A building equipment operation and maintenance rule generation method based on semantic parsing and knowledge fusion

Abstract

This invention discloses a method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion, including: extracting, expanding, and collecting basic information on building equipment and documents; standardizing and parsing equipment documents and performing knowledge preprocessing to establish a building equipment information knowledge base; automatically extracting, fusing, verifying, and completing knowledge items related to building equipment operation and maintenance from the information knowledge base using a large language model combined with retrieval enhancement generation technology, and updating the knowledge base; instantiating and transforming knowledge items into rules to form operation and maintenance rules, and associating and mapping them with the equipment's digital model to form a set of operation and maintenance rules that resolve conflicts. This invention improves the efficiency of generating building equipment operation and maintenance rules, avoids reliance on manual configuration, avoids the problems of traditional extraction methods struggling to understand complex semantics and logical relationships, and avoids the difficulty of traditional methods in accurately forming executable operation and maintenance rules.

Description

Technical Field

[0001] This invention relates to the field of building operation and maintenance technology, and in particular to a method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion. Background Technology

[0002] Against the backdrop of the continuous evolution of IoT and smart building technologies, building operation and maintenance management systems have gradually acquired the capabilities for real-time equipment data acquisition, status monitoring, and alarm management. However, in actual engineering implementation, the construction and maintenance of building equipment operation and maintenance rules in the system still heavily rely on manual configuration. Operation and maintenance personnel typically need to consult equipment operation manuals, maintenance instructions, operating procedures, and manufacturer technical documents, and combine expert experience to complete the compilation and formulation of rules. Because related technical assets are mostly in unstructured forms such as natural language, tables, and process descriptions, and because the formats and terminology of documents from different manufacturers vary significantly, rule configuration efficiency is low, standards are difficult to unify, and knowledge is difficult to reuse. Furthermore, with the increasing number of building equipment and system complexity, equipment operation rules are no longer limited to single parameter thresholds, but also include multi-dimensional content such as operating conditions, interlocking logic, and maintenance strategies. Especially in asset-heavy scenarios with dense equipment assets and complex operation and maintenance logic, such as large public buildings, transportation hubs, hospitals, commercial complexes, and industrial parks, traditional manual methods are no longer sufficient to meet the rapid configuration needs of massive equipment operation and maintenance rules.

[0003] Existing technologies attempt to extract structured parameter information from device documentation using keyword matching, template recognition, or traditional information extraction algorithms. However, these methods typically only extract local parameters or content with a fixed format, lacking the ability to effectively understand complex semantic relationships, contextual logic, and conditional constraints. When device documentation suffers from inconsistent formatting, incomplete descriptions, or even low-quality, fragmented text due to poor scanning, traditional methods struggle to accurately formulate executable operation and maintenance rules.

[0004] In addition, existing methods generally lack the ability to associate with digital models of building equipment and IoT points, and still require manual completion of static mapping between rules and equipment and monitoring points, making it difficult to achieve automated generation and implementation of operation and maintenance rules. Summary of the Invention

[0005] The purpose of this invention is to provide a method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion, so as to solve the problem that traditional manual methods are difficult to meet the rapid configuration requirements of massive building equipment operation and maintenance rules.

[0006] To address the aforementioned technical problems, this invention provides a method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion, comprising:

[0007] Step S1, Building Equipment Basic Information Extraction, Expansion and Document Collection: Extract basic information and extended attributes of building equipment from the BIM model of operation and maintenance, supplement missing extended attributes, and collect documents for building equipment to form a building equipment document set.

[0008] Step S2, Building Equipment Information Knowledge Base Establishment: Standardize and preprocess the building equipment document collection to form a building equipment information knowledge base;

[0009] Step S3: Automatically extract, generate, optimize, and update knowledge items for building equipment operation and maintenance into the building equipment operation and maintenance knowledge base: Through a large language model combined with retrieval enhancement generation technology, knowledge items for building equipment operation and maintenance are automatically extracted from the building equipment information knowledge base, and the knowledge items are integrated, verified, and updated to the constructed building equipment operation and maintenance knowledge base.

[0010] Step S4, Building Equipment Operation and Maintenance Rule Set Generation: By instantiating and transforming the knowledge items updated in the building equipment operation and maintenance knowledge base into rules, operation and maintenance rules are formed. The mapping and binding relationship between each operation and maintenance rule and its corresponding equipment digital model is automatically retrieved, and conflicting operation and maintenance rules are automatically resolved or items awaiting manual review are generated, thus forming a building equipment operation and maintenance rule set.

[0011] Furthermore, the method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion provided by the present invention includes, in step S1, the method for extracting, expanding, and collecting basic building equipment information and documents, comprising:

[0012] Step S1.1: Extract basic information and extended attributes of building equipment from the BIM model of building operation and maintenance. The basic information includes equipment code and equipment name, and the extended attributes include equipment type, specifications, system, space and related relationships.

