A technical document association search method and system

Abstract

The application discloses a kind of technical document association retrieval method and system, it is related to data retrieval technical field.A kind of technical document association retrieval system, including have: document structured module, relationship identification module, trigger rule module, association network module, propagation confidence module, association verification module and result output module.The application is by executing relationship state propagation calculation in multidimensional association network, makes association relationship along network structure gradually expands, and state determination is carried out in combination with relationship trigger rule model, realizes the automatic derivation of complex association path.Through the combination calculation to relationship confidence parameter, and in the recursive updating of relationship propagation process, the reliable degree of association relationship in propagation path is gradually transmitted and adjusted.

Description

Technical Field

[0001] This invention relates to the field of data retrieval technology, and in particular to a method and system for associating and retrieving technical documents. Background Technology

[0002] With the continuous expansion of industrial equipment scale and the increasing complexity of operating conditions, a large number of technical documents have been generated surrounding the design, installation, operation, and maintenance processes of equipment. These technical documents typically cover information such as technical parameters, operating constraints, applicable conditions, and version evolution, and are continuously updated from different sources at different stages. In practical applications, the same technical object corresponds to multiple document versions or supplementary explanations, and there are relationships between different documents such as inheritance, substitution, or applicable conditions. Therefore, how to accurately extract technical information matching specific operating conditions from a large number of technical documents and identify the inherent relationships between documents has become a key issue in technical document management and application.

[0003] In existing technologies, the processing of technical documents often employs retrieval methods based on keyword matching or simple classification, focusing on the degree of textual matching of document content and lacking the ability to comprehensively analyze structured information such as technical parameters, technical constraints, and applicable conditions. When there are version substitution relationships, conditional exception rules, or constraint conflicts between different documents, existing methods struggle to effectively identify the mechanisms of these relationships. Especially when multiple operating conditions overlap, problems such as inaccurate search results, inconsistent judgments of association relationships, or neglect of key constraints can easily arise, thus affecting the reliability of technical decisions. Summary of the Invention

[0004] This invention proposes a method for unified modeling and comprehensive analysis of complex relationships between technical documents in scenarios involving multiple documents, multiple constraints, and multiple conditions. This method aims to solve the technical problem of determining the relationships between technical documents under multiple document and multiple constraint conditions and making consistency judgments on the technical parameters and constraints involved in specific application scenarios.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A technical document association retrieval method, applied in a technical document management system, is executed by a processor and includes: Multiple technical documents are acquired and parsed to obtain a structured set of technical elements, which includes technical object information, technical parameter information, technical constraint information, and document attribute information. Based on a set of structured technical elements, the basic relationships between technical documents are identified. These basic relationships include exception handling relationships, substitution inheritance relationships, and constraint conflict relationships. Relationship confidence parameters are then established for these basic relationships. Based on the basic relationships and their corresponding relationship confidence parameters, a relationship triggering rule model is constructed; and by combining the time attributes and applicable condition attributes of the basic relationships, a multi-dimensional relationship network among technical documents is constructed. In response to a retrieval request, the target technology object and target application scenario parameters are determined, relation state propagation calculation is performed in a multidimensional association network, and the relation confidence parameters in the propagation process are combined and calculated to obtain candidate associations and propagation confidence. The structural validity is obtained based on candidate association relationships. A parameter coupling matrix is ​​constructed based on the technical parameter information and technical constraint information in the structured technical element set. The parameter consistency is obtained based on the parameter coupling matrix, and the association validity of the candidate association relationships is determined. Based on the target application scenario parameters, propagation confidence, and association validity, valid associations are filtered, and search results are generated and output.

[0006] As a preferred technical solution of the present invention, the identification of the basic relationship includes: based on the technical object information in the structured technical element set, performing association matching on the same technical object involved in different technical documents; based on the technical object matching, performing comparative analysis based on technical parameter information and technical constraint information, and determining that there is an exception handling relationship between the first technical document and the second technical document when the technical parameter information or technical constraint information in the first technical document deviates from the corresponding information in the second technical document and has applicable conditions; determining that there is a substitution inheritance relationship between the first technical document and the second technical document when the technical object in the first technical document corresponds to the technical object in the second technical document, and the technical parameter information in the first technical document substitutes for or inherits the technical parameter information in the second technical document; and determining that there is a constraint conflict relationship between the different technical documents when the technical constraint information in different technical documents involving the same technical object is inconsistent.

[0007] As a preferred technical solution of the present invention, the establishment of the relationship confidence parameter includes: for exception handling relationships, substitution inheritance relationships, and constraint conflict relationships, determining the matching degree of the corresponding basic relationships based on technical parameter information, technical constraint information, and document attribute information in the structured technical element set; the matching degree is determined based on the degree of difference between technical parameter information involving the same technical object, the degree of consistency between technical constraint information, and the time attribute in the document attribute information; according to the matching degree, assigning a corresponding relationship confidence parameter to each basic relationship to characterize the credibility of the corresponding basic relationship.

[0008] As a preferred embodiment of the present invention, the construction of the relationship triggering rule model includes: defining triggering conditions for different types of basic relationships based on the association between exception handling relationships, substitution inheritance relationships, and constraint conflict relationships; when a constraint conflict relationship involving the same technical object exists, determining the triggering conditions for the exception handling relationship based on technical constraint information and document attribute information; when a substitution inheritance relationship exists, determining the state change conditions of the technical constraint information corresponding to the substitution inheritance relationship based on technical parameter information and technical constraint information; when the exception handling relationship meets its applicable conditions, determining the state change conditions of the substitution inheritance relationship based on document attribute information; and establishing an association mapping between different types of basic relationships based on the triggering conditions and state change conditions to construct the relationship triggering rule model.

[0009] As a preferred technical solution of the present invention, the construction of the multidimensional association network includes: based on the basic relationship, taking technical documents as nodes, and taking exception handling relationship, substitution inheritance relationship and constraint conflict relationship as the association relationship between nodes; associating each association relationship with the corresponding relationship confidence parameter, time attribute and applicable condition attribute; and establishing a corresponding relationship between the relationship triggering rule model and the association relationship, so that the association relationship has the ability to change state based on the relationship triggering rule model, forming a multidimensional association network for relationship state propagation calculation.

[0010] As a preferred embodiment of the present invention, the execution of the relationship state propagation calculation includes: determining the basic relationship directly associated with the technical document corresponding to the target technical object in a multidimensional association network based on the target technical object; determining the state of the directly associated basic relationship based on the relationship triggering rule model, and determining the extended association relationship triggered by the directly associated basic relationship; for the extended association relationship, continuing to determine the state based on its corresponding relationship triggering rule model, so as to determine the subsequent extended association relationship step by step along the multidimensional association network; and determining the candidate association relationship corresponding to the target technical object based on the state change results of each association relationship during the relationship state propagation calculation process.

[0011] As a preferred embodiment of the present invention, the combined calculation of the relationship confidence parameters includes: obtaining the relationship confidence parameters corresponding to the current association during the relationship state propagation calculation process, and the relationship confidence parameters corresponding to the extended association triggered by the current association; based on the numerical correlation between the relationship confidence parameters corresponding to the current association and the relationship confidence parameters corresponding to the extended association, performing combined calculations according to preset combination rules to obtain the propagation confidence corresponding to the extended association; when the relationship state propagation calculation proceeds step by step along the multidimensional association network, updating the propagation confidence corresponding to the subsequent extended association based on the recursive relationship between the propagation confidence corresponding to the previous level association and the relationship confidence parameters corresponding to the current level association.

[0012] As a preferred embodiment of the present invention, the construction of the parameter coupling matrix includes: extracting parameter variables and constraint relationships between parameters corresponding to each technical object based on technical parameter information and technical constraint information in a structured set of technical elements; performing parameter correspondence analysis on parameter variables corresponding to different technical objects to determine the coupling relationship between parameters of different technical objects; constructing a parameter coupling matrix based on the constraint relationship between parameters and the coupling relationship between parameters, wherein the matrix elements of the parameter coupling matrix are used to characterize the degree of coupling between parameters of different technical objects; and determining the corresponding matrix elements of the technical objects involved in the candidate association relationship in the parameter coupling matrix based on the candidate association relationship.

