Method for generating complex query of special equipment inspection based on semantic constraint chain

Abstract

The disclosure provides a special equipment inspection complex query generation method based on a semantic constraint chain, comprising: analyzing a natural query sentence input by a user to identify inspection equipment information, and constructing a multi-process semantic graph based on the inspection equipment information. The inspection stage jump relationship of the same inspection equipment at different process nodes is extracted, and a cross-process semantic bridging relationship is established for the process nodes, so as to embed the cross-process semantic bridging relationship in the multi-process semantic graph to obtain an extended semantic graph structure. The semantic nodes are searched and reorganized through the extended semantic graph structure to expand the semantic constraint chain of a single process to obtain a cross-process semantic constraint chain. The cross-process semantic constraint chain is mapped to a multi-table association path in a database, and a table connection relationship is established through the cross-process semantic bridging relationship to generate a target query sentence covering multiple inspection processes, thereby ensuring the continuity and integrity of the complex query path in a multi-process environment.

Description

Technical Field

[0001] This invention relates to the field of equipment inspection technology, and in particular to a method for generating complex queries for special equipment inspection based on semantic constraint chains. Background Technology

[0002] In scenarios where periodic inspections and defect re-inspections of special equipment are carried out simultaneously, existing technologies show that the same equipment has different business paths at different inspection stages. When the semantic constraint chain is constructed within a single process, it is difficult to cover the node jump relationship across processes, which leads to the interruption of the query path at the process boundary and makes it difficult to form a complete retrieval loop. In other words, there is a problem of semantic constraint chain breakage in nested scenarios across inspection processes.

[0003] Therefore, existing technologies have technical problems such as the semantic constraint chain breaking at the process boundary in nested cross-inspection process scenarios, making it difficult to form continuous query paths across processes. Consequently, it is difficult to achieve semantic node jumpable connection between different inspection stages, and even more difficult to guarantee the continuity and integrity of complex query paths in multi-process environments. Summary of the Invention

[0004] Therefore, it is necessary to propose a method for generating complex queries for special equipment inspection based on semantic constraint chains to address the aforementioned technical problems.

[0005] The present invention adopts the following technical solution.

[0006] The first aspect of this invention discloses a method for generating complex queries for special equipment inspection based on semantic constraint chains, the method comprising: The user-input natural query statement is parsed to identify inspection equipment information, and a multi-process semantic graph is constructed based on the inspection equipment information; Extract the inspection stage jump relationship of the same inspection equipment at different process nodes, and establish cross-process semantic bridging relationship for the process nodes, so as to embed the cross-process semantic bridging relationship into the multi-process semantic graph to obtain an extended semantic graph structure. The extended semantic graph structure is used to perform path search and path reorganization on semantic nodes to extend the semantic constraint chain of a single process and obtain a cross-process semantic constraint chain. The cross-process semantic constraint chain is mapped to a multi-table association path in the database, and the inter-table connection relationship is established through the cross-process semantic bridging relationship to generate a target query statement covering multiple inspection processes.

[0007] Furthermore, the parsing of the user-input natural query statement to identify inspection equipment information, and the construction of a multi-process semantic graph based on the inspection equipment information, includes: The system obtains the user's input natural query statement and performs word segmentation, part-of-speech tagging, and business dictionary matching on the natural query statement to parse the natural query statement into multiple business semantic units to be processed. The inspection type and inspection stage are extracted from the inspection business semantic unit, and the inspection type and inspection stage are assigned to distinguish multiple different inspection processes. Extract the device object from the inspection business semantic unit, and use the device object as a binding object to control the inspection business semantic units of the same device under different inspection processes to bind to the same device object, so as to output multiple process belonging semantic unit groups.

[0008] Furthermore, the step of parsing the natural query statement input by the user to identify the inspection equipment information, and constructing a multi-process semantic graph based on the inspection equipment information, also includes: Using the inspection business semantic unit as the semantic node and the device object as the root node, a corresponding process semantic sub-graph is established for each process belonging to the semantic unit group according to the order of the inspection stages. Using the device object as the main index, semantic nodes in different process semantic subgraphs are compared to establish cross-process identifiers for semantic nodes belonging to the same device object but located in different inspection processes. Based on the process semantic subgraph and the cross-process identifiers and semantic nodes of the same device object, the multi-process semantic graph is constructed; The inspection equipment information includes the inspection type, inspection stage, and equipment object.

[0009] Furthermore, the step of extracting the inspection stage jump relationship of the same inspection equipment at different process nodes and establishing cross-process semantic bridging relationships for the process nodes, so as to embed the cross-process semantic bridging relationships into the multi-process semantic graph, to obtain an extended semantic graph structure, includes: The semantic node set corresponding to the same inspection device is located and extracted from the multi-process semantic graph, and semantic nodes with process jumps are selected from the semantic node set. The semantic nodes with process jumps are identified as cross-process candidate connection nodes, and the inspection stage sequence of the cross-process candidate connection nodes is analyzed to determine whether the cross-process candidate connection nodes have entered the subsequent inspection stage from the previous inspection stage.

[0010] Furthermore, the step of extracting the inspection stage jump relationships corresponding to the same inspection equipment at different process nodes, and establishing cross-process semantic bridging relationships for the process nodes, so as to embed the cross-process semantic bridging relationships into the multi-process semantic graph to obtain an extended semantic graph structure, also includes: If the cross-process candidate connection node is determined to enter the subsequent inspection stage from the previous inspection stage, the inspection stage jump relationship is established, and it is determined whether the inspection stage jump relationship meets the preset bridging conditions. If so, the cross-process semantic bridging relationship is established. The cross-process semantic bridging relationship is embedded into the multi-process semantic graph in the form of nodes, and the order of the inspection stages and the connection relationship of the semantic nodes in the multi-process semantic graph are preserved during the embedding process to obtain the extended semantic graph structure.