[0013] Step S1.2: For equipment in the BIM model that lacks extended attributes, the corresponding equipment object is matched and located from the as-built drawings through the equipment basic information, and its system and spatial information are obtained. Combined with the material list, equipment list, procurement data and construction documents, the specifications and models of the matching equipment are retrieved, and the mapping relationship between the equipment basic information and the specifications and models is established to supplement the missing extended attributes in the BIM model.

[0014] Step S1.3: Based on the specifications and models of the equipment in the BIM model, automatically retrieve and collect the equipment documents from the project completion data to form a building equipment document set for the target equipment.

[0015] Furthermore, in the method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion provided by the present invention, the method for establishing the building equipment information knowledge base in step S2 includes:

[0016] Step S2.1: Analyze the layout structure of each equipment document in the building equipment document collection, identify and extract the chapter hierarchy and content organization structure, and form structured parsing data containing text content and layout information;

[0017] Step S2.2: Based on the structured parsing data obtained in step S2.1, the device documents from different sources and in different formats are uniformly converted and standardized. The text content, table data, image descriptions and their relationships are encoded and stored according to the preset document graph data model to form standardized document data.

[0018] Step S2.3 involves performing knowledge segmentation on the standardized document data from step S2.2 to form several knowledge units with independent semantics, establishing the association between knowledge units and equipment objects, vectorizing the knowledge units and their associations to construct multi-dimensional semantic feature representations, and establishing a vector index library to form a building equipment information knowledge base for equipment operation and maintenance knowledge retrieval and rule generation.

[0019] Furthermore, in step 2.1, the method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion provided by the present invention uses OCR recognition technology to extract text content for scanned documents and image-type equipment documents, and combines layout analysis algorithms to restore the original chapter hierarchy and content organization structure of the document, forming structured parsing data containing text content and layout information.

[0020] Furthermore, the method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion provided by the present invention, in step 3, the method for automatically extracting, generating, optimizing, and updating knowledge items for building equipment operation and maintenance to the building equipment operation and maintenance knowledge base includes:

[0021] Step S3.1: Construct a workflow for extracting building operation and maintenance rule knowledge with a large language model as the core and combined with retrieval enhancement generation technology to automatically extract building equipment operation and maintenance knowledge items from the building equipment document set;

[0022] Step 3.2: Based on the building equipment operation and maintenance knowledge base and external knowledge source retrieval, the automatically extracted knowledge items are integrated, verified, and supplemented, and then updated to the building equipment operation and maintenance knowledge base.

[0023] Furthermore, the method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion provided by the present invention, in step 3.1, the method for automatically extracting building equipment operation and maintenance knowledge items from the building equipment document set includes:

[0024] Step S3.1.1: Determine the knowledge items to be extracted based on the preset equipment type knowledge template. The knowledge items include equipment objects, system types, monitoring parameters, state variables, threshold conditions, time constraints, interlocking relationships, operation and maintenance actions, and fault handling requirements. Among them, operation and maintenance actions include alarms, shutdowns, inspections, maintenance reminders, and work order generation.

[0025] Step S3.1.2: For each knowledge item determined in step S3.1.1, a retrieval task is automatically generated by combining the equipment name, specifications and model and knowledge item type. A mixed retrieval is performed from the building equipment information knowledge base to obtain the K knowledge fragments with the highest semantic similarity to the retrieval task, where K represents the number of candidate knowledge fragments participating in knowledge extraction and reasoning analysis.

[0026] Step S3.1.3: Use the large language model to perform semantic understanding and information extraction on the K knowledge fragments obtained in step S3.1.2, identify the parameter values, threshold ranges, execution conditions, action requirements, time periods and associated object information corresponding to the knowledge items, and form the knowledge item extraction results;

[0027] Step S3.1.4 involves performing consistency verification and standardization on the knowledge item extraction results generated in step S3.1.3. Consistency verification is used to detect whether there are conflicts in the names, units, parameter values, and logical descriptions of the same knowledge item in different chapters, tables, diagrams, and appendices. When there are multiple candidate results, the target result is determined based on the credibility of the knowledge source, the frequency of occurrence, and the semantic consistency of the context. Standardization is used to uniformly map parameter names, units of measurement, equipment codes, status descriptions, and maintenance actions according to preset equipment knowledge standards, eliminating semantic differences caused by synonyms, abbreviations, and different expressions.

[0028] Furthermore, the method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion provided by the present invention, in step S3.1.3, for knowledge content that has cross-chapter references, multiple condition constraints, causal relationship descriptions or implicit interlocking control logic, reasoning analysis is performed through the semantic association relationship between K knowledge fragments to identify the logical dependency and constraint relationship between knowledge items and generate the knowledge item extraction result of the corresponding rule expression.

[0029] Furthermore, in the method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion provided by the present invention, step 3.2, the method of fusing, verifying, and completing the automatically extracted knowledge items and updating them to the building equipment operation and maintenance knowledge base, includes:

[0030] Step S3.2.1: Construct a building equipment operation and maintenance knowledge base to store structured operation and maintenance knowledge data generated in historical projects. This knowledge base covers knowledge entities such as equipment classification, equipment parameters, industry standards, operation and maintenance experience, and failure modes, and establishes semantic relationships between these knowledge entities. Automatically extracted knowledge items are then integrated, verified, and supplemented before being added to the building equipment operation and maintenance knowledge base. During the knowledge item addition process, duplicate detection and consistency verification are performed to identify duplicate equipment knowledge entities, conflicting parameters, and abnormal data. Corrections are made through rule verification or manual review to ensure the consistency and correctness of the knowledge base data.

