Target essential process design method based on double effects of demand and constraint
By using the element-objective transformation method and combining multi-dimensional constraints, the problem of unsystematic mapping between user needs and process parameters was solved, achieving high efficiency, accuracy, and dynamic adaptability in process design, and improving the design efficiency and quality of forging processes.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-02
- Publication Date
- 2026-04-07
AI Technical Summary
In existing forging process design, the mapping relationship between user needs and process parameters lacks systematization, resulting in long process development cycles, high trial production costs, and difficulty in fully considering constraints. This leads to deviations in the process scheme during actual production, especially with delayed response under special working conditions.
We adopt a target-based process design method under the dual influence of demand and constraints. Through the transformation of elements into targets, user requirements are transformed into executable process targets. The design targets are dynamically adjusted by using the element chain, combined with multi-dimensional constraints.
It improves the efficiency of demand conversion, ensures the integrity and operability of process design objectives, can dynamically adapt to new materials and special working conditions, reduces subjectivity, and improves the accuracy and efficiency of process design.
Smart Images
Figure CN121809027A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of intelligent manufacturing technology, and relates to product process design, particularly to a target-based process design method under the dual influence of demand and constraints. For example, in the field of aerospace composite material manufacturing, by combining factory process documents and technical standards, process parameters can be dynamically adjusted to improve yield. Background Technology
[0002] In the process of intelligent transformation of modern manufacturing, process reasoning technology, as a core module of intelligent process design software, plays a crucial role in achieving accurate process decision-making, shortening product development cycles, and ensuring manufacturing quality. In the forging process field, existing process design target analysis methods are user-demand oriented, using demand transformation or intent understanding to establish design goals. The determination of process parameters mainly relies on engineers' experience and trial-and-error methods. This experience-based approach has technical limitations. Furthermore, process knowledge is scattered across literature, company documents, and expert experience, lacking a unified structured expression and thus failing to support intelligent decision-making.
[0003] This leads to: (1) Low demand conversion efficiency: There is a lack of systematic mapping relationship between user demand (such as quality performance, production efficiency, etc.) and final process parameters, resulting in long process development cycle and high trial production cost; (2) Incomplete consideration of constraints: Existing methods are difficult to effectively integrate multi-source constraint information (such as material properties, equipment capabilities, environmental limitations, etc.) from technical standards, patent literature, factory practice, etc., which can easily cause deviations in the process scheme in actual production; Moreover, the constraint processing adopts static rules, which cannot dynamically respond to multi-objective demands; In particular, the response to special working conditions is delayed, which increases the number of design process iterations and costs. Summary of the Invention
[0004] The purpose of this invention is to address at least some of the problems existing in the prior art by providing a target-based process design method under the dual influence of demand and constraints. This method is a process design target reasoning method that integrates demand and constraints using elements as intermediate variables. Through an element-target transformation method, a complete and clear expression of the design target is achieved through an element chain. The entire process can transform vague user requirements into executable process targets.
[0005] To achieve the above objectives, the present invention adopts the following technical solutions.
[0006] This invention provides a target-based process design method under the dual influence of demand and constraints, which includes the following steps:
[0007] S1, obtain user requirements information and process constraint information;
[0008] S2, transforming user demand information into corresponding elements; this step includes the following sub-steps:
[0009] S21, classify user demand information to obtain demand task elements;
[0010] S22, Based on the requirement task elements, construct the design requirement text;
[0011] S23, Based on the design requirements text, match the corresponding elements from the requirements-element mapping case library;
[0012] S24, cluster the matched features to obtain standard feature groups;
[0013] S3, standardize the process constraint information to obtain constraint groups;
[0014] S4, Construct constraint-feature groups for constraints and feature dependencies;
[0015] S5. Based on the constraint-element group, determine the relationship between each element, construct the element chain, and determine the design goal.
[0016] In step S1 above, relevant user needs and process constraints information are collected by consulting publicly available process literature, company documents, patent databases, and conducting in-depth communication with users, combined with actual observations and summaries of the company.
[0017] Step S2 above aims to transform user demand information into corresponding elements. These elements are defined intermediate variables, which can be categorized into physical elements, process elements, and media elements based on their attributes. Physical elements refer to physical parameters (such as forging temperature, die temperature, hammering pressure, hammering interval, etc.), equipment resources, human resources, and material resources. Process elements refer to factors related to the technological process, such as cooling time, furnace loading sequence, furnace loading quantity, and furnace loading direction. Media elements refer to factors related to the process medium, such as cooling method, lubrication method, and release agent selection.
[0018] In step S21 above, the structured template for user requirement information input is automatically categorized into: quality performance requirements, production efficiency requirements, and economic performance requirements. Through these three types of requirement input, the specific classification of requirements and corresponding requirement information are analyzed, and requirement task elements, which are quantifiable and executable process indicators, are constructed based on the categorized requirement information.
[0019] In step S22 above, design requirement text is constructed based on the requirement task elements and the categorized requirement information. The design requirement text generally includes four sub-features: requirement object, requirement category, requirement objective, and requirement content. Keywords for each feature can be extracted from the description text of the requirement task elements. Requirement content refers to the information in the design requirement text that meets user needs (including information communicated with users and information extracted by the designer from other conditions); requirement objective is the goal of achieving the user's requirement content.
[0020] In step S23 above, in order to reduce the empirical errors of the designers and improve the objectivity and efficiency of demand transformation, the present invention adopts a case-based reasoning method to realize demand-element transformation.