[0011] Furthermore, the step of performing path search and path reorganization on semantic nodes through the extended semantic graph structure to extend the semantic constraint chain of a single process and obtain a cross-process semantic constraint chain includes: Query the query target corresponding to the natural query statement, and take the initial verification stage in the natural query statement as the starting semantic node and the query target as the target semantic node; Based on the starting semantic node and the target semantic node, the cross-process semantic bridging relationship traversed by the query path is determined through the extended semantic graph structure to obtain the cross-process path search boundary; Within the cross-process path search boundary, multiple arrival nodes corresponding to the starting semantic node are searched according to the inspection process order and cross-process semantic bridging relationship, and a candidate cross-process semantic path set is constructed based on the starting semantic node and multiple arrival nodes.

[0012] Furthermore, the step of performing path search and path reorganization on semantic nodes through the extended semantic graph structure to extend the semantic constraint chain of a single process and obtain a cross-process semantic constraint chain also includes: Extract the constraints corresponding to each candidate cross-process semantic path in the candidate cross-process semantic path set, and use the constraints in the preceding inspection process as inherited constraints. When the preceding inspection process enters the candidate cross-process semantic path, the inherited constraints are passed to the constraints corresponding to the candidate cross-process semantic path to obtain the cross-process constraint integrated path set. The cross-process constraint integration path is reorganized by focusing on semantic nodes located in different inspection processes. During the reorganization, the order of inspection processes and the cross-process semantic bridging relationship in the candidate cross-process semantic path are preserved to generate the cross-process semantic constraint chain.

[0013] Furthermore, the step of mapping the cross-process semantic constraint chain to a multi-table join path in the database, and establishing inter-table join relationships through the cross-process semantic bridging relationship to generate a target query statement covering multiple verification processes, includes: Multiple semantic nodes of different types are extracted from the cross-process semantic constraint chain, and the query path is used as the link attribution benchmark. Each type of semantic node is mapped to the corresponding data table to obtain the cross-process data table mapping set. Identify the data tables and fields corresponding to the cross-process semantic bridging relationship from the cross-process data table mapping set, and construct the multi-table association path using the cross-process semantic bridging relationship as the inter-table association; The inheritance constraints are transformed into query conditions, and the fields corresponding to the multi-table association paths, the cross-process semantic bridging relationships, and the preset filtering conditions are integrated into the target query statement based on the query conditions.

[0014] A second aspect of this invention discloses a complex query generation device for special equipment inspection based on semantic constraint chains, used to implement the complex query generation method for special equipment inspection based on semantic constraint chains as described in any one of the first aspects, the device comprising: The semantic graph construction module is used to parse the natural query statements input by the user to identify the inspection equipment information, and to construct a multi-process semantic graph based on the inspection equipment information. The bridging relationship establishment module is used to extract the inspection stage jump relationship of the same inspection equipment at different process nodes, and establish cross-process semantic bridging relationships for the process nodes, so as to embed the cross-process semantic bridging relationships into the multi-process semantic graph to obtain an extended semantic graph structure. The constraint chain reorganization module is used to perform path search and path reorganization on semantic nodes through the extended semantic graph structure, so as to extend the semantic constraint chain of a single process and obtain a cross-process semantic constraint chain. The query statement generation module is used to map the cross-process semantic constraint chain to a multi-table association path in the database, and establish inter-table connection relationships through the cross-process semantic bridging relationship to generate target query statements covering multiple inspection processes.

[0015] A third aspect of the present invention discloses a terminal, including a processor and a storage medium; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the method described in the first aspect.

[0016] A fourth aspect of the present invention discloses a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in the first aspect.

[0017] The present invention has the following advantages: (1) This invention parses the natural query statements input by users to identify inspection equipment information and constructs a multi-process semantic graph based on the inspection equipment information. Then, based on the constructed multi-process semantic graph, it extracts the inspection stage jump relationship corresponding to the same inspection equipment at different process nodes, and establishes cross-process semantic bridging relationships for process nodes to embed the cross-process semantic bridging relationships into the multi-process semantic graph, thus obtaining an extended semantic graph structure. To address the technical problem of semantic constraint chains breaking at process boundaries in cross-inspection process nesting scenarios, a cross-process continuous semantic graph structure is established, enabling semantic nodes between different inspection stages to jump and connect.

[0018] (2) This invention extends the semantic constraint chain of a single process by performing path search and path reorganization on semantic nodes through an extended semantic graph structure, thereby obtaining a cross-process semantic constraint chain. Finally, the cross-process semantic constraint chain is mapped to a multi-table association path in the database, and inter-table connections are established through cross-process semantic bridging relationships to generate target query statements covering multiple inspection processes. By constructing a semantic constraint chain with cross-process semantic connection capabilities, not only is it possible to achieve jumpable connections between semantic nodes in different inspection stages, but it also ensures the continuity and integrity of complex query paths in a multi-process environment. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this 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 described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating the method for generating complex queries for special equipment inspection based on semantic constraint chains provided by the present invention.

[0021] Figure 2 This is a schematic diagram of the structure of the special equipment inspection complex query generation device based on semantic constraint chain provided by the present invention. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] like Figure 1As shown, in one embodiment, a method for generating complex queries for special equipment inspection based on semantic constraint chains includes the following steps: Step S110: Parse the natural query statement entered by the user to identify the inspection equipment information, and construct a multi-process semantic graph based on the inspection equipment information.

[0024] In some embodiments, the special equipment inspection complex query generation method based on semantic constraint chain provided by the present invention includes the following steps in step S110: Step S111: Obtain the natural query statement input by the user, and perform word segmentation, part-of-speech tagging and business dictionary matching on the natural query statement to parse the natural query statement into multiple business semantic units to be processed.

[0025] Step S112: Extract the inspection type and inspection stage from the inspection business semantic unit, and determine the attribution of the inspection type and inspection stage to distinguish multiple different inspection processes.

[0026] Step S113: Extract the device object from the inspection business semantic unit and use the device object as the binding object to control the inspection business semantic units of the same device under different inspection processes to bind to the same device object, so as to output multiple process belonging semantic unit groups.

[0027] In some embodiments, the special equipment inspection complex query generation method based on semantic constraint chain provided by the present invention further includes the following steps in step S110: Step S114: Using the inspection business semantic unit as the semantic node and the device object as the root node, establish a corresponding process semantic sub-graph for each process belonging to the semantic unit group according to the order of the inspection stages.