[0031] Step S3.2.2: Perform a completeness check on the knowledge item extraction results automatically generated in step S3.1. Check each knowledge item according to the preset knowledge template to see if valid content has been extracted, and identify knowledge items with missing or ambiguous information.

[0032] Step S3.2.3: For knowledge items with missing or ambiguous information, semantic matching and retrieval are performed first from the building equipment operation and maintenance knowledge base constructed in step S3.2.1 based on the characteristics of equipment name, equipment specifications and model, equipment type and system to which it belongs, to obtain historical project operation and maintenance knowledge of the same or similar equipment, and to complete the missing knowledge items; when there are multiple candidate results, the completion result is determined according to the credibility of the knowledge source, semantic similarity and knowledge completeness.

[0033] Step S3.2.4: When the building equipment operation and maintenance knowledge base completion mechanism described in step S3.2.3 cannot meet the knowledge item completion requirements, an online retrieval task is automatically generated. Supplementary information is obtained through Internet search engines, equipment supplier official websites and industry technology platforms. The large language model is used to evaluate the credibility of the retrieval results, perform cross-validation and structure transformation, and generate candidate knowledge items for completion.

[0034] Step S3.2.5: When the network search completion mechanism in step S3.2.4 fails to meet the knowledge integrity requirements, a manual verification and supplementation mechanism is triggered. Domain experts or maintenance personnel confirm, supplement, and correct the missing knowledge items, and update the supplemented results to the building equipment maintenance knowledge base to achieve continuous iterative updates and optimization of the building equipment maintenance knowledge base.

[0035] Furthermore, in the method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion provided by the present invention, step S4, the method for generating the building equipment operation and maintenance rule set includes:

[0036] Step S4.1: The knowledge items updated in the building equipment operation and maintenance knowledge base are combined with the basic equipment information and extended attributes to perform rule instantiation and transformation to automatically generate corresponding equipment object operation monitoring rules, abnormal alarm rules, interlock control rules, inspection rules and maintenance rules, and organize them into a hierarchical operation and maintenance rule set according to equipment level, system level and building level.

[0037] Step S4.2 involves performing conflict detection and optimization on the operation and maintenance rules in the hierarchical operation and maintenance rule set generated in step S4.1, identifying threshold conflicts, action conflicts, time conflicts, and interlocking logic conflicts between rules; when a conflict is detected, it is automatically resolved based on device priority, system level, and preset strategies, or an item is generated for manual review, so as to form a consistent and executable operation and maintenance rule set.

[0038] Step S4.3: Based on the equipment code, name, model and system information, automatically retrieve the corresponding equipment digital model and establish the association between the operation and maintenance rules and the equipment digital model. Map the monitoring parameters in the operation and maintenance rules to the attributes, measurement points and status variables of the equipment digital model to realize the online binding of the operation and maintenance rules and the equipment digital model.

[0039] Furthermore, in step S4, the method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion provided by the present invention further includes:

[0040] Step S4.4: Deploy the operation and maintenance rules that have completed the binding of the digital model of the equipment to the building operation and maintenance management platform, digital twin platform or IoT rule engine, and automatically associate the real-time operation data of IoT points according to the data mapping relationship to realize online continuous monitoring of operation and maintenance rules; when the triggering conditions of operation and maintenance rules are met, the corresponding operation and maintenance actions are automatically executed.

[0041] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0042] This invention provides a method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion. The method involves the steps of extracting and expanding basic building equipment information and collecting documents, establishing a building equipment information knowledge base, establishing, updating, and optimizing the building equipment operation and maintenance knowledge base, and generating a set of building equipment operation and maintenance rules. This is achieved by establishing a building equipment information knowledge base through deep semantic standardization parsing of documents such as equipment operation manuals and maintenance instructions. Knowledge items related to equipment operation and maintenance are automatically extracted, generated, fused, verified, and supplemented within the knowledge base. Then, the knowledge items are instantiated and transformed to generate operation and maintenance rules. This improves the efficiency of building equipment operation and maintenance rule generation and solves the problems in existing technologies, such as the high dependence on manual configuration for building equipment operation and maintenance rules and the difficulty of understanding complex semantics and logical relationships using traditional information extraction methods. It also avoids the problem in existing technologies where equipment documents have inconsistent formats, incomplete descriptions, or even low-quality and incomplete text due to poor scanning quality, making it difficult for traditional methods to accurately form executable operation and maintenance rules.