[0021] The process of constructing the requirement-element mapping case library is as follows: Cases involving requirement-element mapping from historical design cases are recorded as valid cases. These valid cases are derived from publicly available process documents or enterprise process documents and are developed by professional designers, ensuring high reliability and accuracy. For each valid case, the corresponding design requirements and the elements derived from the requirement mapping are extracted from the valid case design document, serving as the basic data source for the requirement-element mapping case library. To facilitate subsequent retrieval and calculation of the data in the case library, this invention also stores the data in a unified format. In the established case library, each data entry consists of a requirement description and an element description.
[0022] This step first employs the Case-Based Reasoning (CBR) method. Based on the design requirement text, similar requirement descriptions are matched from the requirement-element mapping case library, and corresponding elements are obtained. Then, the design requirement text and similar requirement descriptions are converted into vector representations, and their cosine similarity is calculated. Elements in requirement descriptions with a cosine similarity greater than a set element threshold are used as the corresponding elements of the design requirement text. Due to the similarity of requirements, elements in the element groups of the case data can serve as candidate elements that satisfy this design requirement mapping. Considering that both the design requirement text and similar requirement descriptions are recorded in TXT text format, natural language processing methods (such as Word2Vec, GloVe, BERT, RNN, LSTM, etc.) are used to process the design requirement text and similar requirement descriptions to obtain their corresponding vector representations, and the cosine similarity between them is calculated. Elements in requirement descriptions with a cosine similarity greater than a set element threshold are used as the corresponding elements of the design requirement text.
[0023] The purpose of step S24 above is to organize the elements. After obtaining the elements corresponding to the relevant design requirement text from the case through case matching, these elements still need to be further organized to construct the final output element group.
[0024] First, perform deduplication and supplementation operations on the elements. Since there may be some duplicate elements among the multiple elements matched based on different design requirements, deduplication is required to retain only a single element. At the same time, check whether all matched elements can fully cover all design requirement texts. If there are coverage gaps, the elements need to be supplemented. This can be done by re-matching cases for the uncovered requirements or by the designer supplementing the elements based on experience.
[0025] Then, the re-clustering and association of elements are carried out. The existing elements are clustered and divided according to the types of physical elements, process elements, and media elements, and integrated into physical element groups, process element groups, and media element groups, respectively.
[0026] In step S3 above, the process constraint information can be standardized based on technical standards and specifications, patent databases, factory process documents, and the experience of experts and workers to obtain constraint groups. Constraint groups include equipment constraints, process constraints, material constraints, and environmental constraints.
[0027] In step S4 above, the so-called constraint-feature dependency refers to the process of associating constraints with corresponding features through reasonable logical operations, thereby supplementing the features. Specifically, this requires executing a search procedure divided into two parts:
[0028] Part 1: Searching for dependent elements from the standard element set. The core principle is to find elements that can be further augmented with attributes or relationships, and then construct a set of dependent elements.
[0029] Part Two: Finding executable dependency design constraints from constraint variables. The core principle is to select design constraints corresponding to variables that are highly consistent with the target element, thereby forming a set of dependency constraints for the target element.
[0030] The search procedure can be determined by the designer based on their own knowledge and practical experience.
[0031] Step S4 includes the following sub-steps:
[0032] S41, retrieve the dependent starting point element from the standard element group; based on the starting point element, match one or more constraints to be attached from the constraint group; if the starting point element cannot be matched with a constraint, then the element is an independent element; at the same time, for any matched constraint, match one or more dependent endpoint elements from the standard element group.
[0033] Repeat the above operation for the remaining features in the standard feature group until all features are matched;
[0034] S42 associates the constraints with dependencies with the elements to establish dependent target elements.
[0035] In step S41, dependent starting elements are retrieved from the standard element group obtained from the demand transformation in step S2, i.e., elements that can be further added with attributes or relationships. This invention determines the starting element based on the information entropy of the elements, using the element with the lowest information entropy in the standard element group as the starting element. Then, constraints associated with the retrieved starting element are matched from the constraint group as constraints attached to the starting element; if no constraint is matched for the starting element, the element is considered an independent element. Simultaneously, for any matched constraint, one or more endpoint elements associated with that constraint are matched from the standard element group. This step uses constraints with a correlation degree greater than a set correlation degree threshold as constraints related to the element. This invention uses the cosine similarity between the element and the constraint as the correlation degree between the element and the constraint.
[0036] Repeat the above search and matching operations for the remaining elements in the standard element group to match all elements. Elements that do not match the constraints are considered independent elements.
[0037] It also assesses the impact of matching constraints and elements; based on the impact, it analyzes the priority of constraints. It evaluates the degree of change that design constraints may cause to the design objectives, identifying which constraints are critical and likely to have a significant impact on achieving the objectives. In this step, the impact... The calculation formula is:
[0038] ;
[0039] in, This represents the standard deviation of the i-th constraint; It represents the standard deviation of the j-th element under the influence of the i-th constraint.
[0040] In step S42, constraints and elements with dependency relationships are associated to establish dependency target elements. The key is to determine the dependency form between constraints and elements. Simultaneously, four dependency forms need to be defined based on the number of constraints and elements. The dependency rules (e.g., direct dependency, ordered constraint dependency, ordered element dependency, and cross-dependent constraint-element dependency) and the representation of the dependency target element differ for each dependency form. The four dependency forms are: one-to-one dependency, one-to-many dependency, many-to-one dependency, and many-to-many dependency.