[0013] As a preferred embodiment of the present invention, the determination of the association validity includes: analyzing the propagation path of candidate associations in a multidimensional association network, and determining the structural validity based on the changes in connectivity or path characteristics of the propagation paths of candidate associations before and after removing intermediate associations; obtaining the corresponding matrix elements of the technical objects involved in the candidate associations in the parameter coupling matrix, and calculating the degree of parameter constraint consistency between technical objects based on the degree of parameter coupling represented by the matrix elements to characterize parameter consistency; adjusting the parameter consistency by weighting based on the propagation confidence corresponding to the candidate associations to obtain corrected parameter consistency; and determining the candidate association as a valid association when the structural validity meets a preset condition and the corrected parameter consistency is greater than a preset threshold.

[0014] A technical document association retrieval system, comprising: Document structuring module: Retrieves multiple technical documents, parses them, and obtains a set of structured technical elements; Relationship identification module: Based on a structured set of technical elements, it identifies the fundamental relationships between technical documents; Triggering rule module: Constructs a relationship triggering rule model based on the basic relationship and the corresponding relationship confidence parameters; The network module combines the time and applicable condition attributes of the basic relationships to construct a multi-dimensional network of relationships between technical documents. Propagation confidence module: In response to the retrieval request, it determines the target technology object and target application scenario parameters, performs relation state propagation calculation in the multidimensional association network, and combines the relation confidence parameters in the propagation process to obtain candidate associations and propagation confidence. The association verification module obtains structural validity based on candidate association relationships, constructs a parameter coupling matrix based on technical parameter information and technical constraint information in the structured technical element set, obtains parameter consistency based on the parameter coupling matrix, and determines the association validity of candidate association relationships. Results output module: Based on the target application scenario parameters, propagation confidence, and association validity, it filters valid associations and generates and outputs search results.

[0015] The present invention has the following advantages: This invention identifies exception handling relationships, substitution inheritance relationships, and constraint conflict relationships based on a structured set of technical elements, achieving unified modeling of multiple types of relationships between technical documents. This accurately reflects the relationships between different documents in terms of parameter changes, version evolution, and constraint differences. By establishing relationship confidence parameters and performing comprehensive calculations based on the degree of difference in technical parameters, the degree of consistency in technical constraints, and document time attributes, different basic relationships possess quantifiable levels of credibility.

[0016] This invention constructs a relationship triggering rule model to map the mechanisms of action between exception handling relationships, substitution inheritance relationships, and constraint conflict relationships, enabling different types of relationships to trigger and influence each other under specific conditions. By constructing a multi-dimensional association network that includes relationship type, time attribute, and applicable condition attribute, technical documents and their associations are uniformly represented as a network structure with multi-dimensional attributes, making the associations between documents scalable and traceable.

[0017] This invention achieves automatic derivation of complex association paths by performing relation state propagation calculations in a multidimensional association network, allowing the associations to expand step-by-step along the network structure, and combining this with a relation triggering rule model for state determination. By combining and calculating relation confidence parameters and recursively updating them during relation propagation, the confidence level of associations in the propagation path is progressively transmitted and adjusted.

[0018] This invention constructs a parameter coupling matrix to uniformly model the parameter variables and their constraints between different technical objects, thereby achieving a quantitative expression of the coupling relationship between parameters. By calculating parameter consistency based on the parameter coupling matrix and correcting it in conjunction with propagation confidence, the rationality of candidate associations at the parameter level can be comprehensively evaluated. By performing structural analysis on the propagation path of candidate associations and determining structural validity based on changes in path connectivity, the determination of associations not only relies on numerical calculations but also incorporates network structural characteristics.

[0019] This invention combines structured technical elements, relationship propagation mechanisms, and parameter coupling analysis to form a complete data processing flow, realizing closed-loop processing from technical document parsing to relational output, thereby improving the overall efficiency and consistency of the technical document association retrieval process. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings in the following description are only schematic diagrams of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort. Figure 1 This is a schematic diagram of the structure of a technical document association retrieval system used in an embodiment of the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0022] Example 1: A technical document association retrieval method, applied in a technical document management system, is executed by a processor and includes the following steps: Step S1: Obtain multiple technical documents and parse them to obtain a structured set of technical elements, which includes technical object information, technical parameter information, technical constraint information, and document attribute information; In one embodiment of the present invention, the step of "acquiring multiple technical documents" refers to reading document data related to the industrial system (such as an "industrial cooling pump system") from a pre-stored set of technical documents. This document data includes design specification documents, operation and maintenance documents, installation process documents, exception handling documents, and model upgrade documents. Specifically, the technical documents include documents such as the "P-210 Cooling Pump Design Specification," "Supplementary Instructions for High-Temperature Operation," "P-210B Upgrade Instructions," "Low Vibration Installation Requirements," and "High Flow Rate Operation Guidelines."

[0023] The "parse of technical documents" refers to the structuring of the acquired technical documents, transforming the original unstructured or semi-structured text into data records for subsequent calculations. The parsing process is based on preset technical element identification rules, which are used to identify fields related to technical objects, technical parameters, and constraints from the document. The parsed results are stored in the form of structured technical element records, with each record corresponding to a set of descriptive information about a specific technical object in the document. To ensure consistency in subsequent relationship identification and calculation, all types of data in the structured technical element set adopt a unified data field organization method, specifically including the following: The technical object information is used to characterize the entity objects involved in the technical document and serves as the foundational data for subsequent cross-document matching and association. The technical object information originates from the object name, component description, and model identifier fields in the technical document and is obtained through parsing. In this embodiment, the technical object information includes: object name, such as "cooling pump P-210", "motor M-21", "sealing assembly S-21", and "bearing assembly B-21"; object identifier, such as equipment model code or component number; object type, such as complete machine, component, or sub-assembly; object hierarchy, such as cooling pump P-210 including motor, impeller, and sealing assembly; the system to which the object belongs, such as all belonging to the industrial cooling pump system; and object version information, such as the model evolution relationship between P-210 and P-210B. For example, in the "P-210 Cooling Pump Design Specification", a technical object record is obtained: Object name: Cooling pump P-210; Object type: Complete machine; System: Industrial cooling pump system; Sub-objects included: Motor M-21, Sealing assembly S-21; This type of information is used for subsequent technical object matching between different documents.

[0024] The technical parameter information is used to characterize the performance indicators and operating characteristics of the technical object, and is derived from parameter description paragraphs, table data, or design specification descriptions in technical documents. The technical parameter information is organized in the form of parameter items, each of which includes at least: parameter name; parameter value; parameter unit; the technical object to which it belongs; and applicable conditions (if any). In this embodiment, the following parameter information is obtained from the analysis of the P-210 cooling pump: rated flow rate: 120 m³ / h; rated head: 45 m; rated speed: 2900 rpm; motor power: 15 kW (corresponding to 18.5 kW in P-210B); operating temperature range: 5℃~85℃; vibration limit: ≤2.8 mm / s. The above parameter information is derived from design specifications and operating guidance documents, and is stored using a unified field for subsequent parameter difference analysis and coupling calculations.

[0025] The technical constraint information is used to characterize the restrictive relationships that a technical object must meet under specific conditions. Unlike technical parameter information, technical constraint information emphasizes conditional constraints or restrictive rules. Technical constraint information originates from restrictive descriptions, precautions, operating procedures, or safety requirements in technical documents, and the correspondence between "conditions" and "constraint content" is identified through parsing. In this embodiment, the parsed technical constraint information includes: under continuous operating conditions, the bearing temperature rise must not exceed 40°C; when the medium temperature is higher than 70°C, the sealing assembly must use the high-temperature resistant model S-21H; in low-vibration scenarios, the vibration value must not exceed 2.0 mm / s; when the flow rate is greater than 100 m³ / h, the motor power must be greater than or equal to 15 kW. Each piece of technical constraint information includes: the constrained object (e.g., sealing assembly, bearing assembly); the constraint condition (e.g., "medium temperature > 70°C"); the constraint content (e.g., "use model S-21H"); and the threshold or range (e.g., temperature rise ≤ 40°C). This type of data is used for subsequent identification of constraint conflicts and judgment of parameter consistency.

[0026] The document attribute information describes the characteristic attributes of the technical document itself and is an important basis for judging the applicability and priority of relationships. Document attribute information originates from document metadata and explanatory information in the text. In this embodiment, document attribute information includes: document number, such as DOC-P210-001; document name, such as "P-210 Cooling Pump Design Specification"; document type, such as design specification, supplementary instructions, upgrade instructions; publication date and effective date; document version information, such as V1.0, V2.0; scope of application, such as normal temperature conditions or high temperature conditions; document status, such as currently valid or superseded; priority attributes, such as mandatory or recommended. For example, the document attributes of "Supplementary Instructions for High Temperature Operation" are parsed as: document type: supplementary instruction document; scope of application: operating conditions with medium temperature higher than 70℃; effective date: later than design specification; priority: under specific conditions, priority over general specifications. This type of information is used in subsequent steps to determine exception handling relationships and relationship triggering conditions.

