treatment temperature", etc.); each value in the set D is taken from the range D. i The legal range of values ​​for the corresponding variable under engineering specifications and physical laws is defined (e.g., wall thickness x). thickness The domain is [2.0mm, 10.0mm]); each constraint c in the constraint set C. j The logical relationships between variables are defined, such as "if the material is aluminum alloy, the welding current shall not exceed 200A" and "the ratio of wall thickness to hole diameter shall not be less than 0.5".

[0049] The conflict identification sub-unit is configured to traverse all constraints c j ∈C, check if the current combination of variable assignments satisfies each constraint. If a certain constraint c j If the condition is not met, add it to the set of violated constraints C. violated Record the specific variable that violates the constraint and the deviation Δ between its current value and the allowable range. j .

[0050] The conflict severity calculation sub-unit is configured to calculate the conflict severity index based on the set of violated constraints.

[0051] The resolution strategy generates sub-units configured for C. violated For each constraint violated, a set of resolution strategies is generated using a backtracking search algorithm. The algorithm starts by trying to adjust the variable with the widest range boundary (because this type of variable has the largest adjustment space and the least impact on the overall design), and then gradually narrows the range of each variable until a set of variable assignments that satisfies all constraints is found. For each resolution strategy, the system records the name of the parameter to be adjusted, the suggested adjustment direction (increase / decrease), the suggested adjustment magnitude, and a description of the expected effect.

[0052] Module principle: The core principle of this module is constraint satisfaction problem solving and conflict quantification assessment. The design parameter conflict problem is formalized into a CSP model, allowing conflict detection to be transformed into logical judgments of constraints, and conflict resolution strategy generation to be transformed into a CSP solution process. Conflict severity index I. conflict The design comprehensively considers two factors: constraint priority and deviation amount. Higher priority constraints result in greater severity when violated; under the same constraint, the larger the deviation, the greater the severity. The exponential decay function makes the sensitivity of severity to deviation amount decrease as deviation increases, which aligns with engineering intuition—when the deviation is already large, further increasing the deviation tends to saturate the effect on severity, because at this point, significant modifications are necessary regardless.

[0053] Iterative optimization feedback module The iterative optimization feedback module is the interaction hub between the system and each design terminal. It is responsible for converting the calculation results into actionable feedback information and driving the iterative optimization process.

[0054] Module composition: This module includes: a feedback data packet encapsulation subunit, a priority calculation subunit, a message push subunit, and an iterative convergence judgment subunit.

[0055] Module functions: The feedback data packet encapsulation subunit is configured to encapsulate the matching score S output by the collaborative matching degree calculation module. match The conflict severity index I output by the conflict detection and resolution module conflict The set of resolution strategies is encapsulated to generate a structured feedback data packet. Each data packet contains: the target design scheme ID, a list of conflict parameters involved, the severity of each conflict, the recommended resolution strategy, the current comprehensive matching score, and the natural language suggestion text generated by the system (such as "It is recommended to increase the wall thickness from 1.5mm to 2.5mm to meet the minimum wall thickness requirement of the casting process").

[0056] The priority calculation subunit is configured to calculate the feedback priority weight P for each feedback data packet. feedback The message push sub-unit is configured according to P feedback Feedback data packets are pushed in real time to the corresponding design terminals via WebSocket long-lived connections or message queues (such as RabbitMQ or Kafka) in descending order of value. Terminal ID routing information is included with the push to ensure that each feedback is only delivered to the relevant designer.

[0057] The iterative convergence judgment subunit is configured to monitor the system state after each iteration. A maximum number of iterations T is set. max (T) max =10) and matching threshold S threshold (S)threshold =0.85). After the t-th iteration, if the weighted matching score of all design schemes satisfies... And the severity index of the conflict If (i.e., no constraint is violated), then the collaborative design is considered converged, the system automatically locks the design scheme and generates a design review report. If T is reached... max If the iteration still fails to converge, the system outputs the current set of optimal solutions and triggers a manual intervention prompt, notifying the project leader to coordinate offline.