[0041] Step S4 completes the constraint-element dependency operation. Through the constraint-element dependency process, the design goal, jointly defined by the elements and constraints, is finally obtained. This goal is presented in the form of an element chain formed by several constraints attached to the elements.
[0042] After the demand-element transformation and constraint-element dependency stages, although the basic element group of the design goal has been obtained, the elements within the group are still in a discrete state. The process design goal needs to be further transformed and generated from these discrete elements, and the element-goal transformation process is actually a process of organizing and associating elements. In this process, it is necessary to clarify which elements can be combined into a complete design goal, which is essentially an analysis of the relationships between elements.
[0043] In step S5 above, a complete element chain represents a design objective, and different elements in the element chain describe different aspects of the same design objective. The final overall design objective may include several independent complete element chains.
[0044] First, it is necessary to identify the relationships between elements. Elements can have various relationships, and different relationships will lead to different organizational forms between elements. This invention summarizes the possible relationships between elements and explains their meanings, specifically including the following four types of relationships: mutually independent relationships (no direct association), subordinate-inclusion relationships (including hierarchical dependencies such as subordinate and semi-subordinate), conflicting and mutually exclusive relationships (such as exclusionary and semi-exclusionary oppositional relationships), and conditional dependency relationships (such as sequential or parallel logical constraints). Among these, the association strength between two elements in a mutually independent relationship is 0; the association strength between two elements in a subordinate, exclusionary, sequential, or parallel relationship is 1; the association strength of a semi-subordinate or semi-exclusionary relationship is represented by a defined fuzzy number, with a value range of (0,1). Overall, this aims to systematically describe the interaction patterns between different elements, providing a structured perspective for relationship modeling and system analysis.
[0045] By establishing an element association matrix, various elements (such as physical, technical, and media elements) and constraints (such as equipment, materials, and process constraints) are integrated to clarify their dependencies. Then, using elements as nodes (which may have dependent constraints), the starting node of the element chain is selected, and initial design process parameters are determined based on search data, technical standards, patents, etc. Next, the element chain is established, and the design is gradually improved by dynamically adjusting the process parameters. Finally, through iterative verification (such as updating nodes and checking element independence), a complete set of element chains is output, forming an executable design goal. The entire process emphasizes the dynamic relationship between elements and constraints, ensuring a logical closed loop and operability from requirements to design goals.
[0046] Finally, the relationships between elements are treated as directed edges, and nodes with related relationships are connected in sequence to construct a constraint-dependent element chain to represent the design goal.
[0047] Based on the above analysis, step S5 includes the following sub-steps:
[0048] S51, Input Constraints - Feature Group;
[0049] S52, Determine the relationship between elements in a constraint group;
[0050] S53, Construct the element association matrix: Based on the association relationships between each element, determine the corresponding association strength and the number of relationship type identifiers, and map the element association relationships into an element association matrix;
[0051] S54, Select the starting node of the element chain;
[0052] S55, Establish an element chain: Based on the element association matrix, search for elements that are related to the starting node as new nodes in the element chain, and connect the two to form an element chain; repeat the above search until all elements with a non-zero association strength with the starting node have been found and added to the element chain as nodes.
[0053] S56, Update Nodes, Continue to Complete the Feature Chain Established in Step S55: Select a new starting node from the nodes connected to the starting node, and use it as the updated starting node; Repeat step S55, find the connected features and add them as nodes to the feature chain; Repeat the above search until the association strength between the updated starting node and the remaining features is 0, then stop the search and obtain a complete feature chain.
[0054] Remove elements contained in the constructed element chain from the constraint-feature group, reconstruct the element association matrix, and then repeat steps S54-S56 to construct the next complete element chain; until all elements are contained in the element chain or the remaining elements are independent of each other, and then output the set of complete element chains.
[0055] In step S52 above, the relationships between elements are obtained by judging the relationships of subordination, conflict and mutual exclusion, conditional dependence and mutual independence.
[0056] In step S53 above, the feature association matrix is represented as follows:
[0057]
[0058] in, as elements and Number of relation type identifiers; as elements and The strength of the association.
[0059] In step S54 above, the strategy for selecting the starting node is as follows:
[0060] Based on the elements corresponding to the row:
[0061] (1) The element corresponding to the row with the fewest zero elements in the element association matrix is taken as the starting node;
[0062] (2) If there are more than two rows with the fewest zero elements, the corresponding element is selected as the starting node according to the selection rule that physical elements are superior to process elements, and process elements are superior to medium elements.
[0063] (3) If there are more than two rows with the fewest zero elements under the physical element category, more than two rows with the fewest zero elements under the process element category, or more than two rows with the fewest zero elements under the medium element category, then the element with the smallest number under the same category shall be taken as the starting node.
[0064] Alternatively, the elements corresponding to the column can be used as the basis:
[0065] (1) The element corresponding to the column with the fewest zero elements in the element association matrix is taken as the starting node;
[0066] (2) If there are more than two columns with the fewest zero elements, the corresponding element is selected as the starting node according to the selection rule that physical elements are superior to process elements, and process elements are superior to medium elements.
[0067] (3) If there are more than two columns with the fewest zero elements under the physical element category, more than two columns with the fewest zero elements under the process element category, or more than two columns with the fewest zero elements under the medium element category, then the element with the smallest number under the same category shall be taken as the starting node.