[0043] This invention provides a method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion. By introducing a large language model knowledge extraction mechanism based on RAG technology and a TOP K-semantic retrieval enhancement strategy into the steps of establishing, updating, and optimizing the building equipment operation and maintenance knowledge base, and by automatically mapping and binding operation and maintenance rules with equipment digital models and IoT points in the step of generating the building equipment operation and maintenance rule set, it achieves the automatic conversion of unstructured documents such as equipment operation manuals, maintenance instructions, and as-built data into executable operation and maintenance rules, as well as the deep integration of operation and maintenance rules with equipment digital models. This significantly reduces the reliance on human experience in configuring building equipment operation and maintenance rules, improves the efficiency and standardization of operation and maintenance rule generation, and enhances the adaptability and scalability of operation and maintenance rules in complex building systems. It can be widely applied to intelligent operation and maintenance management in equipment-intensive scenarios such as large public buildings, transportation hubs, hospitals, commercial complexes, and industrial parks. It overcomes the problem that traditional algorithms struggle to accurately generate executable operation and maintenance rules, and also solves the problem that traditional methods lack the ability to associate with building equipment digital models and IoT points, making it difficult to achieve automated generation and implementation of operation and maintenance rules. Attached Figure Description

[0044] Figure 1 This is a flowchart of a method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion. Detailed Implementation

[0045] The present invention will now be described in detail with reference to the accompanying drawings. The advantages and features of the present invention will become clearer from the following description. It should be noted that the drawings are all in a very simplified form and use non-precise proportions, and are only used to facilitate and clarify the illustration of the embodiments of the present invention.

[0046] Please refer to Figure 1 This invention provides a method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion, the technical solution of which includes the following steps:

[0047] Step S1: Extraction, expansion, and document collection of basic building equipment information. Specifically, this includes:

[0048] Step S1.1: Extract basic information and extended attributes of building equipment from the BIM model of building operation and maintenance. The basic information includes equipment code and equipment name. To extract basic information and extended attributes of building equipment from the BIM model of building operation and maintenance, a BIM model can be constructed according to preset operation and maintenance modeling standards. Modelers, based on as-built drawings, equipment lists, and relevant as-built data, represent the building equipment in a model form and label its basic information to achieve unique identification of equipment objects. Preferably, the BIM model may also include extended attributes such as equipment type, specifications, system, space, topology level, and relationships for labeling to support subsequent document collection and rule generation.

[0049] Step S1.2: For equipment lacking extended attributes in the BIM model, based on the basic equipment information extracted in step S1.1, match and locate the corresponding equipment object from the as-built drawings, obtain its system and spatial information, and in conjunction with the material list, equipment list, procurement data and construction documents, retrieve the specifications and models of the matching equipment, establish the mapping relationship between the basic equipment information and the specifications and models, and supplement the missing extended attributes in the BIM model.

[0050] Step S1.3: Based on the specifications and models of the equipment in the BIM model, automatically retrieve and collect the equipment's instruction manuals, technical specifications, operation and maintenance manuals, and related technical standards from the project completion data to form a building equipment document set for the target equipment. If no document for the corresponding specifications and models is found, expand the search based on the equipment name, category, manufacturer information, and similar model matching rules to finally form a building equipment associated document set for the target equipment.

[0051] Step S2, establishing a building equipment information knowledge base. This specifically includes:

[0052] Step S2.1 involves parsing the layout structure of each equipment document in the building equipment document collection. This includes identifying and extracting chapter titles, table of contents levels, paragraph structure, page numbers, table content, image descriptions, and chart relationships, forming structured parsed data containing both text content and layout information. For scanned documents and image-based equipment documents, OCR technology is used to extract the text content, and layout analysis algorithms are combined to restore the original chapter hierarchy and content organization structure, resulting in structured parsed data containing both text content and layout information.

[0053] Step S2.2: Based on the structured parsing data obtained in step S2.1, the device documents from different sources and in different formats are uniformly converted and standardized. The text content, table data, image descriptions and their relationships are encoded and stored according to the preset document graph data model to form standardized document data.

[0054] Step S2.3 involves segmenting the standardized document data from Step S2.2 into several knowledge units with independent semantics, taking into account chapter structure and topic information. The relationships between these knowledge units and equipment objects are then established. Furthermore, the knowledge units and their relationships are vectorized and encoded to construct multi-dimensional semantic feature representations. A vector index library is then established, forming a standardized building equipment information knowledge base for equipment operation and maintenance knowledge retrieval and rule generation. The processes of segmenting knowledge units, establishing relationships, vectorizing and encoding, constructing multi-dimensional semantic feature representations, and establishing a vector index library constitute the knowledge preprocessing process.

[0055] Step S3 involves automatically extracting, generating, optimizing, and updating knowledge items related to building equipment operation and maintenance into the building equipment operation and maintenance knowledge base. Specifically, this includes:

[0056] Step S3.1 involves constructing a knowledge extraction workflow for building operation and maintenance rules, centered on a general large language model and combined with Retrieval Enhanced Generation (RAG) technology. This workflow automatically extracts building equipment operation and maintenance knowledge items (hereinafter referred to as knowledge items) from the building equipment document set. Step S3.1 includes:

[0057] Step S3.1.1: Determine the knowledge items to be extracted based on the preset equipment type knowledge template. These knowledge items include equipment objects, system types, monitoring parameters, state variables, threshold conditions, time constraints, interlocking relationships, maintenance actions, and fault handling requirements. The maintenance actions include alarms, shutdowns, inspections, maintenance reminders, and work order generation.