[0028] Step S115: Using the device object as the main index, compare the semantic nodes in the semantic subgraphs of different processes to establish cross-process identifiers for semantic nodes belonging to the same device object but located in different inspection processes.

[0029] Step S116: Construct a multi-process semantic graph based on the process semantic subgraph and cross-process identifiers and semantic nodes of the same device object.

[0030] The inspection equipment information includes the inspection type, inspection stage, and equipment object.

[0031] In a specific embodiment, the special equipment inspection complex query generation method based on semantic constraint chain provided by the present invention includes steps 1 to 4: Step 1: Perform cross-process semantic unit identification and multi-process semantic graph construction.

[0032] The input natural language query is parsed to identify the inspection types, inspection stages, and device objects involved, and corresponding process semantic subgraphs are constructed for each. In each process semantic subgraph, semantic nodes and their dependencies are established according to the order of inspection stages. At the same time, identifier associations are established for nodes of the same device in different processes, forming a semantic graph structure with multiple parallel processes. This includes the following sub-steps: Sub-step 1.1: Perform the validation and segmentation of the natural language query statement.

[0033] Specifically, the system performs word segmentation, part-of-speech tagging, and business dictionary matching on natural language queries, breaking down statements such as "whether the defect re-inspection after the elevator's periodic inspection is qualified" and "defect rectification record in the annual inspection of pressure vessels" into processable inspection business semantic units. During execution, the system pre-sets a special equipment inspection business dictionary, which includes at least fields such as equipment category, inspection type, inspection stage, defect status, rectification status, and re-inspection conclusion.

[0034] Sub-step 1.2: Bind the inspection type, inspection stage, and equipment object.

[0035] Specifically, the system assigns the identified inspection type units and inspection stage units to different processes, such as periodic inspection, supervisory inspection, defect re-inspection, and rectification confirmation. Simultaneously, the equipment object unit is designated as the primary binding object, allowing semantic units of the same equipment in different processes to be grouped under the same equipment object. For example, when "periodic inspection," "defect re-inspection," and "rectification conclusion" appear simultaneously in natural language, the system does not treat it as a single process but rather forms two separate semantic unit groups.

[0036] Sub-step 1.3: Construction of the semantic subgraph of the execution process.

[0037] Specifically, for each process belonging to a semantic unit group, a corresponding process semantic subgraph is established. Each semantic unit is a semantic node, the order of inspection stages is used as the basis for node arrangement, the device object node is the root node of the process semantic subgraph, and the inspection type node, inspection stage node, defect status node, rectification status node, and query target node form dependencies in sequence.

[0038] Sub-step 1.4: Perform cross-process identifier association and multi-process semantic graph generation for the same device.

[0039] Specifically, using equipment object nodes as the primary index, the system compares identifiers such as equipment name, equipment number, user unit, registration number, and inspection record number in different process semantic subgraphs. This establishes cross-process identifier associations for semantic nodes belonging to the same special equipment but in different inspection processes. During execution, if the same equipment has a "defect discovery node" in the periodic inspection process and a "rectification confirmation node" or "re-inspection conclusion node" in the defect re-inspection process, these nodes are included in their corresponding node sets. Finally, the system outputs a semantic graph structure of multiple parallel processes, including multiple process semantic subgraphs, cross-process identifier associations for the same equipment, and corresponding node sets.

[0040] In this embodiment, the expression for the cross-process identifier association strength is: ; In the formula, This indicates the degree of consistency between device serial numbers; 1 indicates complete consistency, and 0 indicates inconsistency. This indicates the degree of consistency in device names, with a value ranging from 0 to 1. Indicates the degree of consistency in the use of units, with a value ranging from 0 to 1; This indicates the degree of consistency in the registration numbers, with a value ranging from 0 to 1. This indicates the degree of correlation between the inspection record number or report number, with a value ranging from 0 to 1. This indicates the degree to which the chronological order of the test matches the sequence of business processes, with a value ranging from 0 to 1. This represents the normalized value for the cross-process interval, ranging from 0 to 1. This formula integrates device identity consistency and the sequence of verification processes to determine whether different process semantic subgraphs should belong to the same device's continuous query path.

[0041] Step S120: Extract the inspection stage jump relationship of the same inspection equipment at different process nodes, and establish cross-process semantic bridging relationship for process nodes, so as to embed the cross-process semantic bridging relationship into the multi-process semantic graph to obtain the extended semantic graph structure.

[0042] In some embodiments, the special equipment inspection complex query generation method based on semantic constraint chain provided by the present invention includes the following steps in step S120: Step S121: Locate and extract the set of semantic nodes corresponding to the same inspection device from the multi-process semantic graph, and filter out the semantic nodes with process jumps from the set of semantic nodes.

[0043] Step S122: Semantic nodes with process jumps are identified as cross-process candidate connection nodes, and the inspection stage sequence of cross-process candidate connection nodes is analyzed to determine whether the cross-process candidate connection node has entered the subsequent inspection stage from the previous inspection stage.

[0044] In some embodiments, the special equipment inspection complex query generation method based on semantic constraint chain provided by the present invention further includes the following steps in step S120: Step S123: If a cross-process candidate connection node is determined to enter the subsequent inspection stage from the previous inspection stage, an inspection stage jump relationship is established, and it is determined whether the inspection stage jump relationship meets the preset bridging conditions. If so, a cross-process semantic bridging relationship is established.

[0045] Step S124: Embed the cross-process semantic bridging relationship into the multi-process semantic graph in the form of nodes, and preserve the order of the inspection stages and the connection relationship of semantic nodes in the multi-process semantic graph during embedding to obtain the extended semantic graph structure.