[0068] Through the above steps, the transformation from element to objective is achieved. The final result of the element-objective transformation is to convert the element group into several complete element chains and a textual description of the element chains. In order for the design objective to be understood by the designer, it is also necessary to summarize the textual description corresponding to the design objective based on the elements contained in each complete element chain.
[0069] This element-based approach to expressing design objectives provides more complete information, more standardized and comprehensive descriptions, and a clearer hierarchical structure, while also ensuring the independence of each design objective. The resulting process design objectives can be described from different dimensions, including physical, technical, and environmental aspects. In the subsequent solution generation phase, the elements within these element-based process design objectives can be directly mapped to the corresponding design activities to facilitate solution generation.
[0070] Compared with the prior art, the present invention has the following beneficial effects:
[0071] 1) This invention improves the efficiency of demand transformation into process design objectives: The lack of a systematic mapping relationship between user needs (such as quality performance, production efficiency, etc.) and final process parameters leads to problems such as long process development cycles and high trial production costs; This invention automatically maps fuzzy needs into executable process parameters through a structured demand-element transformation model, reducing subjectivity;
[0072] 2) This invention generates process objectives by synergistically limiting multi-dimensional constraints: Traditional methods have difficulty considering multiple dimensions of constraints such as equipment constraints, process constraints, material constraints, and environmental constraints at the same time; This invention introduces elements as intermediate variables, and combines them with elements in the form of constraint-element dependency, which is convenient for subsequent transformation into complete process design objectives;
[0073] 3) This invention can improve dynamic adaptability: When faced with new materials (such as high-strength aluminum alloys) or special working conditions (such as aerospace precision parts), traditional methods are difficult to adjust the process chain in a timely manner; the method provided by this invention systematically integrates multi-dimensional requirements and constraints, and through the dynamic correlation between elements and constraints, it can realize the dynamic adjustment of the process route. It is an innovative method of knowledge-driven decision-making, which has certain feasibility and applicability. Attached Figure Description
[0074] Figure 1 This is a schematic diagram of the target element-based process design method based on the dual effects of demand and constraints of the present invention.
[0075] Figure 2 This is a schematic diagram of the principle of the target element-based process design method based on the dual effects of demand and constraints of the present invention.
[0076] Figure 3 A diagram illustrating the categorization of user needs information;
[0077] Figure 4 A flowchart illustrating the transformation of demand into factors of production;
[0078] Figure 5 A flowchart illustrating the process of building a standard element group;
[0079] Figure 6 A schematic diagram of the constraint-element dependency construction process;
[0080] Figure 7 A schematic diagram of the process for building the element chain;
[0081] Figure 8 This is a schematic diagram of the correlation strength analysis of elements.
[0082] Terminology Explanation
[0083] 1. Case-Based Reasoning (CBR): Case-based reasoning (CBR) is an experience-based artificial intelligence (AI) and problem-solving method. Its core idea is that similar problems have similar solutions. It solves new problems by drawing on past cases. It imitates the human experience-based reasoning method (such as doctors referring to past cases to diagnose new patients) and is widely used in fields such as engineering design, medical diagnosis, and fault handling. Detailed Implementation
[0084] The technical solutions of various embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0085] This embodiment uses a high-strength titanium alloy bladed disk (Ti-6Al-4V) for aero-engines as the target product. User requirements include excellent microstructure consistency, safety and reliability, and high dimensional accuracy. A specific process flow needs to be designed, but the effective information related to the equipment attributes expressed in the user requirements is limited, and challenges include high material deformation resistance, poor fluidity, and susceptibility to cracking. This example uses the forging process of a high-strength titanium alloy bladed disk (Ti-6Al-4V) for aero-engines to illustrate the application of the proposed process design method based on target elementization under the dual effects of requirements and constraints.
[0086] like Figure 1 As shown, this embodiment provides a target-based process design method under the dual influence of demand and constraints, which includes the following steps:
[0087] S1, obtain user requirements information and process constraint information;
[0088] S2 transforms user demand information into corresponding elements;
[0089] S3, standardize the process constraint information to obtain constraint groups;
[0090] S4, Construct constraint-feature groups for constraints and feature dependencies;
[0091] S5. Based on the constraint-element group, determine the relationship between each element, construct the element chain, and determine the design goal.
[0092] In step S1 above, relevant user needs and process constraints information are collected by consulting publicly available process literature, company documents, patent databases, and conducting in-depth communication with users, combined with actual observations and summaries of the company.
[0093] In step S2 above, the purpose is to transform user demand information into corresponding elements. These elements are defined intermediate variables, which can be categorized into physical elements, process elements, and media elements based on their attributes. Physical elements refer to physical parameters (such as forging temperature, die temperature, hammering pressure, hammering interval, etc.), equipment resources, human resources, and material resources. Process elements refer to factors related to the technological process, such as cooling time, furnace loading sequence, furnace loading quantity, and furnace loading direction. Media elements refer to factors related to the process medium, such as cooling method, lubrication method, and release agent selection.
[0094] This step includes the following sub-steps:
[0095] S21, classify user demand information to obtain demand task elements.
[0096] User requirements are input into a structured template, which is automatically categorized into three types: quality performance requirements, production efficiency requirements, and economic performance requirements. Based on these three types of input, the specific classification of requirements and corresponding requirement information are analyzed. Then, requirement task elements, which are quantifiable and actionable process indicators, are constructed according to the categorized requirement information.