[0058] Step S3.1.2: For each knowledge item determined in step S3.1.1, a retrieval task is automatically generated by combining the equipment name, specifications, and knowledge item type. A mixed retrieval is performed from the building equipment information knowledge base described in step S2.3 to obtain the K knowledge fragments (i.e., knowledge units) with the highest semantic similarity (TOP) to the retrieval task. Here, K represents the number of candidate knowledge fragments participating in knowledge extraction and reasoning analysis, and its value is an integer greater than 1, preferably 3-10. Preferably, the prompt for the retrieval task is "What is the [knowledge item] of [equipment name] with specifications [equipment specifications]?".

[0059] Step S3.1.3 utilizes a large language model to perform semantic understanding and information extraction on the K knowledge fragments obtained in Step S3.1.2, identifying information such as parameter values, threshold ranges, execution conditions, action requirements, time periods, and associated objects corresponding to each knowledge item, thus forming knowledge item extraction results. Specifically, for knowledge content with cross-chapter references, multiple conditional constraints, causal relationship descriptions, or implicit interlocking control logic, reasoning analysis is performed based on the semantic relationships between the K knowledge fragments to identify logical dependencies and constraints between knowledge items, generating corresponding rule-expressed knowledge item extraction results.

[0060] Step S3.1.4 involves performing consistency verification and standardization on the knowledge item extraction results generated in step S3.1.3. Consistency verification checks for conflicts in the names, units, parameter values, and logical descriptions of the same knowledge item across different chapters, tables, diagrams, and appendices. When multiple candidate results exist, the target result is determined based on the credibility of the knowledge source, frequency of occurrence, and semantic consistency within the context. Standardization, based on preset equipment knowledge standards, uniformly maps parameter names, units of measurement, equipment codes, status descriptions, and maintenance actions, eliminating semantic differences caused by synonyms, abbreviations, and different expressions.

[0061] Step 3.2: Based on the building equipment operation and maintenance knowledge base and external knowledge source retrieval, the knowledge item extraction results of step S3.1 are fused, verified, and supplemented, and then updated to the building equipment operation and maintenance knowledge base.

[0062] Step S3.2.1: Construct a building equipment operation and maintenance knowledge base to store structured operation and maintenance knowledge data generated in historical projects. This knowledge base covers knowledge entities such as equipment classification, equipment parameters, industry standards, operation and maintenance experience, and failure modes, and establishes semantic relationships between these knowledge entities. Automatically extracted knowledge items are then integrated, verified, and supplemented before being added to the building equipment operation and maintenance knowledge base (i.e., data entry). During the data entry process, duplicate detection and consistency verification are performed to identify duplicate equipment knowledge entities, conflicting parameters, and abnormal data. Corrections are made through rule verification or manual review to ensure the consistency and correctness of the knowledge base data.

[0063] Step S3.2.2 involves performing a completeness check on the knowledge item extraction results automatically generated in step S3.1. Each knowledge item is checked against a preset knowledge template to determine if valid content has been extracted, and missing, empty, or uncovered knowledge items are identified. For knowledge items with extracted content, a large language model is used to evaluate the clarity, explicitness, and executability of the content based on preset review prompts. Knowledge items with missing content, vague descriptions, or those that cannot be directly used to generate operation and maintenance rules are identified, and a knowledge completeness assessment result is generated. Missing or uncovered knowledge items can be categorized as fuzzy information knowledge items. The review prompts may be: "Please determine whether the following equipment operation and maintenance knowledge is complete, clear, and executable. If there are missing parameters, unclear conditions, vague action descriptions, missing time constraints, or situations where operation and maintenance rules cannot be directly formed, please point out the specific problems. Content to be evaluated: [Knowledge Content]".

[0064] In step S3.2.3, for knowledge items with missing or ambiguous information, semantic matching and retrieval are performed from the building equipment operation and maintenance knowledge base constructed in step S3.2.1 based on features such as equipment name, equipment specifications and model, equipment type and system to which it belongs, to obtain historical project operation and maintenance knowledge of the same or similar equipment, and to complete the missing knowledge items; when there are multiple candidate results, the completion result is determined according to the credibility of the knowledge source, semantic similarity and knowledge completeness.

[0065] Step S3.2.4: When the building equipment operation and maintenance knowledge base completion mechanism described in step S3.2.3 cannot meet the knowledge item completion requirements, an online retrieval task is automatically generated. Supplementary information is obtained through Internet search engines, equipment supplier official websites and industry technology platforms. The credibility assessment, cross-validation and structured transformation of the retrieval results are performed using a large language model to generate candidate knowledge items for completion.

[0066] Step S3.2.5: When neither the building equipment operation and maintenance knowledge base completion mechanism in step S3.2.3 nor the network retrieval completion mechanism in step S3.2.4 can meet the knowledge completeness requirements, a manual verification and supplementation mechanism is triggered. Domain experts or operation and maintenance personnel confirm, supplement and correct the missing knowledge items, and update the supplemented results to the building equipment operation and maintenance knowledge base after review, so as to realize the continuous iterative update and optimization of the building equipment operation and maintenance knowledge base.