[0058] Module principle: The core principle of this module is closed-loop feedback control and priority scheduling. The collaborative design process is viewed as a feedback control system: each design terminal generates a design scheme (the controlled object), the system calculates the matching degree and conflicts (sensors), generates a resolution strategy (controller), and pushes it to the terminal to drive modifications (actuators), forming a closed loop. The priority scheduling mechanism ensures that system resources (designers' attention) are prioritized for the most urgent and important issues, avoiding wasting time on a large number of low-priority feedbacks. Example 2: like Figure 2 As shown, the collaborative management method for metal product design and development provided by this invention includes the following steps: Step S1: Multi-source data access and standardization. After system startup, the multi-source heterogeneous data access module listens for data upload requests from various design terminals. When a CAD model file is received, the CAD data parsing subunit calls the corresponding format parser to extract geometric feature parameters; when material parameter data is received, the material parameter access subunit verifies data integrity and attaches metadata tags; when process constraint rules are received, the process rule access subunit compiles them into an internal constraint representation. All data is ultimately converted into a unified JSON Schema format by the data standardization subunit and written to the data lake.

[0059] Step S2: Semantic Feature Extraction. The semantic feature extraction module reads the standardized engineering dataset from the data lake. The text encoding subunit performs word segmentation, stop word removal, and BERT encoding on the design requirement text, outputting a 128-dimensional text semantic vector. The parameter encoding subunit performs Min-Max normalization on the engineering parameters and then maps them to a 128-dimensional parameter encoding vector through a fully connected layer. The vector concatenation subunit concatenates the two to form a 256-dimensional design semantic feature vector.

[0060] Step S3: Collaborative matching degree calculation. The collaborative matching degree calculation module receives design semantic feature vectors from different design terminals. The feature sub-vector splitting unit splits the 256-dimensional vector into three 128-dimensional sub-vectors (material, process, geometry) according to predefined dimensional partitioning rules. The dimensional weight configuration unit reads the current weight configuration. The weighted matching degree calculation unit calculates the cosine similarity of each scheme pair in the three dimensions in turn, and sums them by weight to obtain the final S. match Rating matrix.

[0061] Step S4: Conflict Detection and Resolution. The conflict detection and resolution module encodes the parameter constraints of each scheme into CSP triples. The conflict identification sub-unit traverses all constraints, marks the violated constraints, and records the deviation. The conflict severity calculation sub-unit calculates I... conflict The resolution strategy generation sub-unit performs a backtracking search for each violated constraint, generating a set of resolution strategies containing parameter adjustment suggestions.

[0062] Step S5: Iterative optimization feedback. The iterative optimization feedback module encapsulates the feedback data packet and calculates P. feedback Prioritization is performed, and the results are pushed to each design terminal via WebSocket after being sorted by priority. Upon receiving feedback, each terminal highlights conflicting parameters in its local CAD / CAE tool and displays a resolution suggestion panel for designers' reference. Designers modify their solutions based on the suggestions and resubmit, at which point the system automatically enters the next iteration until convergence or manual intervention is triggered.

[0063] Example 3: A collaborative management system for the design and development of metal products is integrated into a management terminal. The management terminal specifically includes a housing 1 and a front cover 4 fixed to the front side of the housing 1. A main control board 3 is fixed inside the housing 1. The main control board 3 directly integrates a multi-source heterogeneous data access module, a semantic feature extraction module, a collaborative matching degree calculation module, a conflict detection and resolution module, and an iterative optimization feedback module. Alternatively, the main control board 3 is connected to the multi-source heterogeneous data access module, the semantic feature extraction module, the collaborative matching degree calculation module, the conflict detection and resolution module, and the iterative optimization feedback module through a plug module 2. A touch screen 5 electrically connected to the housing 1 is provided on the inner wall of the front cover 4.