[0097] Once a multi-dimensional requirement system for process optimization is established, it is also necessary to transform complex user requirements into quantifiable and actionable process indicators, i.e., requirement task elements, through a hierarchical and categorized approach. For example... Figure 3 As shown, for example, quality performance requirements, as the primary goal of process design, mainly include: improved mechanical properties (e.g., tensile strength ≥800MPa), microstructure control (e.g., grain size ASTM 6-8 grade), and dimensional accuracy requirements (e.g., gear tooth profile tolerance ±0.05mm); production efficiency requirements reflect manufacturing process efficiency, mainly including: process cycle optimization (e.g., merging multiple firing cycles to reduce man-hours by 30%) and improved resource utilization (e.g., increasing material utilization from 40% to 55%); economic performance requirements encompass cost elements throughout the entire product lifecycle, mainly including: direct cost control (e.g., mold life ≥50,000 pieces) and total lifecycle cost (including subsequent machining, maintenance, and other comprehensive costs).
[0098] In this embodiment, the user requirement information is classified to obtain the requirement task elements, as shown in Table 1.
[0099] Table 1 Design Target Information
[0100]
[0101] S22, Based on the requirement task elements, construct the design requirement text.
[0102] Based on the requirement task elements and the categorized requirement information, a design requirement text is constructed. This text typically includes four sub-features: requirement object, requirement category, requirement objective, and requirement content. Keywords for each feature can be extracted from the description text of the requirement task elements. Requirement content refers to the information in the design requirement text that meets user needs (including information communicated with users and information extracted by the designer from other contexts). Requirement objectives are the goals for achieving the user's required content.
[0103] S23, based on the design requirements text, match the corresponding elements from the requirements-element mapping case library.
[0104] The process of building the demand-element mapping case library is as follows: Figure 4 As shown, historical design cases involving requirement-element mapping are recorded as valid cases. These valid cases are derived from publicly available process documents or company process documents and are conducted by professional designers, ensuring high reliability and accuracy. For each valid case, the corresponding design requirements and the elements derived from the requirement mapping are extracted from the valid case design document, serving as the basic data source for the requirement-element mapping case library. To facilitate subsequent retrieval and calculation of the case library data, this invention also stores the data in a unified format. In the established case library, each data entry consists of a requirement description and an element description.
[0105] like Figure 4 As shown, this step first uses the case-based reasoning (CBR) method. Based on the design requirement text, similar requirement descriptions are matched from the requirement-element mapping case library, and the corresponding elements can be obtained. For the specific implementation method, please refer to the literature Textual case-based reasoning - Weber, Rosina O. et al-2005.
[0106] Then, the natural language processing method (Word2Vec) is used to convert the design requirement text and similar requirement descriptions into vector representations, which are represented as follows: and And calculate the cosine similarity between the two. The closer the cosine value is to 1, the closer the two vectors are; 1 indicates that the two vectors are the same. Elements with a value greater than the set element threshold are used as the corresponding elements in the design requirement text.
[0107] S24. Cluster the matched features to obtain standard feature groups.
[0108] After obtaining the elements corresponding to the relevant design requirements text from the case studies through case matching, these elements still need to be further organized to construct the final output element group.
[0109] like Figure 5 As shown, the first step is to perform deduplication and supplementation of elements. Since there may be some duplicate elements among the multiple elements matched based on different design requirements, deduplication is required to retain only a single element. At the same time, it is checked whether all matched elements can fully cover all design requirement texts. If there are coverage gaps, elements need to be supplemented. This can be done by re-matching cases for the uncovered requirements or by the designer supplementing elements based on experience.
[0110] Then, the re-clustering and association of elements are carried out. The existing elements are clustered and divided according to the types of physical elements, process elements, and media elements, and integrated into physical element groups, process element groups, and media element groups, respectively.
[0111] In this embodiment, step S2 transforms the requirements of the target product into elements of the design target, and the results are shown in Table 2.
[0112] Table 2 Demand-Factor Transformation Results
[0113]
[0114] In step S3 above, the process constraint information can be standardized based on technical standards and specifications, patent databases, factory process documents, expert and worker experience, etc., to obtain constraint groups.
[0115] In this embodiment, the constraint groups obtained by standardizing the process constraint information are shown in Table 3.
[0116] Table 3. Results of Design Constraint Normalization
[0117]
[0118] In step S4 above, through reasonable logical operations, constraints are associated with corresponding elements, thereby achieving the process of supplementing elements. For example... Figure 6 As shown, this step mainly considers five aspects: retrieval, matching, evaluation, analysis, and integration (association).
[0119] This step includes the following sub-steps:
[0120] S41, retrieve the dependent starting point element from the standard element group; based on the starting point element, match one or more constraints to be attached from the constraint group; if the starting point element cannot be matched with a constraint, then the element is an independent element; at the same time, for any matched constraint, match one or more dependent endpoint elements from the standard element group.
[0121] In this embodiment, elements 1-14 are all dependent elements.
[0122] In step S41, a starting element that can be attached is retrieved from the standard element group obtained from the demand transformation in step S2, that is, an element that can be further added with attributes or relationships. This invention determines the starting element to be retrieved based on the element complexity (here, information entropy is used to represent complexity), and takes the element with the lowest information entropy in the standard element group as the starting element.
[0123] Each element contains more than one engineering attribute (such as process method, process parameters, raw materials, processing equipment, personnel configuration, etc.). The information entropy calculation formula is as follows:
[0124] ;
[0125] in, This represents the probability that the k-th engineering attribute among the elements will be selected. ; Indicates the number of project attributes.