[0027] The structured set of technical elements consists of the aforementioned technical object information, technical parameter information, technical constraint information, and document attribute information, and is organized uniformly in the form of a set of records. Each record includes at least: the document identifier; technical object information; the corresponding parameter set; the corresponding constraint set; and document attribute information.

[0028] Step S2: Based on the structured set of technical elements, identify the basic relationships between technical documents. The basic relationships include exception handling relationships, substitution inheritance relationships, and constraint conflict relationships. Establish relationship confidence parameters for the basic relationships. The identification of the underlying relationships includes: matching the same technical object in different technical documents based on the technical object information in the structured technical element set; based on the technical object matching, comparing and analyzing technical parameter information and technical constraint information; when the technical parameter information or technical constraint information in the first technical document deviates from the corresponding information in the second technical document and has applicable conditions, an exception handling relationship is determined between the first and second technical documents; when the technical object in the first technical document corresponds to the technical object in the second technical document, and the technical parameter information in the first technical document substitutes for or inherits the technical parameter information in the second technical document, a substitution and inheritance relationship is determined between the first and second technical documents; when the technical constraint information in different technical documents involving the same technical object is inconsistent, a constraint conflict relationship is determined between the different technical documents.

[0029] The establishment of the relationship confidence parameter includes: for exception handling relationships, substitution inheritance relationships, and constraint conflict relationships, determining the matching degree of the corresponding basic relationships based on the technical parameter information, technical constraint information, and document attribute information in the structured technical element set; the matching degree is determined based on the degree of difference between technical parameter information involving the same technical object, the degree of consistency between technical constraint information, and the time attribute in the document attribute information; according to the matching degree, assigning a corresponding relationship confidence parameter to each basic relationship to characterize the credibility of the corresponding basic relationship.

[0030] In this step, the "basic relationship" refers to a semantically defined association type formed between different technical documents around the same technical object, based on variations in technical parameters, differences in technical constraints, or differences in document attributes. The basic relationships include three categories: Exception handling relationship: refers to a supplementary relationship that deviates from the general technical rules when specific conditions are met. It is used to describe the relationship type of "replacing the original rule under specific working conditions".

[0031] Substitutional inheritance relationship: refers to a relationship type in which subsequent documents inherit some parameters or structures from the technical objects in the previous document during the evolution or upgrade of technical objects, while at the same time substituting some parameters.

[0032] Conflicting constraint relationships: These refer to a type of relationship where constraints defined in different technical documents cannot be satisfied simultaneously for the same technical object in the same applicable scenario.

[0033] The three types of relationships mentioned above correspond semantically to "conditional deviation", "version evolution" and "constraint contradiction", respectively, and are used to construct the basic edge types of the associated network in subsequent steps.

[0034] The "technical object matching" is a prerequisite for basic relationship identification, and its implementation is based on technical object information. The matching process performs a unified comparison according to object identifiers, object names, and object hierarchical relationships. In this embodiment, the cooling pump P-210 is taken as the core object. The object information parsed from different documents is as follows: Object in "P-210 Cooling Pump Design Specification": Cooling pump P-210; Object in "Supplementary Instructions for High Temperature Operation": Cooling pump P-210; Object in "P-210B Upgrade Instructions": Cooling pump P-210B (with a version correspondence to P-210); Object in "Low Vibration Installation Requirements": Cooling pump and its bearing assembly; Object in "High Flow Operation Guidelines": Cooling pump and its drive motor; Through the consistency of object names and the correspondence of models, it is determined that the above documents all involve the same technical object or its derived objects, thus entering the subsequent relationship identification process.

[0035] After the technical object matching is completed, a comparative analysis is performed based on the technical parameter information and technical constraint information to identify whether there is an exception handling relationship. In this embodiment: the "P-210 Cooling Pump Design Specification" stipulates that the sealing component adopts the standard model S-21; the "Supplementary Instructions for High Temperature Operation" stipulates that when the medium temperature is higher than 70℃, the sealing component must adopt the high temperature resistant model S-21H; the above two pieces of information have deviations in parameters and constraints for the same technical object (sealing component), and the deviation has a clear applicable condition (medium temperature > 70℃). Therefore, the following are identified: First technical document: "Supplementary Instructions for High Temperature Operation"; Second technical document: "P-210 Cooling Pump Design Specification"; Relationship type: exception handling relationship; Deviation content: the sealing component model is changed from S-21 to S-21H; Applicable condition: high temperature operation; In this embodiment, the "deviation" specifically refers to: under the same technical object and the same parameter category, the parameter value or constraint content is inconsistent with the original document definition, and this inconsistency is established under specific conditions.

[0036] The substitution inheritance relationship is determined based on the version evolution information of the technical object and the parameter comparison results. In this embodiment: the "P-210 Cooling Pump Design Specification" defines the motor power as 15kW; the "P-210B Upgrade Instructions" states that the motor power is increased to 18.5kW while maintaining the original interface size and structure. Through comparative analysis of the technical parameter information, it is found that: the two documents are consistent in interface size and structural parameters, forming an inheritance relationship; however, a substitution occurs in the motor power parameters, forming a parameter substitution relationship. Therefore, the following are identified: First technical document: "P-210B Upgrade Instructions"; Second technical document: "P-210 Cooling Pump Design Specification"; Relationship type: substitution inheritance relationship; Inherited content: structural dimensions and installation interface; Substituted content: motor power parameters; In this embodiment, "forming substitution or inheritance" specifically refers to: comparing the parameter set of the same technical object, when some parameters are consistent and some parameters change, the consistent part is defined as the inherited content, and the changed part is defined as the substitution content.

[0037] Constraint conflicts are determined through consistency analysis of technical constraint information in different documents. In this embodiment: the "Low Vibration Installation Requirements" stipulates that the vibration value must not exceed 2.0 mm / s; the "High Flow Operation Guidelines" allow higher rotational speeds under high flow conditions, corresponding to a vibration value of 2.6 mm / s; when the user application scenario is "high flow and low vibration requirements," the above two constraints cannot be met simultaneously, thus forming a constraint conflict. The identification results include: First technical document: "Low Vibration Installation Requirements"; Second technical document: "High Flow Operation Guidelines"; Relationship type: Constraint conflict; Conflicting constraint items: Vibration constraint and high flow operation requirements; Conflict condition: High flow and low vibration scenarios coexist; In this embodiment, "inconsistency" specifically refers to: under the same technical object and the same application conditions, the constraints defined in different documents cannot be simultaneously met numerically or logically.

[0038] The "relationship confidence parameter" is used to quantify the credibility of the basic relationship and is an important input data for subsequent propagation calculations. This parameter is determined based on a combination of multiple influencing factors, with each relationship corresponding to a numerical confidence level. In this embodiment, the relationship confidence parameter is determined based on the following three types of data: Parameter difference degree, representing the magnitude of difference in the same parameter between two documents (e.g., a motor power change from 15kW to 18.5kW, the difference degree is obtained through normalization); Constraint consistency degree, representing the consistency of constraints in two documents (e.g., under non-high-temperature conditions, the constraint consistency between the design specification and supplementary instructions is relatively high); and Time attribute influence, representing the time priority relationship between documents (e.g., a document with a newer publication date has a higher weight in relationship judgment). In specific implementation, the above influencing factors are numerically processed respectively. For example, the parameter difference degree is normalized according to the proportion of the parameter difference to the original parameter value; the constraint consistency degree is graded according to whether the constraints are completely consistent, partially consistent, or completely inconsistent; the time attribute is weighted according to the difference in document publication time; and then each influencing factor is weighted according to a preset weight to obtain the relationship confidence parameter. For example, regarding the exception handling relationship between the supplementary instructions for high temperature and the design specifications: parameter difference level: moderately high; constraint consistency level: high (deviation only under specific conditions); time attribute: the supplementary instructions are a later-released document; considering the above factors, the confidence level of this relationship is determined to be 0.86. This relationship confidence level parameter, as attribute data of the basic relationship, is stored together with the relationship type and applicable conditions for subsequent relationship propagation and association validity judgment.