[0064] The back of the housing 1 is provided with two sets of heat dissipation slots 7, and the housing 1 is provided with two sets of oppositely arranged heat dissipation fans 8 at the heat dissipation slots 7. The opposite arrangement means that one set of heat dissipation fans 8 blows air into the housing 1, and the other set of heat dissipation fans 8 blows air into the outside of the housing 1, forming a stable flow channel for circulating cooling air.

[0065] The inner wall of the housing 1 located at the heat dissipation groove 7 is also rotatably connected to a baffle 6 for controlling the opening of the heat dissipation groove 7. A bidirectional telescopic device 10 is fixed to the inner wall of the housing 1. The telescopic end of the bidirectional telescopic device 10 is rotatably connected to a connecting rod 9. The other end of the connecting rod 9 is rotatably connected to the eccentric side wall of the baffle 6.

[0066] The bidirectional telescopic device 10 includes an outer shell 11 fixed to the inner wall of the housing 1 and two piston plates 12 symmetrically slidably connected to the inner wall of the outer shell 11. A piston rod 13 is fixed to the side wall of the piston plate 12, and the piston rod 13 is rotatably engaged with the connecting rod 9.

[0067] A spring 14 is fastened to the inner wall of the outer casing 11, and the other end of the spring 14 is fastened to the side wall of the piston plate 12. The inner wall of the outer casing 11 is filled with thermal expansion fluid.

[0068] In this embodiment, the specific type of thermally expanding fluid is not limited, but is preferably a mixture of nitrogen dioxide and dinitrogen tetroxide, with the following equilibrium reaction chemical formula: In this embodiment, when the ambient temperature inside the housing 1 is relatively low, the thermal expansion fluid is in a contracted state, and the bidirectional expansion joint 10 is in a contracted state. This causes the baffle 6 to rotate via the connecting rod 9, completely sealing the heat dissipation groove 7. When the ambient temperature inside the housing 1 is relatively high, the thermal expansion fluid is in an expanded state, causing the bidirectional expansion joint 10 to extend. The baffle 6 rotates to open the heat dissipation groove 7. The higher the temperature, the greater the volume expansion of the thermal expansion fluid, and the greater the opening of the heat dissipation groove 7.

[0069] This device, based on conventional heat dissipation, incorporates a baffle 6. On one hand, it can seal the heat dissipation groove 7 when not in use or when the temperature of the housing 1 is low, preventing external impurities from entering and allowing the interior of the housing 1 to heat up rapidly to reach the optimal operating temperature. On the other hand, through the control of the thermal expansion fluid, it can adaptively control the opening of the heat dissipation groove 7 according to the temperature inside the housing 1, achieving an adaptive match between temperature and heat dissipation efficiency. At the same time, the thermal expansion fluid is a mixture of nitrogen dioxide and dinitrogen tetroxide, utilizing its reversible reaction to increase the thermal expansion coefficient of the fluid, thereby ensuring the reliability of its drive and sensing.

[0070] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A collaborative management system for the design and development of metal products, characterized in that, include: The multi-source heterogeneous data access module is used to receive and standardize CAD geometric model data, material property parameters, manufacturing process constraint rules, and historical project knowledge base data from multiple design terminals, and output engineering datasets in a unified format. The semantic feature extraction module communicates with the multi-source heterogeneous data access module. It is used to perform word segmentation and word vector encoding on the unstructured design requirement text in the engineering dataset, normalize and align the structured engineering parameters, and map the two to a unified semantic vector space to generate design semantic feature vectors. The collaborative matching degree calculation module communicates with the semantic feature extraction module and has a built-in weighted cosine similarity model. It is used to receive the design semantic feature vectors corresponding to two or more design schemes and calculate the weighted matching degree score between the schemes in three dimensions: material selection, process route and geometric features. The conflict detection and resolution module is connected to the collaborative matching degree calculation module and the multi-source heterogeneous data access module, respectively. It is used to encode the parameter constraints in each design scheme into variable-domain-constraint triples based on the constraint satisfaction problem model, automatically detect conflicts between cross-disciplinary design parameters, and output a set of resolution strategies. The iterative optimization feedback module communicates with the conflict detection and resolution module. It encapsulates the matching score, conflict detection results and resolution strategy into a feedback data packet, which is then pushed to the corresponding design terminal in real time through the message middleware, driving each terminal to iteratively modify the design scheme.