[0126] Then, constraints associated with the retrieved starting element are matched from the constraint group as constraints attached to the starting element; if no constraint is matched with the starting element, the element is considered an independent element. Simultaneously, for any matched constraint, one or more ending elements associated with that constraint are matched from the standard element group.
[0127] In this embodiment, the cosine similarity between an element and a constraint is used as the correlation between the element and the constraint. Constraints whose correlation between an element and a constraint is greater than a set correlation threshold are considered as constraints related to the element; if the correlation between an element and any constraint is not greater than the set correlation threshold, it means that the element cannot match any constraint and is an independent element.
[0128] After completing the above operations, one round of constraint-feature constraint is finished. Then, repeat the above search and matching operations on the remaining features in the standard feature group to achieve matching for all features. Features that cannot match any constraints are considered independent features.
[0129] This step can also evaluate the impact of the matched constraints and elements; based on the impact, the priority of the constraints can be analyzed.
[0130] In this step, the degree of influence The calculation formula is:
[0131] ;
[0132] in, This represents the standard deviation of the i-th constraint; It represents the standard deviation of the j-th element under the influence of the i-th constraint.
[0133] Based on the degree of impact, assess the extent to which design constraints may change the design objectives, and determine which constraints are critical and may have a significant impact on the achievement of the objectives.
[0134] S42 associates the constraints with dependencies with the elements to establish dependent target elements.
[0135] This step primarily involves connecting constraints and elements with dependency relationships. It also requires defining four dependency forms based on the number of constraints and elements. The dependency rules (e.g., direct dependency, ordered constraint dependency, ordered element dependency, and cross-dependent constraint-element dependency) and the representation of the dependency target element differ for each dependency form. The four dependency forms are: one-to-one dependency, one-to-many dependency, many-to-one dependency, and many-to-many dependency. One-to-one dependency refers to the dependency of one element on one constraint; one-to-many dependency refers to the dependency of multiple elements on one constraint; many-to-one dependency refers to the dependency of one element on multiple constraints; and many-to-many dependency refers to the dependency of multiple elements on multiple constraints.
[0136] In this embodiment, based on the material processing steps, the elements and the constraints with dependencies are connected to obtain the corresponding dependent target elements, as shown in Table 4.
[0137] Table 4 Results of Attached Target Elements
[0138]
[0139] In step S5 above, a complete element chain represents a design objective, and different elements in the element chain describe different aspects of the same design objective. The final overall design objective may include several independent complete element chains.
[0140] like Figure 7 As shown, this step includes the following sub-steps:
[0141] S51, Input constraints - feature group.
[0142] S52, Determine the constraints - the relationships between elements in the element group.
[0143] In this step, the relationships between the various elements are defined. For example... Figure 8 As shown, the relationships between the elements specifically include the following four types of relationships: mutual independence (no direct relationship), subordinate and inclusive relationship (including subordinate, semi-subordinate and other hierarchical dependencies), conflicting and mutually exclusive relationship (such as exclusionary and semi-exclusionary oppositional relationships), and conditional dependency relationship (such as sequential or parallel logical constraints).
[0144] S53, Construct the element association matrix: Based on the association relationships between each element, determine the corresponding association strength and the number of relationship type identifiers, and map the element association relationships into an element association matrix.
[0145] Based on the defined relationships between the elements, the strength of the relationship between the two is determined. In this embodiment, for example... Figure 8 As shown, the association strength between two elements with an independent relationship is 0; the association strength between two elements with a subordinate, exclusive, sequential, or parallel relationship is 1; the association strength of a semi-subordinate or semi-exclusive relationship is represented by a defined fuzzy number, which takes values in the range of (0,1).
[0146] Simultaneously, it is necessary to define the number of relationship types between the various elements. Among these, subordinate and sequential relationships are unidirectional, exhibiting directionality in calculation. The other types of relationships are all bidirectional, and the direction is not distinguished during calculation. The direction of the relationship is indicated by the sign of the identifier number.
[0147] In this embodiment, the association strength and relationship type identifier number between the two elements are defined as shown in Table 5.
[0148] Table 5. Relationship between two elements, number of identifiers, and correlation strength.
[0149]
[0150] Then, an element association matrix is constructed based on the association strength and identifier number between each element. The elements in the matrix are obtained by multiplying the association strength and identifier number between any two corresponding elements. By establishing the element association matrix, various types of elements (such as physical elements, process elements, and media elements) are integrated, and their dependencies are clarified.
[0151] The element association matrix constructed in this embodiment is shown below:
[0152]
[0153] Taking element 1 as an example, according to the relationship matrix, the elements associated with element 1 are: element 2 (semi-subordinate relationship, association strength is 0.2), element 7 (sequential relationship), element 8 (sequential relationship), and element 9 (sequential relationship). The remaining elements are independent of element 1. Taking element 5 as an example, according to the relationship matrix, the elements associated with element 5 are: element 3 (subordinate relationship, element 5 contains element 3), element 4 (subordinate relationship, element 5 contains element 4), and element 6 (subordinate relationship, element 5 contains element 6).
[0154] S54, Select the starting node of the element chain.