[0046] In a specific embodiment, the method for generating complex queries for special equipment inspection based on semantic constraint chains provided by the present invention includes step 2, which involves extracting cross-process semantic connection nodes and constructing bridging relationships. Based on the multi-process semantic graph constructed in step 1, corresponding nodes of the same equipment in different processes are extracted, their jump relationships between inspection stages are identified, and cross-process semantic bridging relationships are established for these nodes. These bridging relationships are embedded into the original semantic graph, enabling connections between different process semantic subgraphs through bridging nodes, forming an extended semantic graph structure with cross-process jump capabilities. This includes the following sub-steps: Sub-step 2.1: Perform the filtering and confirmation of the set of nodes corresponding to the same device.

[0047] Specifically, based on the cross-process identification association of the same equipment formed in step 1, the set of nodes corresponding to the same equipment is first located in the semantic graph structure of multiple parallel processes. Then, semantic nodes that can undertake process jumps are selected from the corresponding node set. For scenarios where periodic inspections and defect re-inspections of special equipment are executed concurrently, if there is a "defect discovery node" or "non-conformance conclusion node" in the periodic inspection process, and there is a "rectification and disposal node" or "re-inspection application node" in the defect re-inspection process, then the above nodes are determined as candidate cross-process connection nodes. During the selection, the single equipment name is not used as the sole criterion. Instead, the consistency of equipment number, the relevance of inspection records, the continuity of inspection stages, and the continuity of defect status are considered comprehensively to avoid mistakenly merging different equipment under the same user unit into the same cross-process connection path.

[0048] Sub-step 2.2: Perform the jump relationship identification during the verification phase.

[0049] Specifically, a phase sequence analysis is performed on the nodes in the cross-process candidate connection node set to determine whether there is a business jump relationship from the preceding inspection process to the subsequent inspection process. During execution, the defect discovery node, non-conformance conclusion node, and rectification requirement node in the periodic inspection process are used as preceding trigger nodes, and the rectification handling node, re-inspection application node, and re-inspection conclusion node in the defect re-inspection process are used as subsequent receiving nodes. In addition, if the preceding trigger node is earlier than the subsequent receiving node in terms of business phase, and there is the same defect object or the same rectification item between them, then an inspection phase jump relationship is established.

[0050] In this embodiment, the inspection stage attributes of candidate connection nodes are first read, and then sorted according to the stage order of "inspection findings before rectification and handling, rectification and handling before re-inspection application, and re-inspection application before re-inspection conclusion". Subsequently, the jump strength of each pair of preceding and following nodes is calculated, and finally, node pairs whose jump strength reaches a threshold are retained. The threshold is set between 1 and 2.5, with a higher value used when the equipment categories are complex and the processes are numerous, and a lower value used when the processes are relatively clear.

[0051] The expression for the jump strength during the testing phase is as follows: ; In the formula, Indicates the candidate connection strength; This indicates the degree of consistency between stages, with a value ranging from 0 to 1, determined by whether the preceding stage can trigger the subsequent stage. It indicates the degree of fit of the defect management closed loop, with a value range of 0-1, and is determined by the completeness of the closed loop of "discovery, rectification, re-inspection, and archiving"; This represents the normalized value for the time span, ranging from 0 to 1. This indicates the degree of parallel process complexity, with a value ranging from 0 to 1. This formula is used to determine whether a node can logically and naturally jump from the verification phase to another node, and to suppress erroneous jumps caused by excessive time spans and too many parallel processes.

[0052] Furthermore, the expression for the candidate connection strength is: ; In the formula, Indicates the strength of cross-process identifier association, with a value range of 0-1; This indicates the strength of business dependencies between nodes, with a value ranging from 0 to 2. This indicates the degree of continuation of the defect status, with a value range of 0-1, and is determined by the defect number, defect location, defect category, and rectification status. This represents the normalized value of the cross-process interval, ranging from 0 to 1. A larger interval indicates a lower reliability of the connection. This indicates the degree of interference from other parallel inspection processes on the same device, with a value ranging from 0 to 1. This formula uses device identity consistency as a foundation, business dependency and defect continuation as enhancement terms, and process interval and parallel process interference as suppression terms, to obtain candidate nodes capable of handling cross-process jumps.

[0053] Sub-step 2.3: Define cross-process bridging nodes and build bridging relationships.

[0054] Specifically, node pairs that satisfy the bridging conditions in the set of jump relationships during the inspection phase are defined as cross-process bridging node combinations. Nodes in the preceding process serve as the bridging start nodes, and nodes in the subsequent process serve as the bridging end nodes. Bridging relationships are not ordinary node dependencies, but rather cross-process semantic connections used to link different process semantic subgraphs. For example, the "non-conforming conclusion node" in the periodic inspection process can be bridged to the "re-inspection application node" in the defect re-inspection process through the "rectification requirement triggering condition"; the "defect location node" in the periodic inspection process can be bridged to the "rectification confirmation node" in the defect re-inspection process.

[0055] In this embodiment, the expression for the bridging weight is: ; In the formula, This indicates the jump strength of the inspection phase obtained in sub-step 2.2; Indicates the candidate connection strength; This indicates the degree of consistency of query targets between bridging nodes, with a value ranging from 0 to 1. This indicates the degree of fit of the defect management closed loop, with a value range of 0-1; The bridging ambiguity level is represented by a value ranging from 0 to 1, and is obtained by normalizing the number of subsequent nodes that can be connected to the same preceding node. This formula combines the confidence level of stage jumps, the confidence level of node connections, and the consistency of query targets into a bridging weight, and reduces the risk of erroneous connections caused by multiple jumps by using the bridging ambiguity level.

[0056] Sub-step 2.4: Perform bridging relationship embedding and extended semantic graph generation.

[0057] Specifically, the set of cross-process bridging nodes and the set of cross-process bridging relationships are embedded into the multi-process parallel semantic graph structure formed in step 1, creating jumpable connections between the originally independent semantic subgraphs of periodic inspection, defect re-inspection, and rectification confirmation processes. During embedding, the stage sequence relationships and node dependencies in the original process semantic subgraphs are preserved, while new cross-process bridging relationships are added to prevent bridging relationships from replacing the original internal relationships within the processes. For each bridging relationship, the system sets a cross-process reachable path marker to indicate that subsequent semantic constraint chains can continue to be passed from preceding process nodes to subsequent process nodes.