[0067] Step S4: Generation and deployment of building equipment operation and maintenance rule sets. This specifically includes:

[0068] Step S4.1 involves converting the structured equipment operation and maintenance knowledge items output in step S3.2.5 into rule instantiations. This conversion, combined with the mapping relationship between basic equipment information and specifications established in step S1.2, inherits extended attributes such as spatial attributes, system attributes, and topology levels of the equipment. It automatically generates corresponding operation monitoring rules, anomaly alarm rules, interlocking control rules, inspection rules, and maintenance rules for the equipment objects, and organizes them into a hierarchical operation and maintenance rule set according to equipment level, system level, and building level.

[0069] Step S4.2 involves performing conflict detection and optimization on the operation and maintenance rules in the hierarchical operation and maintenance rule set generated in step S4.1, identifying threshold conflicts, action conflicts, time conflicts, and interlocking logic conflicts between rules; when a conflict is detected, it is automatically resolved based on device priority, system level, and preset strategy, or an item to be manually reviewed is generated to form a consistent and executable operation and maintenance rule set.

[0070] Step S4.3: Based on the equipment code, name, model and system information, automatically retrieve the corresponding equipment digital model and establish the association between the operation and maintenance rules and the equipment digital model. Map the monitoring parameters in the operation and maintenance rules to the attributes, measurement points and status variables of the equipment digital model to realize the online binding of the operation and maintenance rules and the equipment digital model. If there are multiple candidate mappings, verify and calibrate them by combining spatial location, system topology and historical operation data to ensure the accuracy of the mapping.

[0071] Step S4.4 involves deploying the operation and maintenance rules that have been bound to the digital model of the equipment to the building operation and maintenance management platform, digital twin platform, or IoT rule engine. Based on the data mapping relationship, the rules are automatically associated with the real-time operation data of IoT points to achieve continuous online monitoring. When the rule triggering conditions are met, the corresponding operation and maintenance actions such as alarms, dispatching (i.e., maintenance reminders or work order generation) or linkage control operations are automatically executed.

[0072] This invention provides a method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion. The method involves the steps of extracting and expanding basic building equipment information and collecting documents, establishing a building equipment information knowledge base, establishing, updating, and optimizing the building equipment operation and maintenance knowledge base, and generating a set of building equipment operation and maintenance rules. This is achieved by establishing a building equipment information knowledge base through deep semantic standardization parsing of documents such as equipment operation manuals and maintenance instructions. Knowledge items related to equipment operation and maintenance are automatically extracted, generated, fused, verified, and supplemented within the knowledge base. Then, the knowledge items are instantiated and transformed to generate operation and maintenance rules. This improves the efficiency of building equipment operation and maintenance rule generation and solves the problems in existing technologies, such as the high dependence on manual configuration for building equipment operation and maintenance rules and the difficulty of understanding complex semantics and logical relationships using traditional information extraction methods. It also avoids the problem in existing technologies where equipment documents have inconsistent formats, incomplete descriptions, or even low-quality and incomplete text due to poor scanning quality, making it difficult for traditional methods to accurately form executable operation and maintenance rules.

[0073] This invention provides a method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion. By introducing a large language model knowledge extraction mechanism based on RAG technology and a TOP K-semantic retrieval enhancement strategy into the steps of establishing, updating, and optimizing the building equipment operation and maintenance knowledge base, and by automatically mapping and binding operation and maintenance rules with equipment digital models and IoT points in the step of generating the building equipment operation and maintenance rule set, it achieves the automatic conversion of unstructured documents such as equipment operation manuals, maintenance instructions, and as-built data into executable operation and maintenance rules, as well as the deep integration of operation and maintenance rules with equipment digital models. This significantly reduces the reliance on human experience in configuring building equipment operation and maintenance rules, improves the efficiency and standardization of operation and maintenance rule generation, and enhances the adaptability and scalability of operation and maintenance rules in complex building systems. It can be widely applied to intelligent operation and maintenance management in equipment-intensive scenarios such as large public buildings, transportation hubs, hospitals, commercial complexes, and industrial parks. It overcomes the problem that traditional algorithms struggle to accurately generate executable operation and maintenance rules, and also solves the problem that traditional methods lack the ability to associate with building equipment digital models and IoT points, making it difficult to achieve automated generation and implementation of operation and maintenance rules.

[0074] This invention is not limited to the specific embodiments described above. Obviously, the embodiments described above are only a part of the embodiments of this invention, not all of them. All other embodiments obtained by those skilled in the art based on the described embodiments of this invention are within the scope of protection of this invention. Those skilled in the art can make other modifications and variations to this invention. Therefore, if these modifications and variations of this invention fall within the scope of the claims of this invention, then this invention also intends to include these modifications and variations.