[0039] Step S3: Based on the basic relationships and the corresponding relationship confidence parameters, construct a relationship triggering rule model; The construction of the relationship triggering rule model includes: defining triggering conditions for different types of basic relationships based on the association between exception handling relationships, substitution inheritance relationships, and constraint conflict relationships; when a constraint conflict relationship involving the same technical object exists, determining the triggering conditions for the exception handling relationship based on technical constraint information and document attribute information; when a substitution inheritance relationship exists, determining the state change conditions of the technical constraint information corresponding to the substitution inheritance relationship based on technical parameter information and technical constraint information; when the exception handling relationship meets its applicable conditions, determining the state change conditions of the substitution inheritance relationship based on document attribute information; and establishing an association mapping between different types of basic relationships based on the triggering conditions and state change conditions to construct the relationship triggering rule model.

[0040] In this step, the "relationship triggering rule model" refers to: for basic relationships, according to predefined rules, a structured description of the interaction mechanism between different types of basic relationships, used to guide the state changes and expansion generation of relationships in subsequent steps. Specifically, the relationship triggering rule model describes: the triggering logic between different basic relationships; the activation state of basic relationships under specific conditions; the conditions for deriving new relationships from existing relationships; and the priority and constraint transmission methods between different relationships. The relationship triggering rule model is organized in the form of a "set of rule entries," and each rule entry includes the following data: rule number; input relationship type; triggering condition; state change condition; output relationship type or output relationship state; applicable technical object; applicable condition range; and relationship confidence participation method.

[0041] In this embodiment, the three types of basic relationships are associated in the following ways: The constraint conflict relationship triggers the exception handling relationship. When a constraint conflict exists, the applicable conditions corresponding to the conflicting constraints are analyzed to determine whether an exception handling relationship exists to resolve the conflict. The substitution inheritance relationship transmits the constraint state. When a technical object is substituted or upgraded, the original constraint conditions remain consistent in the inherited part but change in the substituted part, thus causing a change in the constraint state. The exception handling relationship adjusts the scope of application of the substitution inheritance relationship. When the exception handling relationship meets specific conditions, the applicable conditions of the substitution inheritance relationship are modified, thereby changing the effective scope of the substitution relationship. These three types of associations together constitute the basic logic of the relationship triggering rule model.

[0042] When a constraint conflict is identified for the same technical object, it is necessary to determine whether there is an exception handling relationship to resolve the conflict. In this embodiment: the "Low Vibration Installation Requirements" stipulates that the vibration value is ≤2.0 mm / s; the "High Flow Operation Guidelines" allow the vibration value to reach 2.6 mm / s under high flow conditions; the above two constraints constitute a conflict in the "high flow and low vibration requirements" scenario. At this time, analysis is performed by combining document attribute information and technical constraint information: the "Supplementary Instructions for High Temperature Operation" has adjusted the sealing structure, which will affect the vibration characteristics; the document attributes indicate that the supplementary instructions are applicable to specific operating conditions and were published later than the general specifications; through rule definition: when there is a path to modify the constraint item involved in the constraint conflict through specific conditions, based on the applicable conditions in the technical constraint information and the time attribute in the document attribute information, the corresponding exception handling relationship is identified as a conflict mitigation path. Therefore, the following rule entries are formed in this embodiment: Input relationship type: constraint conflict relationship; Triggering condition: the technical object corresponding to the conflict constraint has a supplementary description document with conditions; Output relationship type: exception handling relationship; Output result: under specific working conditions, the constraint in the supplementary description replaces the original constraint; When a substitution inheritance relationship exists, it is necessary to distinguish the constraint states of the inherited and substituted parts of the technical object. In this embodiment: the "P-210B Upgrade Instructions" indicate that the interface size is inherited from P-210; the motor power is replaced by 18.5kW instead of 15kW; combined with the technical constraint information: the original constraint: "When the flow rate is greater than 100m³ / h, the motor power must be ≥15kW"; under the substitution inheritance relationship: for the inherited part (interface size), the constraint remains unchanged; for the substituted part (motor power), the parameter benchmark corresponding to the constraint changes; therefore, the relationship triggering rule model defines: when a substitution inheritance relationship exists, the technical constraint information involving the parameter is updated according to the parameter item that has changed in the technical parameter information; the constraints corresponding to the parameters that have not changed remain in their original state. The corresponding rule entries include: input relationship type: substitution inheritance relationship; state change condition: parameter substitution occurs; output result: update the parameter benchmark value in the corresponding constraint; When an exception handling relationship meets its applicable conditions, it will affect the scope of application of the substitution inheritance relationship. In this embodiment: the sealing component under high-temperature conditions is changed from S-21 to S-21H; P-210B inherits the sealing interface structure of P-210; based on this: if the application scenario meets the high-temperature conditions, the substitution rule of the sealing component should take precedence over the inheritance rule; that is, in P-210B, the sealing component needs to be executed according to the high-temperature supplementary instructions, rather than according to the original design specifications; therefore, the relationship triggering rule model defines: when the applicable conditions of the exception handling relationship are met, the relevant parameters or constraints in the substitution inheritance relationship are modified according to the scope of application and priority in the document attribute information. The corresponding rule entries include: input relationship type: exception handling relationship + substitution inheritance relationship; triggering condition: the applicable conditions of the exception handling relationship are met; output result: adjust the scope of application of relevant parameters or constraints in the substitution inheritance relationship; In this embodiment, the relationship triggering rule model is stored in the form of a rule set, with each rule corresponding to a mechanism of interaction between relationships. The data in the rule set includes: mapping relationships between different relationship types; triggering conditions corresponding to each mapping relationship; specific content of state changes; the scope of application of the rule; and the priority of rule execution.

[0043] In this step, each basic relationship has a corresponding "relationship state," which indicates whether the relationship is effective under specific conditions and how it functions. Relationship states include: valid state: the relationship is valid under the current conditions; conditionally valid state: the relationship is valid only when a specific condition is met; invalid state: the relationship is not valid under the current conditions. The relationship triggering rule model determines and updates the above states, enabling the basic relationships to change states during subsequent propagation.

[0044] Step S4: Combine the time attribute and applicable condition attribute of the basic relationship to construct a multi-dimensional relationship network between technical documents; The construction of the multidimensional association network includes: based on the basic relationship, using technical documents as nodes, and using exception handling relationship, substitution inheritance relationship and constraint conflict relationship as the association relationship between nodes; associating each association relationship with the corresponding relationship confidence parameter, time attribute and applicable condition attribute; and establishing a correspondence between the relationship triggering rule model and the association relationship, so that the association relationship has the ability to change state based on the relationship triggering rule model, forming a multidimensional association network for relationship state propagation calculation.

[0045] In this step, the "multidimensional association network" refers to a data structure that uses technical documents as basic units, connects them through fundamental relationships, and adds multidimensional data such as time attributes, applicable condition attributes, and relationship confidence parameters to each association, thereby forming a data structure that supports subsequent relationship propagation and association analysis. The "multidimensionality" is reflected in: Relationship type dimension: distinguishing between exception handling relationships, substitution inheritance relationships, and constraint conflict relationships; Time dimension: reflecting the temporal sequence and effective order of documents; Condition dimension: reflecting the applicable conditions upon which the relationship depends; Confidence dimension: reflecting the degree of credibility of the relationship's establishment; State dimension: reflecting the validity status of the relationship under current conditions. The combined effect of these various dimensions ensures that the associations between technical documents are no longer merely static connections, but possess conditional constraints and dynamic change capabilities.

[0046] In this embodiment, each technical document is abstracted as a node in the network. Node data includes document attribute information and technical object information. Each node includes at least the following data: node identifier, corresponding document number; document name; document type, such as design specification, supplementary instructions, upgrade instructions; the set of technical objects involved; document effective date and version information; document scope of application; document status, such as currently valid or superseded. Taking this embodiment as an example, the constructed nodes include: Node N1: "P-210 Cooling Pump Design Specification"; Node N2: "Supplementary Instructions for High-Temperature Operation"; Node N3: "P-210B Upgrade Instructions"; Node N4: "Low Vibration Installation Requirements"; Node N5: "High Flow Rate Operation Guidelines". Each node contains its corresponding technical object information and document attribute information for subsequent relationship matching and condition judgment.

[0047] After the nodes are determined, the basic relationships are constructed as associations between nodes. Each association corresponds to an edge data, which includes at least: starting node identifier; ending node identifier; relationship type; involved technical objects; relationship confidence parameter; time attribute; applicable condition attribute; and current relationship status. In this embodiment, the constructed associations include: Node N2 → Node N1: Exception handling relationship; involved objects: sealing components; applicable conditions: medium temperature > 70℃; time attribute: N2's release time is later than N1; Node N3 → Node N1: Substitution inheritance relationship; involved objects: cooling pump unit; inherited content: structural dimensions; substitute content: motor power; Node N4 ↔ Node N5: Constraint conflict relationship; involved objects: cooling pump unit; conflict condition: high flow rate and low vibration requirements; In this way, the basic relationships are transformed into the edge structure in the network.