[0058] During execution, the system first writes the bridging relationships between the corresponding bridging nodes, then checks whether the bridging direction conforms to the inheritance order of the verification phase. Subsequently, it checks whether there are repeated jumps where the same node repeatedly returns to the previous process. Finally, it retains the bridging relationships whose stability reaches the threshold, forming an extended semantic graph structure. The stability threshold is set between 2 and 5, with a higher threshold for more complex processes and more parallel re-checks.

[0059] The expression for the stability of the extended graph is: ; In the formula, This represents the cross-process bridging weight obtained in sub-step 2.3; This indicates the jump strength of the inspection phase obtained in sub-step 2.2; This indicates the completeness of the query path after bridging, with a value range of 0-1, determined by whether the query target can cover the discovery, rectification, re-inspection and archiving stages. This indicates the degree of conflict in the bridging relationship, with a value ranging from 0 to 1. This indicates the risk level of forming a closed loop or repeated jumps, with a value ranging from 0 to 1. This formula is used to determine whether embedding bridging relationships can improve the integrity of cross-process query paths, while suppressing semantic graph instability caused by conflicting bridging and repeated jumps.

[0060] Step S130: By expanding the semantic graph structure, path search and path reorganization are performed on the semantic nodes to expand the semantic constraint chain of a single process and obtain a cross-process semantic constraint chain.

[0061] In some embodiments, the special equipment inspection complex query generation method based on semantic constraint chain provided by the present invention includes the following steps in step S130: Step S131: Query the query target corresponding to the natural query statement, and take the initial verification stage in the natural query statement as the starting semantic node and the query target as the target semantic node.

[0062] Step S132: Based on the starting semantic node and the target semantic node, the cross-process semantic bridging relationship traversed by the query path is determined by expanding the semantic graph structure, so as to obtain the cross-process path search boundary.

[0063] Step S133: Within the cross-process path search boundary, search for multiple arrival nodes corresponding to the starting semantic node according to the inspection process order and cross-process semantic bridging relationship, and construct a candidate cross-process semantic path set based on the starting semantic node and multiple arrival nodes.

[0064] In some embodiments, the special equipment inspection complex query generation method based on semantic constraint chain provided by the present invention further includes the following steps in step S130: Step S134: Extract the constraints corresponding to each candidate cross-process semantic path in the candidate cross-process semantic path set, and use the constraints in the preceding inspection process as inherited constraints, so that when the preceding inspection process enters the candidate cross-process semantic path, the inherited constraints are passed to the constraints corresponding to the candidate cross-process semantic path, thus obtaining the cross-process constraint integrated path set.

[0065] Step S135: The semantic nodes of the cross-process constraint integration path located in different inspection processes are reorganized, and the inspection process order and cross-process semantic bridging relationship in the candidate cross-process semantic path are preserved during the reorganization to generate a cross-process semantic constraint chain.

[0066] In a specific embodiment, the method for generating complex queries for special equipment inspection based on semantic constraint chains provided by the present invention includes step 3, which involves reconstructing cross-process semantic constraint chains and generating continuous paths. Based on the extended semantic graph obtained in step 2, path search and reorganization are performed on semantic nodes to expand the semantic constraint chains that were originally limited to a single process, enabling them to traverse different inspection processes along bridging nodes. During path generation, the constraints of the previous process are passed to the nodes of the next process, forming a set of semantic constraint chains that are continuously passed across processes. This includes the following sub-steps: Sub-step 3.1: Perform start and end node localization and path search boundary determination in the extended semantic graph.

[0067] Specifically, based on the extended semantic graph structure, the query target pointed to by the natural language query is first read, such as: "Query whether the defect re-inspection of a certain special equipment after periodic inspection has been completed" or "Query the re-inspection conclusion and the source of the original defect after rectification". The starting stage in the query semantics is determined as the starting semantic node, and the inspection conclusion, rectification status or re-inspection result to be output is determined as the target semantic node. Subsequently, according to the cross-process reachable path markers generated in step 2, the cross-process bridging nodes that the query path is allowed to pass through are determined. The path is only allowed to extend in the stage inheritance direction of "defect discovery, rectification requirements, re-inspection application, re-inspection conclusion, report archiving". It is not allowed to jump back from the subsequent re-inspection process to the previous periodic inspection process to form repeated back jumps. During execution, the stability of the extended graph and the bridging weight are used to determine the searchable range of the path, and the boundary of unreliable paths is compressed by the closed-loop risk and conflict degree to ensure that the subsequent path search will not wander disorderly between multiple inspection processes.

[0068] In this embodiment, the expression for the path search boundary strength is: ; In the formula, This indicates the stability of the extended graph obtained in step 2; This represents the cross-process bridging weight obtained in step 2; This indicates the completeness of the query path after bridging, and its value ranges from 0 to 1. This indicates the risk level of forming a closed loop or repeated jumps, with a value ranging from 0 to 1. This indicates the degree of conflict in the bridging relationship, with a value ranging from 0 to 1.

[0069] Sub-step 3.2: Perform cross-process candidate semantic path search and stage order verification.

[0070] Specifically, within the path search boundary set, starting from the initial semantic node, reachable nodes are searched sequentially according to internal process dependencies and cross-process bridging relationships. During the search, priority is given to advancing along node dependencies within the same process semantic subgraph. When a cross-process bridging node is reached, the system enters the subsequent process semantic subgraph based on the bridging direction. For situations where the same device has multiple re-inspection processes, multiple defect items, or multiple rectification records, the system generates multiple candidate cross-process semantic paths and verifies the stage order of each path to ensure it conforms to the inspection business logic. The inspection stage order verification not only checks the chronological order but also whether the business meaning can be inherited. For example, the "non-conforming conclusion node" can connect to the "rectification and disposal node," but it cannot directly jump to the "report archiving node" and bypass the re-inspection conclusion node.