Claims

1. A method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion, characterized in that, include: Step S1, Building Equipment Basic Information Extraction, Expansion and Document Collection: Extract basic information and extended attributes of building equipment from the BIM model of operation and maintenance, supplement missing extended attributes, and collect documents for building equipment to form a building equipment document set. Step S2, Building Equipment Information Knowledge Base Establishment: Standardize and preprocess the building equipment document collection to form a building equipment information knowledge base; Step S3: Automatically extract, generate, optimize, and update knowledge items for building equipment operation and maintenance into the building equipment operation and maintenance knowledge base: Through a large language model combined with retrieval enhancement generation technology, knowledge items for building equipment operation and maintenance are automatically extracted from the building equipment information knowledge base, and the knowledge items are integrated, verified, and updated to the constructed building equipment operation and maintenance knowledge base. Step S4, Building Equipment Operation and Maintenance Rule Set Generation: By instantiating and transforming the knowledge items updated in the building equipment operation and maintenance knowledge base into rules, operation and maintenance rules are formed. The mapping and binding relationship between each operation and maintenance rule and its corresponding equipment digital model is automatically retrieved, and conflicting operation and maintenance rules are automatically resolved or items awaiting manual review are generated, thus forming a building equipment operation and maintenance rule set.

2. The method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion according to claim 1, characterized in that, In step S1, the method for extracting, expanding, and collecting basic information on building equipment includes: Step S1.1: Extract basic information and extended attributes of building equipment from the BIM model of building operation and maintenance. The basic information includes equipment code and equipment name, and the extended attributes include equipment type, specifications, system, space and related relationships. Step S1.2: For equipment lacking extended attributes in the BIM model, the corresponding equipment object is matched and located from the as-built drawings through the equipment basic information, and its system and spatial information are obtained. Combined with the material list, equipment list, procurement data and construction documents, the specifications and models of the matching equipment are retrieved, and the mapping relationship between the equipment basic information and the specifications and models is established to supplement the missing extended attributes in the BIM model. Step S1.3: Based on the specifications and models of the equipment in the BIM model, automatically retrieve and collect the equipment documents from the project completion data to form a building equipment document set for the target equipment.

3. The method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion according to claim 1, characterized in that, In step S2, the method for establishing the building equipment information knowledge base includes: Step S2.1: Analyze the layout structure of each equipment document in the building equipment document collection, identify and extract the chapter hierarchy and content organization structure, and form structured parsing data containing text content and layout information; Step S2.2: Based on the structured parsing data obtained in step S2.1, the device documents from different sources and in different formats are uniformly converted and standardized. The text content, table data, image descriptions and their relationships are encoded and stored according to the preset document graph data model to form standardized document data. Step S2.3 involves performing knowledge segmentation on the standardized document data from step S2.2 to form several knowledge units with independent semantics, establishing the association between knowledge units and equipment objects, vectorizing the knowledge units and their associations to construct multi-dimensional semantic feature representations, and establishing a vector index library to form a building equipment information knowledge base for equipment operation and maintenance knowledge retrieval and rule generation.

4. The method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion according to claim 3, characterized in that, In step 2.1, for scanned documents and image-type device documents, OCR recognition technology is used to extract the text content, and the original chapter hierarchy and content organization structure of the document are restored by combining layout analysis algorithms to form structured parsing data containing text content and layout information.

5. The method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion according to claim 1, characterized in that, In step 3, the method for automatically extracting, generating, optimizing, and updating knowledge items for building equipment operation and maintenance into the building equipment operation and maintenance knowledge base includes: Step S3.1: Construct a workflow for extracting building operation and maintenance rule knowledge with a large language model as the core and combined with retrieval enhancement generation technology to automatically extract building equipment operation and maintenance knowledge items from the building equipment document set; Step 3.2: Based on the building equipment operation and maintenance knowledge base and external knowledge source retrieval, the automatically extracted knowledge items are integrated, verified, and supplemented, and then updated to the building equipment operation and maintenance knowledge base.

6. The method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion according to claim 5, characterized in that, In step 3.1, the method for automatically extracting building equipment operation and maintenance knowledge items from the building equipment document collection includes: Step S3.1.1: Determine the knowledge items to be extracted based on the preset equipment type knowledge template. The knowledge items include equipment objects, system types, monitoring parameters, state variables, threshold conditions, time constraints, interlocking relationships, operation and maintenance actions, and fault handling requirements. Among them, operation and maintenance actions include alarms, shutdowns, inspections, maintenance reminders, and work order generation. Step S3.1.2: For each knowledge item determined in step S3.1.1, a retrieval task is automatically generated by combining the equipment name, specifications and model and knowledge item type. A mixed retrieval is performed from the building equipment information knowledge base to obtain the K knowledge fragments with the highest semantic similarity to the retrieval task, where K represents the number of candidate knowledge fragments participating in knowledge extraction and reasoning analysis. Step S3.1.3: Use the large language model to perform semantic understanding and information extraction on the K knowledge fragments obtained in step S3.1.2, identify the parameter values, threshold ranges, execution conditions, action requirements, time periods and associated object information corresponding to the knowledge items, and form the knowledge item extraction results; Step S3.1.4 involves performing consistency verification and standardization on the knowledge item extraction results generated in step S3.1.