[0155] In this embodiment, the selection strategy for the starting node is as follows:
[0156] Based on the elements corresponding to the row:
[0157] (1) The element corresponding to the row with the fewest zero elements in the element association matrix is taken as the starting node;
[0158] (2) If there are more than two rows with the fewest zero elements, the corresponding element is selected as the starting node according to the selection rule that physical elements are superior to process elements, and process elements are superior to medium elements.
[0159] (3) If there are more than two rows with the fewest zero elements under the physical element category, more than two rows with the fewest zero elements under the process element category, or more than two rows with the fewest zero elements under the medium element category, then the element with the smallest number under the same category shall be taken as the starting node.
[0160] Alternatively, the elements corresponding to the column can be used as the basis:
[0161] (1) The element corresponding to the column with the fewest zero elements in the element association matrix is taken as the starting node;
[0162] (2) If there are more than two columns with the fewest zero elements, the corresponding element is selected as the starting node according to the selection rule that physical elements are superior to process elements, and process elements are superior to medium elements.
[0163] (3) If there are more than two columns with the fewest zero elements under the physical element category, more than two columns with the fewest zero elements under the process element category, or more than two columns with the fewest zero elements under the medium element category, then the element with the smallest number under the same category shall be taken as the starting node.
[0164] For example, based on the element association matrix given above, taking the elements corresponding to the rows as the baseline, the row with the fewest 0 elements corresponds to element E6, so element E6 is taken as the starting node P0.
[0165] S55, Establish an element chain: Based on the element association matrix, search for elements that are related to the starting node as new nodes in the element chain, and connect the two to form an element chain; repeat the above search until all elements with a non-zero association strength with the starting node have been found and added to the element chain as nodes.
[0166] In this step, based on the element association matrix, elements that are associated with element P0 are searched and added as new nodes P1 in the element chain. P0 and P1 are then connected to form the element chain. This search is repeated until all elements with a non-zero association strength with P0 have been found and added as nodes to the element chain chain1.
[0167] S56, Update Nodes and Continue to Complete the Feature Chain Established in Step S55: Select a new starting node from the nodes connected to the starting node and use it as the updated starting node; Repeat step S55 to find connected features and add them as nodes to the feature chain; Repeat the above search until the association strength between the updated starting node and the remaining features is 0, then stop the search and obtain a complete feature chain.
[0168] In this step, following the selection strategy for the starting node given earlier, elements P other than P0 are selected from the nodes connected to node P0. i The starting node for the update is selected. After selecting the starting node, repeat step S55 to find connected features and add them as nodes to the feature chain chain1. Repeat the above search until the association strength between the feature corresponding to the updated starting node and the remaining features is 0, then stop the search. At this point, feature chain chain1 is a complete feature chain, as shown in Table 6.
[0169] Table 6. List of Design Objectives for High-Strength Titanium Alloy Bladed Disk Forging Process
[0170]
[0171] Remove elements contained in the constructed element chain from the constraint-feature group, reconstruct the element association matrix, and then repeat steps S54-S56 to construct the next complete element chain; until all elements are contained in the element chain or the remaining elements are independent of each other, and then output the set of complete element chains.
[0172] In this step, the next complete element chain obtained is chain2, as shown in Table 6.
[0173] S59 outputs the complete set of element chains.
[0174] Therefore, through the above steps, the transformation from element to objective is achieved. The final result of the element-objective transformation is to convert the element group into several complete element chains and a textual description of the element chains. In order for the design objective to be understood by the designer, it is also necessary to summarize the textual description corresponding to the design objective based on the elements contained in each complete element chain.
[0175] This embodiment, based on the steps of the feature chain construction procedure, yields two complete feature chains with constraint attachments:
[0176] chain1: Composed of elements 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, and 12 connected by their relationships, with constraints 1, 2, 3, 4, 5, 6, 7, 8, and 9 attached to it, representing design objective 1: forging process of high-strength titanium alloy bladed disk.
[0177] chain2: It is composed of elements 13 and 14 connected by their relationship. The constraints attached to it are constraints 9 and 10, which represent design goal 2: lubrication and parts handling.
[0178] Through the above process, the design objectives of the forging process of this high-strength titanium alloy bladed disk were analyzed, and two design objectives composed of elements and constraints were obtained, which can comprehensively describe the attribute characteristics that the product needs to have.
[0179] In summary, the design objectives expressed by the element-based method in this invention encompass more complete information, are more standardized and comprehensive in description, and have a clearer hierarchical structure, while also ensuring the independence of each design objective. The resulting process design objectives can be described from different dimensions, including physical, technical, and environmental aspects. In the subsequent solution generation stage, the elements in these element-based process design objectives can be directly mapped to the corresponding design activities to carry out solution generation work.
Claims
1. A target-based process design method based on the dual effects of demand and constraints, characterized in that, Includes the following steps: S1, obtain user requirements information and process constraint information; S2, transforming user demand information into corresponding elements; this step includes the following sub-steps: S21, classify user demand information to obtain demand task elements; S22, Based on the requirement task elements, construct the design requirement text; S23, Based on the design requirements text, match the corresponding elements from the requirements-element mapping case library; S24, cluster the matched features to obtain standard feature groups; S3, standardize the process constraint information to obtain constraint groups; S4, Construct constraint-feature groups for constraints and feature dependencies; S5, based on the updated standard element group, determines the relationships between elements, constructs element chains, and determines design objectives.
2. The target element-based process design method based on the dual effects of demand and constraints as described in claim 1, characterized in that, Elements are intermediate variables in the definition, and are divided into physical elements, process elements, and media elements according to their attributes.