[0071] During execution, the score of effective paths needs to be improved by considering the intensity of stage jumps, the continuity of internal dependencies, and the fit of the governance loop. Simultaneously, erroneous paths are suppressed by skipping critical stages and handling parallel processes. For example, the starting and target semantic nodes are read first, then the search proceeds along the internal dependency edges of the process. When a cross-process bridging node is encountered, the bridging direction and weight are read. Subsequently, candidate paths are generated, and their effectiveness is calculated based on the stage order, defect governance loop, and the presence of parallel processes. The effectiveness threshold is set between 8 and 18, with lower values ​​used when the special equipment type is relatively simple and the process is relatively clear, and higher values ​​used when there are multiple re-inspections or multiple defects.

[0072] The expression for the validity of candidate paths is: ; In the formula, Indicates the strength of the path search boundary; This indicates the jump strength of the inspection stage obtained in step 2; G represents the degree of continuity of dependencies between nodes within the process, with a value range of 0-1; G represents the degree of fit of the defect management closed loop, with a value range of 0-1. This represents the normalized value of the number of critical stages skipped in the path, with a value ranging from 0 to 1. This represents the degree of parallel process mixing, with a value ranging from 0 to 1. This formula improves the effective path score by considering the strength of stage jumps, the continuity of internal dependencies, and the fit of governance loops, while suppressing erroneous paths by skipping critical stages and mixing parallel processes.

[0073] Sub-step 3.3: Extract and pass the constraints of the preceding process across processes.

[0074] Specifically, constraints are extracted for each candidate cross-process semantic path. Equipment object constraints, inspection type constraints, defect location constraints, defect category constraints, rectification requirement constraints, and time interval constraints from the preceding process are treated as inheritable constraints. Once the preceding process path enters the cross-process bridging node, these inheritable constraints are passed to subsequent process nodes. This ensures that the re-inspection application, rectification confirmation, and re-inspection conclusion in the defect re-inspection process can inherit the equipment object, defect source, and rectification requirements from the periodic inspection process. Constraints that cannot be inherited, such as display conditions that only belong to the internal sorting of the preceding inspection report, are not passed as filtering conditions for subsequent processes to avoid incorrectly extending local query rules from the preceding process to the subsequent re-inspection process.

[0075] In this embodiment, the expression for the constraint inheritance strength is: ; In the formula, This indicates the effectiveness of the candidate paths obtained in sub-step 3.2; This indicates the candidate connection strength obtained in step 2; Indicates the degree of persistence of the defect state, with a value range of 0-1; This indicates the degree of consistency of query targets between bridging nodes, with a value ranging from 0 to 1. This indicates the degree of semantic offset of the constraint condition after it crosses the process, and its value ranges from 0 to 1. This indicates the degree of bridging ambiguity, with a value ranging from 0 to 1. This formula is used to determine whether the constraints of the preceding process can be stably passed to the subsequent process. If the defect status continues and the query target is consistent, the inheritance strength is increased; if the semantic offset or bridging ambiguity is large, the inheritance strength is decreased.

[0076] Sub-step 3.4 involves performing semantic constraint chain reorganization and continuous chain generation.

[0077] Specifically, based on the set of cross-process constraint inheritance paths, semantic nodes originally scattered across different process semantic subgraphs are reorganized into continuous semantic constraint chains. During reorganization, the order of nodes within each candidate path is preserved, and bridging node markers are inserted at cross-process bridging positions, enabling the constraint chain to clearly represent the continuous structure of "preceding process nodes, bridging nodes, and subsequent process nodes." For example, for the requirement of "querying whether a re-inspection of a special piece of equipment is qualified after a defect is found during periodic inspection," the continuous semantic constraint chain can sequentially include the equipment object node, periodic inspection node, defect discovery node, non-conformance conclusion node, rectification requirement node, cross-process bridging node, re-inspection application node, rectification confirmation node, and re-inspection conclusion node.

[0078] During execution, the system first sorts candidate paths from highest to lowest validity, then removes paths with constraint inheritance strength below a threshold. Subsequently, the nodes in the retained paths are reorganized into a continuous semantic constraint chain according to the inheritance order of the verification phase. Finally, a query path identifier is generated, which is used in step 4 to map the semantic constraint chain to a multi-table database join path. The constraint inheritance strength threshold is recommended to be set between 20 and 60, with a higher threshold for more complex defect re-inspection processes and more parallel rectification items. The link integrity threshold is recommended to be set between 300 and 1200, which can be appropriately lowered if the query target only covers a portion of the process.

[0079] In this embodiment, the expression for link integrity is: ; In the formula, This indicates the constraint inheritance strength obtained in sub-step 3.3; This indicates the effectiveness of the candidate paths obtained in sub-step 3.2; This indicates the coverage of key testing phases, with a value ranging from 0 to 1. This indicates the degree of risk of link interruption, with a value ranging from 0 to 1. This represents the percentage of duplicate nodes, with a value ranging from 0 to 1. This formula uses constraint inheritance strength and path validity to jointly determine whether the recombined semantic constraint chain is continuous and reliable, and suppresses invalid or redundant chains through link interruption risk and the percentage of duplicate nodes.

[0080] Step S140: Map the cross-process semantic constraint chain to a multi-table association path in the database, and establish inter-table connection relationships through cross-process semantic bridging relationships to generate target query statements covering multiple inspection processes.

[0081] In some embodiments, the special equipment inspection complex query generation method based on semantic constraint chain provided by the present invention includes the following steps in step S140: Step S141: Extract multiple semantic nodes of different types from the cross-process semantic constraint chain, and map each type of semantic node to the corresponding data table using the query path as the link affiliation basis to obtain the cross-process data table mapping set.

[0082] Step S142: Identify the data tables and fields corresponding to the cross-process semantic bridging relationship from the cross-process data table mapping set, and construct multi-table association paths using the cross-process semantic bridging relationship as the inter-table association.

[0083] Step S143: Transform the inheritance constraints into query conditions, and integrate the fields corresponding to the multi-table association paths, cross-process semantic bridging relationships, and preset filtering conditions into the target query statement based on the query conditions.