2. The collaborative management system for metal product design and development according to claim 1, characterized in that, The semantic feature extraction module includes: The text encoding subunit is configured to use a pre-trained language model to perform contextual encoding on the design requirement text and output a text semantic vector. The parameter encoding subunit is configured to perform Min-Max normalization on the structured engineering parameters and map them to the same dimensional space as the text semantic vector through a fully connected layer; The vector concatenation subunit is configured to concatenate the text semantic vector and the parameter encoding vector according to the feature dimension to form a unified design semantic feature vector.

3. The collaborative management system for metal product design and development according to claim 1, characterized in that, The weighted cosine similarity model in the collaborative matching degree calculation module has the following expression for calculating the matching degree score: ,in This represents the weighted matching score between design scheme A and design scheme B, where k is the dimension index, with values ​​of 1, 2, and 3, corresponding to the material selection dimension, process route dimension, and geometric feature dimension, respectively. Preset weight coefficients for the k-th dimension Let A and B represent the feature vectors of scheme A and scheme B in the k-th dimension, respectively.

4. The collaborative management system for metal product design and development according to claim 1, characterized in that, The constraint satisfaction problem model in the conflict detection and resolution module formalizes the conflict detection process as a triple CSP=(X,D,C), where: This is a set of design parameter variables, where each variable corresponds to a specific engineering parameter; Let D be the set of values ​​that each variable can take. i For variable x i The range of valid values; Let c be a set of constraints. j Defined as a Boolean function over a subset of variables; When constraint c exists j This causes the variable assignment combination to not satisfy c. j When a conflict is identified, a set of resolution strategies is generated using a backtracking search algorithm, prioritizing the adjustment of variables with the widest range of values.

5. The collaborative management system for metal product design and development according to claim 1, characterized in that, The conflict detection and resolution module calculates the conflict severity index I. conflict Its expression is: ,in The number of constraints that are violated. The preset priority weight for the j-th constraint that is violated. The attenuation coefficient is a positive value. The actual deviation of the j-th constraint is defined as the minimum distance between the current value of the variable and the boundary of the allowable range of the constraint.

6. The collaborative management system for metal product design and development according to claim 1, characterized in that, The iterative optimization feedback module is also configured to calculate the feedback priority weight P. feedback Its expression is: , This is a balancing factor used to adjust the relative weight of the matching score and the severity of conflict in the feedback priority. A weighted score is given to the matching degree between the current design scheme and the optimal reference scheme. The feedback data packet is calculated according to P, which represents the severity index of the conflict in the current design scheme. feedback The values ​​are sorted from highest to lowest and pushed out sequentially.

7. The collaborative management system for metal product design and development according to claim 1, characterized in that, The multi-source heterogeneous data access module is also configured to interface with PLM system, ERP system and MES system, obtain historical data of the entire product life cycle from PLM system, obtain material inventory and cost data from ERP system, obtain process execution data of the actual manufacturing process from MES system, and uniformly convert the above external system data into JSONSchema format for storage and distribution.

8. The collaborative management system for metal product design and development according to claim 1, which is integrated into the management terminal, is characterized in that, The management terminal specifically includes a housing (1) and a front cover (4) fixed to the front side of the housing (1). The main control board (3) is fixed inside the housing (1). The main control board (3) directly integrates a multi-source heterogeneous data access module, a semantic feature extraction module, a collaborative matching degree calculation module, a conflict detection and resolution module, and an iterative optimization feedback module. Alternatively, the main control board (3) is connected to the multi-source heterogeneous data access module, the semantic feature extraction module, the collaborative matching degree calculation module, the conflict detection and resolution module, and the iterative optimization feedback module through a plug module (2). The inner wall of the front cover (4) is provided with a touch screen (5) that is electrically connected to the housing (1).