[0048] The time attribute is used to describe the temporal sequence of documents involved in the association, and it originates from the publication time or effective time in the document attribute information. In this embodiment: the publication time of the "P-210 Cooling Pump Design Specification" is earlier than the "Supplementary Instructions for High Temperature Operation"; the publication time of the "P-210B Upgrade Instructions" is later than the design specification. The time attribute is used to determine: the priority order when multiple rules exist; the order of coverage of the original specification in the exception handling relationship; and the validity of the subsequent document replacing the previous document in the substitution inheritance relationship. The time attribute is stored as an additional attribute of the edges in the network and participates in the relationship state determination in subsequent propagation calculations.

[0049] The applicable condition attributes are derived from technical constraint information and document attribute information, and are used to limit the range of conditions for the establishment of the association relationship. In this embodiment: the applicable condition for the exception handling relationship is "medium temperature > 70℃"; the applicable condition for the constraint conflict relationship is "high flow rate and low vibration requirement"; the applicable condition for the substitution inheritance relationship is "applicable to P-210B model"; the applicable condition attributes are stored in the association relationship in the form of conditional expressions, and are used to determine whether the relationship is in a valid state in subsequent steps.

[0050] In this step, a corresponding relationship is established between the relationship triggering rule model and each association, enabling the association to change its state based on rules. Specifically, each association is bound to a set of rules corresponding to its relationship type; a rule identifier field is added to the association data to indicate which rules control the relationship; when rules are executed, the relationship state is updated based on the current state and attribute data of the association. For example, for the exception handling relationship between nodes N2 and N1, an "exception priority rule" is bound; for the substitution inheritance relationship between nodes N3 and N1, a "parameter substitution rule" is bound; for the constraint conflict relationship between nodes N4 and N5, a "conflict triggering rule" is bound. Through this mapping, each association automatically adjusts its state or triggers a new association when specific conditions are met, according to the rules.

[0051] After the multidimensional association network is constructed, an initial relationship state is assigned to each association. The relationship state is determined as follows: when the applicable conditions are not met, the relationship state is "conditionally valid"; when the applicable conditions are met and there is no conflict, the relationship state is "valid"; when there is a conflict and it is not covered by the exception handling relationship, the relationship state is "invalid". For example, when no high temperature condition is specified, the exception handling relationship N2→N1 is in the conditionally valid state; under high temperature conditions, the relationship becomes valid; in a high flow and low vibration scenario, the constraint conflict relationship between N4 and N5 is in the valid state.

[0052] The multidimensional network constructed through the above steps includes: a set of nodes representing technical documents; a set of relationships representing the basic relationships between documents; a set of relationship attributes including confidence, time attributes, and applicable conditions; a set of rule mappings representing the correspondence between relationships and rules; and a set of states representing the current validity status of relationships.

[0053] Step S5: In response to the retrieval request, determine the target technology object and target application scenario parameters, perform relation state propagation calculation in the multidimensional association network, and combine the relation confidence parameters in the propagation process to obtain candidate association relations and propagation confidence. The execution of the relationship state propagation calculation includes: determining the basic relationships directly related to the technical documents corresponding to the target technical object in a multi-dimensional association network based on the target technical object; determining the state of the directly related basic relationships based on the relationship triggering rule model, and determining the extended association relationships triggered by the directly related basic relationships; for the extended association relationships, continuing to determine the state based on their corresponding relationship triggering rule model, so as to determine the subsequent extended association relationships level by level along the multi-dimensional association network; and determining the candidate association relationships corresponding to the target technical object based on the state change results of each association relationship during the relationship state propagation calculation process.

[0054] The combined calculation of the relationship confidence parameters includes: obtaining the relationship confidence parameters corresponding to the current association during the relationship state propagation calculation process, and the relationship confidence parameters corresponding to the extended association triggered by the current association; based on the numerical correlation between the relationship confidence parameters corresponding to the current association and the relationship confidence parameters corresponding to the extended association, performing combined calculations according to preset combination rules to obtain the propagation confidence corresponding to the extended association; as the relationship state propagation calculation proceeds step by step along the multidimensional association network, updating the propagation confidence corresponding to the subsequent extended association based on the recursive relationship between the propagation confidence corresponding to the previous level association and the relationship confidence parameters corresponding to the current level association.

[0055] In this step, the "retrieval request" is input in a structured format to define the starting point and constraints of the relationship propagation calculation. The retrieval request data comes from user input or preset query conditions. In this embodiment, the retrieval request includes: target technical object: cooling pump P-210; target application scenario parameters: medium temperature: 75℃; operating mode: continuous operation; vibration requirement: low vibration (vibration limit ≤ 2.0mm / s); flow rate requirement: 110m³ / h; the above application scenario parameters are used to determine whether the applicable conditions of the association relationship are met in the subsequent propagation process.

[0056] Based on the target technical object, the basic relationships directly related to it are determined in a multi-dimensional association network. In this embodiment, around the cooling pump P-210, the association relationships directly connected to its corresponding node (N1) are extracted from the network, including N2→N1 (exceptional handling relationship); N3→N1 (substitutional inheritance relationship); N4↔N1, N5↔N1 (constraint relationships associated through object matching); the above relationships constitute the initial set for relationship propagation.

[0057] For directly related basic relationships, the relationship status is determined based on the relationship triggering rule model and the application scenario parameters in the retrieval request. In this embodiment: For the exception handling relationship (N2→N1): Applicable conditions: medium temperature > 70℃; Current scenario parameter: medium temperature is 75℃; the conditions are met, therefore the relationship status is determined to be "valid". For the substitution inheritance relationship (N3→N1): Applicable conditions: applicable to P-210B model; the current retrieval object is P-210, but the substitution inheritance relationship is allowed to be extended to upgraded models; combined with the rule model, the relationship status is determined to be "valid". For the constraint conflict relationship (N4↔N5): Applicable conditions: high flow rate and low vibration; in the current scenario, the flow rate is 110m³ / h and the vibration limit is ≤2.0mm / s; the conflict condition is met, therefore the relationship status is determined to be "valid".

[0058] After determining the status of direct associations, extended associations are generated based on the relationship triggering rule model. In this embodiment: since the exception handling relationship (high-temperature supplementary instructions) is in a valid state, the "exception priority rule" in the triggering rule model extends this relationship to documents related to the alternative inheritance relationship; therefore, an extended association is generated between node N2 (high-temperature supplementary instructions) and node N3 (P-210B upgrade instructions); this extended association indicates that under high-temperature operating conditions, the P-210B model must follow the high-temperature supplementary instructions in the selection of sealing components. The data of the extended association includes: starting node: N2; ending node: N3; relationship source path: N2→N1→N3; relationship type: extended association; triggering rule: linkage rule between exception handling relationship and alternative inheritance relationship; For extended associations, the status is further determined by combining the corresponding relationship triggering rule model, and then propagated level by level along the multi-dimensional association network. In this embodiment: starting from N1, it reaches N2 through the exception handling relationship; the N2 triggers the rule to generate an extended association to N3; from N3, it continues to propagate to other document nodes associated with it (such as subsequent upgrade documents or related operation guidance documents); during each level of propagation, the following operations are performed: determining whether the applicable conditions of the current relationship are met; determining whether the relationship is affected by other relationships (such as conflicting relationships); updating the relationship status according to the rule model; the propagation termination conditions include: no new extended association is generated; the relationship status is determined to be invalid; the preset propagation level limit is reached; During the relation state propagation calculation, all relations with a "valid state" and their corresponding paths are considered as candidate associations. In this embodiment, the candidate associations include: the exception handling relationship between the P-210 design specification and the high-temperature supplementary instructions; the substitution inheritance relationship between P-210 and P-210B; and the extended association relationship between the high-temperature supplementary instructions and P-210B. Each candidate association includes: the association path; the technical object involved; the relationship type; and the applicable conditions. During the propagation process, the relationship confidence parameters are combined and calculated to obtain the propagation confidence. In this embodiment, the combination calculation process includes: obtaining the relationship confidence parameters of the current relationship, for example: N2→N1 (exceptional handling relationship): confidence is 0.86; N3→N1 (substitute inheritance relationship): confidence is 0.82; performing combination calculation on extended association relationships: for the path N2→N1→N3, the corresponding propagation confidence is obtained by recursively calculating the confidence of each relationship on the path; The combined calculation follows these rules: the propagation confidence of the current level is jointly determined by the propagation confidence of the previous level and the current relationship confidence; as the path length increases, the propagation confidence is updated level by level according to the recursive relationship; in this embodiment, after combined calculation, the propagation confidence of the extended association relationship N2→N3 is 0.70; this value reflects the credibility of the association relationship after multi-level propagation.