[0084] In a specific embodiment, the method for generating complex queries for special equipment inspection based on semantic constraint chains provided by the present invention includes step 4, which involves performing cross-process query path mapping and complete query generation. Using the cross-process semantic constraint chain generated in step 3 as input, the constraint chain is mapped to a multi-table association path in a database, where different processes correspond to different data tables. Inter-table connections are established through bridging nodes. Based on this, the filtering conditions corresponding to each node are combined to generate a complete query statement covering multiple inspection processes, enabling continuous retrieval and output of complex inspection data. This includes the following sub-steps: Sub-step 4.1: Perform data table mapping across the process semantic constraint chain.

[0085] Specifically, based on the node type in the cross-process continuous semantic constraint chain, the system maps equipment object nodes to the special equipment basic information table, periodic inspection nodes to the periodic inspection record table, defect discovery nodes to the defect registration table, rectification requirement nodes to the rectification notification table, re-inspection application nodes to the re-inspection application table, rectification confirmation nodes to the rectification confirmation table, re-inspection conclusion nodes to the re-inspection conclusion table, and report archiving nodes to the inspection report archiving table. During mapping, the query path identifier is used as the basis for link attribution to prevent multiple cross-process semantic constraint chains from being mixed and mapped. For each semantic node, the system determines its corresponding data table and fields according to the node name, node type, business stage, and inherited constraints.

[0086] Sub-step 4.2: Execute the multi-table association path generation driven by the bridge node.

[0087] Specifically, in the cross-process data table mapping set, the data tables and fields corresponding to the cross-process bridging nodes are first identified. Then, the bridging nodes are used as the basis for inter-table connections to establish cross-process multi-table association paths. For the connection between the periodic inspection process and the defect re-inspection process, the preceding process data tables may include the periodic inspection record table, defect registration table, and rectification notice table, while the subsequent process data tables may include the re-inspection application table, rectification confirmation table, and re-inspection conclusion table. Bridging fields may include equipment number, inspection record number, defect number, rectification notice number, and re-inspection application number. If the semantic constraint chain explicitly states that "re-inspection is triggered by defects discovered during periodic inspection," then the multi-table association path should preferentially use the defect number and rectification notice number as bridging fields, rather than using only the equipment number for broad connections.

[0088] Sub-step 4.3: Perform the combination of inheritance constraints and the generation of query conditions.

[0089] Specifically, based on the cross-process multi-table association path set, the inherited constraints in step 3 are transformed into specific query conditions. Equipment object constraints are used to limit the equipment number, registration number, or equipment name in the special equipment basic information table; inspection type constraints are used to limit the inspection category in the periodic inspection record table or re-inspection application table; defect status constraints are used to limit the defect category, defect location, and defect severity in the defect registration table; rectification status constraints are used to limit the rectification completion status in the rectification notice table or rectification confirmation table; re-inspection conclusion constraints are used to limit the qualified, unqualified, or need-to-continue-rectification status in the re-inspection conclusion table; and time interval constraints are used to limit the periodic inspection time, rectification period, and re-inspection completion time. For cross-process inherited conditions, the system should pass the equipment object, defect number, and rectification notice number from the preceding process to the query conditions of the subsequent process, without incorrectly passing non-filtering constraints such as the sorting fields and display fields of the preceding report to the subsequent process.

[0090] Sub-step 4.4: Execute the complete query statement generation and continuous retrieval output.

[0091] Specifically, based on the cross-process query condition set, the system combines multi-table join paths, bridging join fields, and filtering conditions into a complete query statement. The query statement should include multiple fields related to the inspection process, such as equipment basic information, periodic inspection records, defect registration records, rectification notice records, re-inspection application records, rectification confirmation records, re-inspection conclusion records, and report archiving records. During generation, the system first determines the main query table, typically the special equipment basic information table or the periodic inspection record table; then, it sequentially connects the defect registration table, rectification notice table, re-inspection application table, rectification confirmation table, and re-inspection conclusion table according to the cross-process multi-table join path; subsequently, it writes filtering conditions such as equipment object, defect status, rectification status, re-inspection conclusion, and time interval, thereby outputting continuous search results covering multiple inspection processes.

[0092] During execution, a combination of templated query generation and rule validation can be used. Templated query generation generates basic query statements in the order of "main table determination, bridging join, condition injection, and result output." Rule validation checks whether related fields are complete, whether filtering conditions come from inherited constraints, whether bridging paths cover cross-process nodes, and whether result fields correspond to the query target. For multiple candidate query statements, the statement with the highest completeness and executability is output first; if multiple statements have similar scores, the corresponding query path identifier is retained for reviewers to compare and select.

[0093] The following describes the complex query generation device for special equipment inspection based on semantic constraint chain provided by the present invention. The complex query generation device for special equipment inspection based on semantic constraint chain described below can be referred to in correspondence with the complex query generation method for special equipment inspection based on semantic constraint chain described above.

[0094] like Figure 2 As shown in one embodiment, a special equipment inspection complex query generation device based on semantic constraint chains includes a semantic graph construction module, a bridging relationship establishment module, a constraint chain reorganization module, and a query statement generation module.

[0095] The semantic graph construction module is used to parse natural queries input by users to identify inspection equipment information and construct a multi-process semantic graph based on the inspection equipment information.

[0096] The bridging relationship establishment module is used to extract the inspection stage jump relationship of the same inspection equipment at different process nodes, and to establish cross-process semantic bridging relationships for process nodes, so as to embed the cross-process semantic bridging relationships into the multi-process semantic graph and obtain the extended semantic graph structure.

[0097] The constraint chain reorganization module is used to perform path search and path reorganization on semantic nodes by extending the semantic graph structure, so as to extend the semantic constraint chain of a single process and obtain a cross-process semantic constraint chain.

[0098] The query statement generation module maps cross-process semantic constraint chains to multi-table join paths in the database and establishes inter-table connections through cross-process semantic bridging relationships to generate target query statements that cover multiple inspection processes.