9. A collaborative management system for the design and development of metal products according to claim 8, characterized in that, The back of the housing (1) is provided with two sets of heat dissipation grooves (7), and the housing (1) is provided with two sets of oppositely arranged heat dissipation fans (8) at the heat dissipation grooves (7). The inner wall of the housing (1) located at the heat dissipation groove (7) is also rotatably connected to a baffle (6) for controlling the opening of the heat dissipation groove (7). The inner wall of the housing (1) is fixed with a bidirectional telescopic device (10). The telescopic end of the bidirectional telescopic device (10) is rotatably connected to a connecting rod (9). The other end of the connecting rod (9) is rotatably connected to the side wall of the baffle (6) at the eccentric position. The bidirectional telescopic device (10) includes an outer shell (11) fixed to the inner wall of the housing (1) and two piston plates (12) symmetrically slidably connected to the inner wall of the outer shell (11). The side wall of the piston plate (12) is fixed with a piston rod (13), and the piston rod (13) is rotatably engaged with the connecting rod (9). The inner wall of the outer shell (11) is fastened with a spring (14), the other end of the spring (14) is fastened to the side wall of the piston plate (12), and the inner wall of the outer shell (11) is filled with thermal expansion fluid.

10. A collaborative management method for the design and development of metal products, applied to the collaborative management system for the design and development of metal products as described in any one of claims 1-9, characterized in that, Includes the following steps: S1 receives engineering data from multiple design terminals through the multi-source heterogeneous data access module, performs data cleaning, format standardization and metadata annotation, and generates a unified format engineering dataset. S2, the semantic feature extraction module performs semantic feature extraction on the engineering dataset, mapping the unstructured design requirement text and structured engineering parameters to a unified semantic vector space, and generating design semantic feature vectors corresponding to each design scheme; S3, through the collaborative matching degree calculation module, uses the weighted cosine similarity model to calculate the weighted matching degree score between different design schemes, and identifies the degree of matching in three dimensions: material selection, process route and geometric features. S4, through the conflict detection and resolution module, encodes and detects the parameter constraints in multi-disciplinary parallel design based on the constraint satisfaction problem model, identifies conflict parameters and generates a set of resolution strategies; S5, through iterative optimization of the feedback module, encapsulates the matching score, conflict detection results and resolution strategy into a feedback data package, sorts them according to feedback priority weights and pushes them to the corresponding design terminals in real time, driving each terminal to iteratively modify the design scheme until all constraints are met and the matching score reaches the preset threshold. In step S2, the unstructured design requirements text and structured engineering parameters are mapped to a unified semantic vector space, specifically including: The design requirement text is segmented into words, stop words are removed, and then it is input into a pre-trained language model to extract context-related word vector representations. After global average pooling, the text semantic vector is obtained. The structured engineering parameters are Min-Max normalized so that the values ​​of each dimension are mapped to the [0,1] interval, and then the parameter encoding vector is obtained by linear transformation through a fully connected layer. The text semantic vector and the parameter encoding vector are concatenated according to the feature dimension to form the design semantic feature vector; In step S5, each terminal is driven to iteratively modify the design scheme, specifically including: Set the maximum number of iterations T max Matching threshold S threshol ; In the t-th iteration (t=1, 2, ..., T) max Each design terminal receives feedback data packets, adjusts the corresponding design parameters according to the set of resolution strategies, and generates an updated design scheme. Repeat steps S2 to S4 to calculate the updated weighted matching score. ; like ≥S threshold ,and If the collaborative design is successful, the final design scheme set is output; otherwise, proceed to the (t+1)th iteration until T is reached. max Then output the current optimal solution and trigger a prompt for manual intervention.

Images