[0059] During the hierarchical propagation process, the propagation confidence is updated recursively. The following factors are considered during this recursion: the propagation confidence of the previous level relationship; the relationship confidence parameter of the current relationship; the impact of path length on the confidence; and when there are conflicting constraints in the path, the propagation confidence is corrected according to the rule model to reflect the impact of the conflict on the association's credibility. Through the above relationship state propagation calculation and confidence combination calculation, we obtain: a set of candidate associations; the propagation path corresponding to each candidate association; and the propagation confidence of each candidate association.

[0060] Step S6: Obtain structural validity based on candidate association relationships, construct a parameter coupling matrix based on technical parameter information and technical constraint information in the structured technical element set, obtain parameter consistency based on the parameter coupling matrix, and determine the association validity of candidate association relationships; The construction of the parameter coupling matrix includes: extracting parameter variables and constraint relationships between parameters corresponding to each technical object based on technical parameter information and technical constraint information in the structured technical element set; performing parameter correspondence analysis on parameter variables corresponding to different technical objects to determine the coupling relationship between parameters of different technical objects; constructing a parameter coupling matrix based on the constraint relationship between parameters and the coupling relationship between parameters, wherein the matrix elements of the parameter coupling matrix are used to characterize the degree of coupling between parameters of different technical objects; and determining the corresponding matrix elements of the technical objects involved in the candidate association relationship in the parameter coupling matrix based on the candidate association relationship.

[0061] The determination of the association validity includes: analyzing the propagation path of candidate associations in a multidimensional association network, and determining structural validity based on the changes in connectivity or path characteristics of the propagation paths of candidate associations before and after removing intermediate associations; obtaining the corresponding matrix elements of the technical objects involved in the candidate associations in the parameter coupling matrix, and calculating the degree of parameter constraint consistency between technical objects based on the degree of parameter coupling represented by the matrix elements to characterize parameter consistency; adjusting the parameter consistency based on the propagation confidence corresponding to the candidate associations to obtain corrected parameter consistency; and determining the candidate association as a valid association when the structural validity meets the preset conditions and the corrected parameter consistency is greater than the preset threshold.

[0062] In this step, "structural validity" refers to whether the propagation path of the candidate association in the multidimensional association network has stable structural support, that is, whether the association depends on the key relationship chain and whether there is a significant structural change after the key relationship is removed. Structural validity is determined based on the propagation path of the candidate association, which originates from the relationship state propagation calculation results. In this embodiment, taking the extended association between the candidate association "High Temperature Condition Supplementary Description (N2) and P-210B Upgrade Description (N3)" as an example, its propagation path is: N2 (High Temperature Supplementary Description) → N1 (Design Specification) → N3 (Upgrade Description); structural analysis of this path shows that the original path is a three-node, two-relationship path; after removing the intermediate association (such as the replacement inheritance relationship N3 → N1), the path is broken; the path connectivity changes from a connected state to a disconnected state; therefore, it is determined that the candidate association depends on the key intermediate relationship, has clear structural support, and its structural validity meets the preset conditions.

[0063] In this embodiment, the "path feature change" specifically includes: whether the path is connected; whether the path length has changed; whether the number of intermediate nodes has decreased; if the path is no longer connected or the path structure changes significantly after removing any intermediate relationship, then the structure validity is determined to be valid.

[0064] In this step, the "parameter coupling matrix" refers to a matrix structure constructed based on the mutual influence relationships and corresponding constraints between parameter variables of different technical objects, used to characterize the degree of correlation between parameters. The parameter coupling matrix reflects: the linkage relationship between parameters of different technical objects; the degree of influence of parameter changes on other parameter constraints; and the degree of coordination of parameter combinations under specific conditions. This matrix originates from technical parameter information and technical constraint information.

[0065] In this embodiment, the following parameter variables are extracted from the structured technical element set: motor power; speed; flow rate; head; vibration value; bearing temperature rise; sealing gap; and medium temperature. Simultaneously, the constraint relationships between these parameters are extracted. For example, when the flow rate is greater than 100 m³ / h, the motor power must be greater than or equal to 15 kW; as the speed increases, the vibration value rises; when the medium temperature rises, the sealing material needs adjustment; there is a positive correlation between speed and bearing temperature rise. These constraint relationships are used to determine the dependencies between the parameters.

[0066] In this embodiment, a correspondence analysis is performed on the parameters between different technical objects. For example, motor power and speed correspond to the same drive system; speed and vibration value correspond to operating performance parameters; medium temperature and sealing component parameters correspond to environmental adaptation parameters. Based on the above correspondence, the coupling relationship between parameters is determined, including: strong coupling relationship: parameter changes directly affect the constraint of another parameter, such as speed and vibration value; medium coupling relationship: parameter changes indirectly affect another parameter, such as motor power and flow rate; weak coupling relationship: the influence between parameters is small, such as sealing gap and flow rate. The determination of the coupling relationship is based on: the dependency description in the technical constraint information; and the range of influence of parameter changes on the constraint conditions. In this embodiment, a matrix is ​​constructed based on a set of parameter variables, where: rows represent the technical object or parameter category to which the parameter variable belongs; columns represent other related parameter variables; matrix elements represent the degree of coupling between two parameters; for example: the coupling degree between motor power and speed is high; the coupling degree between speed and vibration value is high; the coupling degree between medium temperature and sealing component parameters is high; and the coupling degree between motor power and sealing gap is low; the matrix elements use numerical methods to represent the degree of coupling, for example: strong coupling corresponds to higher values; medium coupling corresponds to intermediate values; and weak coupling corresponds to lower values; this matrix is ​​used for subsequent calculation of parameter consistency between objects involved in candidate association relationships.

[0067] For each candidate association, the technical objects and related parameter variables involved are extracted, and the corresponding matrix elements are determined in the parameter coupling matrix. In this embodiment, for the candidate association "extended association between high-temperature supplementary instructions and P-210B upgrade instructions": the parameters involved include: medium temperature, motor power, speed, vibration value, and sealing component parameters; the corresponding elements are extracted from the matrix: the coupling degree between medium temperature and sealing component parameters; the coupling degree between speed and vibration value; and the coupling degree between motor power and flow rate; the above matrix elements are used to calculate the degree of coordination between the parameters.

[0068] The term "parameter consistency" refers to whether the parameters involved in the candidate correlation simultaneously meet the corresponding technical constraints under their coupling relationships and constraints. In this embodiment: under high temperature conditions, the sealing assembly is adjusted to S-21H; under high flow conditions, the motor power is increased to 18.5kW; the speed and vibration value must meet the vibration limit under low vibration requirements; by analyzing the coupling relationships and constraints between the above parameters: the sealing assembly adjustment meets the high temperature constraint; the motor power meets the flow constraint; the vibration value is within the constraint range; therefore, it is determined that the parameter consistency of this candidate correlation meets the requirements.

[0069] In this step, parameter consistency is adjusted based on the propagation confidence corresponding to the candidate association. Specifically, when the propagation confidence is high, parameter consistency maintains the original calculation result; when the propagation confidence decreases, parameter consistency is reduced accordingly. For example, if the propagation confidence is 0.70, the parameter consistency is adjusted proportionally. The corrected parameter consistency is used to more accurately reflect the reliability of the candidate association.

[0070] The candidate associations are ultimately determined by considering both structural validity and the consistency of the corrected parameters. In this embodiment: structural validity meets preset conditions; the consistency of the corrected parameters is greater than a preset threshold; therefore, the candidate association is determined to be a valid association. If either condition is not met, the candidate association will not be considered a valid association for subsequent screening. After this step, the following results are obtained: the structural validity determination result for each candidate association; the parameter consistency and correction results for each candidate association; and the set of valid associations.

[0071] Step S7: Based on the target application scenario parameters, propagation confidence, and association validity, filter valid associations, generate and output search results.