3. The target element-based process design method based on the dual effects of demand and constraints as described in claim 1, characterized in that, In step S21, user demand information is classified into: quality performance demand, production efficiency demand and economic performance demand, and demand task elements are constructed based on the classified demand information.
4. The target element-based process design method based on the dual effects of demand and constraints as described in claim 1, characterized in that, In step S23, the process of constructing the requirement-element mapping case library is as follows: record the cases involving requirement-element mapping in the historical design cases as valid cases; for each valid case, extract the corresponding design requirements and the elements obtained by requirement mapping from the valid case design document, and use this as the basic data source of the requirement-element mapping case library. Each piece of data consists of requirement description and element description.
5. The target element-based process design method based on the dual effects of demand and constraints as described in claim 4, characterized in that, First, the Case-Based Reasoning (CBR) method is used to match similar requirement descriptions from the requirement-element mapping case library based on the design requirement text, and the corresponding elements can be obtained. Then, the design requirement text and similar requirement descriptions are converted into vector representations, and the cosine similarity between the two is calculated. Elements in the requirement descriptions with a cosine similarity greater than a set element threshold are used as the corresponding elements of the design requirement text.
6. The target element-based process design method based on the dual effects of demand and constraints as described in claim 1, characterized in that, In step S24, the deduplication and addition of elements are performed first; then, the existing elements are clustered and divided according to the types of physical elements, process elements, and media elements, and integrated into physical element groups, process element groups, and media element groups, respectively.
7. The target element-based process design method based on the dual effects of demand and constraints as described in claim 1, characterized in that, Step S4 includes the following sub-steps: S41, retrieve the dependent starting point element from the standard element group; based on the starting point element, match one or more constraints to be attached from the constraint group; if the starting point element cannot be matched with a constraint, then the element is an independent element; at the same time, for any matched constraint, match one or more dependent endpoint elements from the standard element group. Repeat the above operation for the remaining features in the standard feature group until all features are matched; S42 associates the constraints with dependencies with the elements to establish dependent target elements.
8. The target element-based process design method based on the dual effects of demand and constraints as described in claim 7, characterized in that, In step S41, the influence of matching constraints and elements is also evaluated; based on the influence, the priority of constraints is analyzed.
9. The target element-based process design method based on the dual effects of demand and constraints as described in claim 8, characterized in that, Constraints whose correlation with a feature is greater than a set correlation threshold are considered as constraints related to the feature; the cosine similarity between a feature and a constraint is used as the correlation between the feature and the constraint. Impact The calculation formula is: ； in, This represents the standard deviation of the i-th constraint; It represents the standard deviation of the j-th element under the influence of the i-th constraint.
10. The target element-based process design method based on the dual effects of demand and constraints according to any one of claims 1 to 9, characterized in that, Step S5 It includes the following steps: S51, Input constraints - feature group; S52, Determine the relationship between elements in a constraint group; S53, Construct the element association matrix: Based on the association relationships between each element, determine the corresponding association strength and the number of relationship type identifiers, and map the element association relationships into an element association matrix; S54, Select the starting node of the element chain; S55, Establish an element chain: Based on the element association matrix, search for elements that are related to the starting node as new nodes in the element chain, and connect the two to form an element chain; repeat the above search until all elements with a non-zero association strength with the starting node have been found and added to the element chain as nodes. S56, Update the node and continue to supplement and complete the element chain established in step S55: Select a new initial node from the nodes connected to the starting node as the updated starting node; Repeat step S55 to find the connected elements and add them to the element chain as nodes; Repeat the above search until the association strength between the element corresponding to the updated initial node and the remaining elements is 0, then stop the search and obtain a complete element chain. Remove elements contained in the constructed element chain from the constraint-feature group, reconstruct the element association matrix, and then repeat steps S54-S56 to construct the next complete element chain; until all elements are contained in the element chain or the remaining elements are independent of each other, and then output the set of complete element chains.
11. The target element-based process design method based on the dual effects of demand and constraints as described in claim 10, characterized in that, In step S52, the relationships between the elements are obtained by judging the relationships of subordinate inclusion, conflict and mutual exclusion, conditional dependence and mutual independence.
12. The target element-based process design method based on the dual effects of demand and constraints as described in claim 10, characterized in that, In steps S54 and S56, the strategy for selecting the starting node is as follows: Based on the elements corresponding to the row: (1) The element corresponding to the row with the fewest zero elements in the element association matrix is taken as the starting node; (2) If there are more than two rows with the fewest zero elements, the corresponding element is selected as the starting node according to the selection rule that physical elements are superior to process elements, and process elements are superior to medium elements. (3) If there are more than two rows with the fewest zero elements under the physical element category, more than two rows with the fewest zero elements under the process element category, or more than two rows with the fewest zero elements under the medium element category, then the element with the smallest number under the same category shall be taken as the starting node. Alternatively, the elements corresponding to the column can be used as the basis: (1) The element corresponding to the column with the fewest zero elements in the element association matrix is taken as the starting node; (2) If there are more than two columns with the fewest zero elements, the corresponding element is selected as the starting node according to the selection rule that physical elements are superior to process elements, and process elements are superior to medium elements. (3) If there are more than two columns with the fewest zero elements under the physical element category, more than two columns with the fewest zero elements under the process element category, or more than two columns with the fewest zero elements under the medium element category, then the element with the smallest number under the same category shall be taken as the starting node.