3. Consistency verification is used to detect whether there are conflicts in the names, units, parameter values, and logical descriptions of the same knowledge item in different chapters, tables, diagrams, and appendices. When there are multiple candidate results, the target result is determined based on the credibility of the knowledge source, the frequency of occurrence, and the semantic consistency of the context. Standardization is used to uniformly map parameter names, units of measurement, equipment codes, status descriptions, and maintenance actions according to preset equipment knowledge standards, eliminating semantic differences caused by synonyms, abbreviations, and different expressions.

7. The method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion according to claim 6, characterized in that, In step S3.1.3, for knowledge content that has cross-chapter references, multiple condition constraints, causal relationship descriptions, or implicit interlocking control logic, reasoning analysis is performed through the semantic association relationship between K knowledge fragments to identify the logical dependency and constraint relationship between knowledge items and generate the knowledge item extraction results expressed in the corresponding rules.

8. The method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion according to claim 5, characterized in that, In step 3.2, the method for fusing, verifying, and completing the automatically extracted knowledge items and updating them in the building equipment operation and maintenance knowledge base includes: Step S3.2.1: Construct a building equipment operation and maintenance knowledge base to store structured operation and maintenance knowledge data generated in historical projects. This knowledge base covers knowledge entities such as equipment classification, equipment parameters, industry standards, operation and maintenance experience, and failure modes, and establishes semantic relationships between these knowledge entities. Automatically extracted knowledge items are then integrated, verified, and supplemented before being added to the building equipment operation and maintenance knowledge base. During the knowledge item addition process, duplicate detection and consistency verification are performed to identify duplicate equipment knowledge entities, conflicting parameters, and abnormal data. Corrections are made through rule verification or manual review to ensure the consistency and correctness of the knowledge base data. Step S3.2.2: Perform a completeness check on the knowledge item extraction results automatically generated in step S3.

1. Check each knowledge item according to the preset knowledge template to see if valid content has been extracted, and identify knowledge items with missing or ambiguous information. Step S3.2.3: For knowledge items with missing or ambiguous information, semantic matching and retrieval are performed first from the building equipment operation and maintenance knowledge base constructed in step S3.2.1 based on the characteristics of equipment name, equipment specifications and model, equipment type and system to which it belongs, to obtain historical project operation and maintenance knowledge of the same or similar equipment, and to complete the missing knowledge items; when there are multiple candidate results, the completion result is determined according to the credibility of the knowledge source, semantic similarity and knowledge completeness. Step S3.2.4: When the building equipment operation and maintenance knowledge base completion mechanism described in step S3.2.3 cannot meet the knowledge item completion requirements, an online retrieval task is automatically generated. Supplementary information is obtained through Internet search engines, equipment supplier official websites and industry technology platforms. The large language model is used to evaluate the credibility of the retrieval results, perform cross-validation and structure transformation, and generate candidate knowledge items for completion. Step S3.2.5: When the network search completion mechanism in step S3.2.4 fails to meet the knowledge integrity requirements, a manual verification and supplementation mechanism is triggered. Domain experts or maintenance personnel confirm, supplement, and correct the missing knowledge items, and update the supplemented results to the building equipment maintenance knowledge base to achieve continuous iterative updates and optimization of the building equipment maintenance knowledge base.

9. The method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion according to claim 1, characterized in that, In step S4, the method for generating the building equipment operation and maintenance rule set includes: Step S4.1: The knowledge items updated in the building equipment operation and maintenance knowledge base are combined with the basic equipment information and extended attributes to perform rule instantiation and transformation to automatically generate corresponding equipment object operation monitoring rules, abnormal alarm rules, interlock control rules, inspection rules and maintenance rules, and organize them into a hierarchical operation and maintenance rule set according to equipment level, system level and building level. Step S4.2 involves performing conflict detection and optimization on the operation and maintenance rules in the hierarchical operation and maintenance rule set generated in step S4.1, identifying threshold conflicts, action conflicts, time conflicts, and interlocking logic conflicts between rules; when a conflict is detected, it is automatically resolved based on device priority, system level, and preset strategies, or an item is generated for manual review, so as to form a consistent and executable operation and maintenance rule set. Step S4.3: Based on the equipment code, name, model and system information, automatically retrieve the corresponding equipment digital model and establish the association between the operation and maintenance rules and the equipment digital model. Map the monitoring parameters in the operation and maintenance rules to the attributes, measurement points and status variables of the equipment digital model to realize the online binding of the operation and maintenance rules and the equipment digital model.

10. The method for generating building equipment operation and maintenance rules based on semantic parsing and knowledge fusion according to claim 9, characterized in that, In step S4, the method for generating the building equipment operation and maintenance rule set further includes: Step S4.4: Deploy the operation and maintenance rules that have completed the binding of the digital model of the equipment to the building operation and maintenance management platform, digital twin platform or IoT rule engine, and automatically associate the real-time operation data of IoT points according to the data mapping relationship to realize online continuous monitoring of operation and maintenance rules; when the triggering conditions of operation and maintenance rules are met, the corresponding operation and maintenance actions are automatically executed.

Images