[0072] In this step, the input data for screening specifically includes: a set of candidate associations; the propagation confidence of each candidate association; the structural validity determination result of each candidate association; the consistency of the correction parameters of each candidate association; and the target application scenario parameters in the retrieval request. In this embodiment, the above data are all constructed based on the association analysis results of the cooling pump P-210 in the scenario of "high temperature, continuous operation, low vibration, and high flow rate".

[0073] The "screening of valid associations" refers to: in the candidate association set, according to preset screening rules, eliminating associations that do not meet technical constraints or application scenario requirements, and retaining only those that meet comprehensive conditions. The screening rules are based on the following three types of data: application scenario matching rules, used to determine whether the applicable conditions of candidate associations are consistent with the parameters of the target application scenario. In this embodiment, for example, for an exception handling relationship (high temperature supplementary explanation), its applicable condition is "medium temperature > 70℃," and the current scenario temperature is 75℃, which meets the condition; for documents only applicable to normal temperature conditions, their applicable conditions do not meet the current scenario and must be eliminated; propagation confidence screening rules, used to screen associations with high propagation path reliability. In this embodiment, the preset propagation confidence threshold is 0.60: when the propagation confidence of a candidate association is greater than or equal to this threshold, the association is retained; when the propagation confidence is lower than this threshold, it is determined that the association has significant uncertainty in the propagation process and is eliminated from the candidate set; association validity screening rules, used to screen associations that simultaneously meet the requirements of structural validity and parameter consistency. In this embodiment: the association is retained only when the structural validity meets the preset conditions and the consistency of the correction parameters is greater than the preset threshold; if the candidate association shows parameter conflict or constraint failure in the parameter coupling matrix analysis, it is not included in the set of valid associations. In this embodiment, the candidate association set is filtered in the following order: First, based on the application scenario parameters, associations that do not meet the applicable conditions are eliminated; among the remaining associations, they are filtered according to the propagation confidence; for the associations that pass the first two steps of filtering, they are further judged based on structural validity and consistency of correction parameters; after the above step-by-step filtering, the final set of valid associations is obtained.

[0074] In this embodiment, the effective associations after screening include: the exception handling relationship between the P-210 design specification and the supplementary instructions for high-temperature operating conditions; meeting the high-temperature application scenario; high propagation confidence; parameter consistency meeting the constraint requirements; the substitution inheritance relationship between P-210 and P-210B; in high-flow scenarios, the upgraded model meets the power requirements; motor power and flow rate are matched in the parameter coupling relationship; the extended association relationship between the supplementary instructions for high-temperature conditions and P-210B; generated by rule triggering; effective under high-temperature conditions; and both propagation confidence and parameter consistency meeting the screening conditions. The above associations constitute the final set of effective associations.

[0075] The term "generating search results" refers to organizing the filtered valid associations according to a preset format to form search result data for use. Each search result includes at least: target technical object; associated technical document identifier and name; association type; propagation path; propagation confidence level; association validity judgment result; applicable scenario description; and association basis description. In this embodiment, an example of a search result is as follows: Target object: Cooling pump P-210; Associated document: "P-210B Upgrade Instructions"; Association type: Extended association; Propagation path: High temperature supplementary instructions → Design specifications → Upgrade instructions; Propagation confidence level: 0.70; Association validity: Meets structural validity and parameter consistency requirements; Applicable scenario: Medium temperature 75℃, high flow rate, low vibration; Basis description: Based on the combined propagation result of high temperature exception handling relationship and substitution inheritance relationship. After generating the search results, the results are sorted to prioritize the display of highly relevant relationships. The sorting criteria include: propagation confidence level; consistency of correction parameters; relationship path length; and document time attribute (preferring the latest documents). In this embodiment, relationships with higher propagation confidence and shorter paths are prioritized.

[0076] The term "output search results" refers to providing the sorted search results in a structured data format for subsequent display or further processing. The output data includes: a list of valid relationships; detailed attribute information for each relationship; corresponding applicable conditions and constraints; and the output results maintain consistency with the aforementioned data structure so that they can be directly used for technical document recommendations or decision support.

[0077] In this embodiment, the preset conditions and preset thresholds are determined based on the historical data statistics and technical constraint information in the structured technical element set, in order to ensure the stability and consistency of the relationship propagation, association validity determination and result screening process.

[0078] The "preset conditions" refer to the set of conditions used to determine whether a correlation is established or whether to proceed to the next processing step. These conditions originate from the applicable scope descriptions in the technical constraint information and document attribute information. The process of obtaining preset conditions includes: extracting the set of technical constraint conditions from the technical constraint information in the structured technical element set, such as: medium temperature greater than 70℃; flow rate greater than 100 m³ / h; vibration value not exceeding 2.0 mm / s; and converting the above conditions into standardized condition expressions, including: condition variables (such as temperature, flow rate, vibration); comparison relationships (greater than, less than, range); and threshold values. Extract the document's applicable scope conditions from its attribute information, such as: applicable to high-temperature conditions; applicable to continuous operation scenarios; applicable to low-vibration requirement scenarios. Map the above descriptions to conditional expressions consistent with technical constraints, such as: high-temperature conditions → medium temperature > 70℃; low-vibration requirement → vibration ≤ 2.0 mm / s. Construct a unified set of conditions by merging the technical constraints and the document's applicable scope conditions to create a unified set of preset conditions used for: determining the applicable conditions of exception handling relationships; determining whether the relationship status is valid; and path selection during the propagation of constraint relationships. In this embodiment, the preset conditions include: medium temperature > 70℃; flow rate > 100 m³ / h; vibration ≤ 2.0 mm / s. These conditions are used for relationship status determination in S5 and for filtering processing in S7.

[0079] The propagation confidence threshold is used to filter candidate associations with high reliability of propagation paths. The propagation confidence threshold is determined based on the statistical distribution of historical relationship propagation results. Its acquisition process includes: obtaining historical propagation confidence data; obtaining propagation confidence data for multiple candidate associations from historical retrieval processes, such as a set of confidence values ​​corresponding to multiple propagation paths; dividing the propagation confidence into intervals, for example: high confidence interval: above 0.75; medium confidence interval: 0.60~0.75; low confidence interval: below 0.60; determining the screening threshold; based on actual application requirements, the lower limit of the medium confidence interval is used as the screening threshold. In this embodiment, a propagation confidence threshold of 0.60 is selected.

[0080] The pre-defined conditions for structural validity are used to determine whether the propagation path of candidate associations has stable structural support. The acquisition methods include: statistically analyzing historical propagation path characteristics; analyzing the propagation paths of historically valid associations to extract the following features: path length; number of intermediate relationships; changes in connectivity after removing intermediate relationships; determining key structure judgment rules; based on the statistical results, determining the following judgment conditions: if the path is no longer connected after removing any intermediate relationship, then the path has structural support; if the path length is greater than 1 and contains at least one intermediate node, then it has propagation characteristics; forming the pre-defined conditions for structural validity. In this embodiment, structural validity satisfies the following conditions: the path contains at least one intermediate relationship; the path is disconnected after removing a key relationship. The parameter consistency threshold is used to determine whether candidate associations meet technical constraints at the parameter level. Its acquisition process includes: calculating the parameter consistency of historical associations; calculating the parameter consistency of historical valid associations using the parameter coupling matrix to obtain a set of consistency values; statistically analyzing the consistency distribution range; obtaining: a high consistency interval; a medium consistency interval; and a low consistency interval; determining the consistency threshold; and selecting the lower limit of the medium consistency interval as the parameter consistency threshold. In this embodiment, the parameter consistency threshold is set to 0.65; when the corrected parameter consistency of a candidate association is greater than this threshold, it is determined that it meets the parameter constraint consistency requirements.

[0081] In this embodiment: the preset conditions are derived from technical constraint information and the scope of application of the document; the propagation confidence threshold is derived from the statistics of historical propagation results; the structural validity judgment condition is derived from the propagation path structure analysis; the parameter consistency threshold is derived from the statistics of parameter coupling matrix calculation results; the above preset conditions and preset thresholds are used consistently throughout the method to ensure the uniformity of the relationship identification, propagation calculation and result screening process.