[0099] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0100] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A method for generating complex queries for special equipment inspection based on semantic constraint chains, characterized in that, The method includes: The user-input natural query statement is parsed to identify inspection equipment information, and a multi-process semantic graph is constructed based on the inspection equipment information; Extract the inspection stage jump relationship of the same inspection equipment at different process nodes, and establish cross-process semantic bridging relationship for the process nodes, so as to embed the cross-process semantic bridging relationship into the multi-process semantic graph to obtain an extended semantic graph structure. The extended semantic graph structure is used to perform path search and path reorganization on semantic nodes to extend the semantic constraint chain of a single process and obtain a cross-process semantic constraint chain. The cross-process semantic constraint chain is mapped to a multi-table association path in the database, and the inter-table connection relationship is established through the cross-process semantic bridging relationship to generate a target query statement covering multiple inspection processes. The step of performing path search and path reorganization on semantic nodes through the extended semantic graph structure to extend the semantic constraint chain of a single process and obtain a cross-process semantic constraint chain includes: Query the query target corresponding to the natural query statement, and take the initial verification stage in the natural query statement as the starting semantic node and the query target as the target semantic node; Based on the starting semantic node and the target semantic node, the cross-process semantic bridging relationship traversed by the query path is determined through the extended semantic graph structure to obtain the cross-process path search boundary; Within the cross-process path search boundary, multiple arrival nodes corresponding to the starting semantic node are searched according to the inspection process order and cross-process semantic bridging relationship, and a candidate cross-process semantic path set is constructed based on the starting semantic node and multiple arrival nodes. The step of performing path search and path reorganization on semantic nodes through the extended semantic graph structure to extend the semantic constraint chain of a single process and obtain a cross-process semantic constraint chain also includes: Extract the constraints corresponding to each candidate cross-process semantic path in the candidate cross-process semantic path set, and use the constraints in the preceding inspection process as inherited constraints. When the preceding inspection process enters the candidate cross-process semantic path, the inherited constraints are passed to the constraints corresponding to the candidate cross-process semantic path to obtain the cross-process constraint integrated path set. The cross-process constraint integration path is reorganized by focusing on semantic nodes located in different inspection processes. During the reorganization, the order of inspection processes and the cross-process semantic bridging relationship in the candidate cross-process semantic path are preserved to generate the cross-process semantic constraint chain.

2. The method for generating complex queries for special equipment inspection based on semantic constraint chains according to claim 1, characterized in that, The process of parsing the user-input natural query to identify inspection equipment information and constructing a multi-process semantic graph based on the inspection equipment information includes: The system obtains the user's input natural query statement and performs word segmentation, part-of-speech tagging, and business dictionary matching on the natural query statement to parse the natural query statement into multiple business semantic units to be processed. The inspection type and inspection stage are extracted from the inspection business semantic unit, and the inspection type and inspection stage are assigned to distinguish multiple different inspection processes. Extract the device object from the inspection business semantic unit, and use the device object as a binding object to control the inspection business semantic units of the same device under different inspection processes to bind to the same device object, so as to output multiple process belonging semantic unit groups.

3. The method for generating complex queries for special equipment inspection based on semantic constraint chains according to claim 2, characterized in that, The process of parsing the user-input natural query to identify inspection equipment information and constructing a multi-process semantic graph based on the inspection equipment information also includes: Using the inspection business semantic unit as the semantic node and the device object as the root node, a corresponding process semantic sub-graph is established for each process belonging to the semantic unit group according to the order of the inspection stages. Using the device object as the main index, semantic nodes in different process semantic subgraphs are compared to establish cross-process identifiers for semantic nodes belonging to the same device object but located in different inspection processes. Based on the process semantic subgraph and the cross-process identifiers and semantic nodes of the same device object, the multi-process semantic graph is constructed; The inspection equipment information includes the inspection type, inspection stage, and equipment object.

4. The method for generating complex queries for special equipment inspection based on semantic constraint chains according to claim 1, characterized in that, The process involves extracting the inspection stage transition relationships of the same inspection equipment at different process nodes and establishing cross-process semantic bridging relationships for these process nodes. These cross-process semantic bridging relationships are then embedded into the multi-process semantic graph to obtain an extended semantic graph structure, including: The semantic node set corresponding to the same inspection device is located and extracted from the multi-process semantic graph, and semantic nodes with process jumps are selected from the semantic node set. The semantic nodes with process jumps are identified as cross-process candidate connection nodes, and the inspection stage sequence of the cross-process candidate connection nodes is analyzed to determine whether the cross-process candidate connection nodes have entered the subsequent inspection stage from the previous inspection stage.

5. The method for generating complex queries for special equipment inspection based on semantic constraint chains according to claim 4, characterized in that, The step of extracting the inspection stage jump relationship of the same inspection equipment at different process nodes and establishing cross-process semantic bridging relationships for the process nodes, so as to embed the cross-process semantic bridging relationships into the multi-process semantic graph to obtain an extended semantic graph structure, further includes: If the cross-process candidate connection node is determined to enter the subsequent inspection stage from the previous inspection stage, the inspection stage jump relationship is established, and it is determined whether the inspection stage jump relationship meets the preset bridging conditions. If so, the cross-process semantic bridging relationship is established. The cross-process semantic bridging relationship is embedded into the multi-process semantic graph in the form of nodes, and the order of the inspection stages and the connection relationship of the semantic nodes in the multi-process semantic graph are preserved during the embedding process to obtain the extended semantic graph structure.

6. The method for generating complex queries for special equipment inspection based on semantic constraint chains according to claim 1, characterized in that, The step of mapping the cross-process semantic constraint chain to a multi-table join path in the database, and establishing inter-table joins through the cross-process semantic bridging relationship to generate a target query statement covering multiple verification processes, includes: Multiple semantic nodes of different types are extracted from the cross-process semantic constraint chain, and the query path is used as the link attribution benchmark. Each type of semantic node is mapped to the corresponding data table to obtain the cross-process data table mapping set. Identify the data tables and fields corresponding to the cross-process semantic bridging relationship from the cross-process data table mapping set, and construct the multi-table association path using the cross-process semantic bridging relationship as the inter-table association; The inheritance constraints are transformed into query conditions, and the fields corresponding to the multi-table association paths, the cross-process semantic bridging relationships, and the preset filtering conditions are integrated into the target query statement based on the query conditions.

7. A terminal, comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the method according to any one of claims 1-6.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the method described in any one of claims 1-6.

Images