[0082] Example 2: A technical document association retrieval system, see [link to example]. Figure 1 As shown, it includes the following modules: Document structuring module: Retrieves multiple technical documents, parses them, and obtains a set of structured technical elements; Relationship identification module: Based on a structured set of technical elements, it identifies the fundamental relationships between technical documents; Triggering rule module: Constructs a relationship triggering rule model based on the basic relationship and the corresponding relationship confidence parameters; The network module combines the time and applicable condition attributes of the basic relationships to construct a multi-dimensional network of relationships between technical documents. Propagation confidence module: In response to the retrieval request, it determines the target technology object and target application scenario parameters, performs relation state propagation calculation in the multidimensional association network, and combines the relation confidence parameters in the propagation process to obtain candidate associations and propagation confidence. The association verification module obtains structural validity based on candidate association relationships, constructs a parameter coupling matrix based on technical parameter information and technical constraint information in the structured technical element set, obtains parameter consistency based on the parameter coupling matrix, and determines the association validity of candidate association relationships. Results output module: Based on the target application scenario parameters, propagation confidence, and association validity, it filters valid associations and generates and outputs search results.

[0083] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for associating and retrieving technical documents, characterized in that, Applied in a technical document management system, executed by a processor and including: Multiple technical documents are acquired and parsed to obtain a structured set of technical elements, which includes technical object information, technical parameter information, technical constraint information, and document attribute information. Based on a set of structured technical elements, the basic relationships between technical documents are identified. These basic relationships include exception handling relationships, substitution inheritance relationships, and constraint conflict relationships. Relationship confidence parameters are then established for these basic relationships. Based on the basic relationships and their corresponding relationship confidence parameters, a relationship triggering rule model is constructed; and by combining the time attributes and applicable condition attributes of the basic relationships, a multi-dimensional relationship network among technical documents is constructed. In response to a retrieval request, the target technology object and target application scenario parameters are determined, relation state propagation calculation is performed in a multidimensional association network, and the relation confidence parameters in the propagation process are combined and calculated to obtain candidate associations and propagation confidence. The structural validity is obtained based on candidate association relationships. A parameter coupling matrix is ​​constructed based on the technical parameter information and technical constraint information in the structured technical element set. The parameter consistency is obtained based on the parameter coupling matrix, and the association validity of the candidate association relationships is determined. Based on the target application scenario parameters, propagation confidence, and association validity, valid associations are filtered, and search results are generated and output.

2. The technical document association retrieval method according to claim 1, characterized in that, The identification of the underlying relationships includes: matching the same technical object in different technical documents based on the technical object information in the structured technical element set; based on the technical object matching, comparing and analyzing technical parameter information and technical constraint information; when the technical parameter information or technical constraint information in the first technical document deviates from the corresponding information in the second technical document and has applicable conditions, an exception handling relationship is determined between the first and second technical documents; when the technical object in the first technical document corresponds to the technical object in the second technical document, and the technical parameter information in the first technical document substitutes for or inherits the technical parameter information in the second technical document, a substitution and inheritance relationship is determined between the first and second technical documents; when the technical constraint information in different technical documents involving the same technical object is inconsistent, a constraint conflict relationship is determined between the different technical documents.

3. The technical document association retrieval method according to claim 1, characterized in that, The establishment of the relationship confidence parameter includes: for exception handling relationships, substitution inheritance relationships, and constraint conflict relationships, determining the matching degree of the corresponding basic relationships based on the technical parameter information, technical constraint information, and document attribute information in the structured technical element set; the matching degree is determined based on the degree of difference between technical parameter information involving the same technical object, the degree of consistency between technical constraint information, and the time attribute in the document attribute information; according to the matching degree, assigning a corresponding relationship confidence parameter to each basic relationship to characterize the credibility of the corresponding basic relationship.

4. The technical document association retrieval method according to claim 1, characterized in that, The construction of the relationship triggering rule model includes: defining triggering conditions for different types of basic relationships based on the association between exception handling relationships, substitution inheritance relationships, and constraint conflict relationships; when a constraint conflict relationship involving the same technical object exists, determining the triggering conditions for the exception handling relationship based on technical constraint information and document attribute information; when a substitution inheritance relationship exists, determining the state change conditions of the technical constraint information corresponding to the substitution inheritance relationship based on technical parameter information and technical constraint information; when the exception handling relationship meets its applicable conditions, determining the state change conditions of the substitution inheritance relationship based on document attribute information; and establishing an association mapping between different types of basic relationships based on the triggering conditions and state change conditions to construct the relationship triggering rule model.

5. The technical document association retrieval method according to claim 1, characterized in that, The construction of the multidimensional association network includes: based on the basic relationship, using technical documents as nodes, and using exception handling relationship, substitution inheritance relationship and constraint conflict relationship as the association relationship between nodes; associating each association relationship with the corresponding relationship confidence parameter, time attribute and applicable condition attribute; and establishing a correspondence between the relationship triggering rule model and the association relationship, so that the association relationship has the ability to change state based on the relationship triggering rule model, forming a multidimensional association network for relationship state propagation calculation.

6. The technical document association retrieval method according to claim 1, characterized in that, The execution of the relationship state propagation calculation includes: determining the basic relationships directly related to the technical documents corresponding to the target technical object in a multi-dimensional association network based on the target technical object; determining the state of the directly related basic relationships based on the relationship triggering rule model, and determining the extended association relationships triggered by the directly related basic relationships; for the extended association relationships, continuing to determine the state based on their corresponding relationship triggering rule model, so as to determine the subsequent extended association relationships level by level along the multi-dimensional association network; and determining the candidate association relationships corresponding to the target technical object based on the state change results of each association relationship during the relationship state propagation calculation process.

7. The technical document association retrieval method according to claim 1, characterized in that, The combined calculation of the relationship confidence parameters includes: obtaining the relationship confidence parameters corresponding to the current association during the relationship state propagation calculation process, and the relationship confidence parameters corresponding to the extended association triggered by the current association; based on the numerical correlation between the relationship confidence parameters corresponding to the current association and the relationship confidence parameters corresponding to the extended association, performing combined calculations according to preset combination rules to obtain the propagation confidence corresponding to the extended association; as the relationship state propagation calculation proceeds step by step along the multidimensional association network, updating the propagation confidence corresponding to the subsequent extended association based on the recursive relationship between the propagation confidence corresponding to the previous level association and the relationship confidence parameters corresponding to the current level association.

8. A method for associating and retrieving technical documents according to claim 1, characterized in that, The construction of the parameter coupling matrix includes: extracting parameter variables and constraint relationships between parameters corresponding to each technical object based on technical parameter information and technical constraint information in the structured technical element set; performing parameter correspondence analysis on parameter variables corresponding to different technical objects to determine the coupling relationship between parameters of different technical objects; constructing a parameter coupling matrix based on the constraint relationship between parameters and the coupling relationship between parameters, wherein the matrix elements of the parameter coupling matrix are used to characterize the degree of coupling between parameters of different technical objects; and determining the corresponding matrix elements of the technical objects involved in the candidate association relationship in the parameter coupling matrix based on the candidate association relationship.

9. The technical document association retrieval method according to claim 1, characterized in that, The determination of the association validity includes: analyzing the propagation path of candidate associations in a multidimensional association network, and determining structural validity based on the changes in connectivity or path characteristics of the propagation paths of candidate associations before and after removing intermediate associations; obtaining the corresponding matrix elements of the technical objects involved in the candidate associations in the parameter coupling matrix, and calculating the degree of parameter constraint consistency between technical objects based on the degree of parameter coupling represented by the matrix elements to characterize parameter consistency; adjusting the parameter consistency based on the propagation confidence corresponding to the candidate associations to obtain corrected parameter consistency; and determining the candidate association as a valid association when the structural validity meets the preset conditions and the corrected parameter consistency is greater than the preset threshold.

10. A technical document association retrieval system, characterized in that, The system employs a technical document association retrieval method according to any one of claims 1 to 9, comprising: Document structuring module: Retrieves multiple technical documents, parses them, and obtains a set of structured technical elements; Relationship identification module: Based on a structured set of technical elements, it identifies the fundamental relationships between technical documents; Triggering rule module: Constructs a relationship triggering rule model based on the basic relationship and the corresponding relationship confidence parameters; The network module combines the time and applicable condition attributes of the basic relationships to construct a multi-dimensional network of relationships between technical documents. Propagation confidence module: In response to the retrieval request, it determines the target technology object and target application scenario parameters, performs relation state propagation calculation in the multidimensional association network, and combines the relation confidence parameters in the propagation process to obtain candidate associations and propagation confidence. The association verification module obtains structural validity based on candidate association relationships, constructs a parameter coupling matrix based on technical parameter information and technical constraint information in the structured technical element set, obtains parameter consistency based on the parameter coupling matrix, and determines the association validity of candidate association relationships. Results output module: Based on the target application scenario parameters, propagation confidence, and association validity, it filters valid associations and generates and outputs search results.

Images