Building physical fuzzy query method based on large language model and medium

Abstract

The invention discloses a building physical fuzzy query method based on a large language model and a medium. The method comprises the following steps: acquiring a current to-be-queried user problem in real time in a building physical fuzzy query system, and analyzing the current to-be-queried user problem to obtain an analysis result of the current to-be-queried user problem; through a multi-type hierarchical fuzzy matching method, comparing the problem analysis result of the current to-be-queried user to generate current matching query key information; inputting the current matching query key information into a pre-constructed building data dictionary knowledge graph and a building physical vector database, and generating a current information fusion candidate answer; and verifying the current information fusion candidate answer through a large language model answer matching degree verification method, and if the verification is passed, obtaining a target fuzzy query answer. The problems of low query efficiency and incompleteness caused by lack of a fuzzy query mechanism for building physical system characteristics are solved, and the building physical fuzzy query efficiency and accuracy are improved.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence data processing, and in particular to a method and medium for fuzzy query of building physics based on a large language model. Background Technology

[0002] In the management, operation and maintenance and energy consumption analysis of building physics systems, users often need to query the status, energy consumption or risk information of system objects, such as the status, energy consumption or risk information of objects such as fans, water pumps or chillers.

[0003] In the process of realizing this invention, the inventors discovered the following defects in the prior art: existing question-answering systems or intelligent agents based on large language models mainly rely on knowledge graphs, but lack fuzzy query mechanisms for the characteristics of building physics systems. They generally suffer from inconsistencies in terminology, hierarchical differences, or semantic ambiguity, resulting in insufficient coverage and low accuracy of query results. Summary of the Invention

[0004] This invention provides a building physics fuzzy query method and medium based on a large language model, so as to improve the accuracy and efficiency of building physics fuzzy query.

[0005] According to one aspect of the present invention, a fuzzy query method for building physics based on a large language model is provided, comprising:

[0006] In the building physics fuzzy query system, the current user's question to be queried is obtained in real time, and the current user's question to be queried is parsed to obtain the parsing result of the current user's question to be queried.

[0007] By comparing the parsing results of the current user's question with the preset multi-type hierarchical fuzzy matching method, the current matching query key information is generated.

[0008] The multi-type hierarchical fuzzy matching method includes: keyword precise matching sub-method, synonym extended matching sub-method, hyponym / hypernym extended matching sub-method, and relation extended matching sub-method;

[0009] The key information of the current matching query is input into a pre-built building data dictionary knowledge graph and building physics vector database to generate candidate answers for current information fusion.

[0010] The current information fusion candidate answer is verified by a preset large language model answer matching degree verification method. If the verification is successful, the target fuzzy query answer is obtained.

[0011] According to another aspect of the present invention, a building physics fuzzy query device based on a large language model is provided, comprising:

[0012] The module for determining the parsing result of the current user's question is used in the building physics fuzzy query system to obtain the current user's question in real time, and to parse the current user's question to obtain the parsing result of the current user's question.

[0013] The current matching query key information generation module is used to compare the parsing results of the current user's question with the preset multi-type hierarchical fuzzy matching method to generate the current matching query key information;

[0014] The multi-type hierarchical fuzzy matching method includes: keyword precise matching sub-method, synonym extended matching sub-method, hyponym / hypernym extended matching sub-method, and relation extended matching sub-method;

[0015] The current information fusion candidate answer generation module is used to input the key information of the current matching query into a pre-built building data dictionary knowledge graph and building physics vector database to generate the current information fusion candidate answer;

[0016] The target fuzzy query answer determination module is used to verify the current information fusion candidate answer using a preset large language model answer matching degree verification method. If the verification passes, the target fuzzy query answer is obtained.

[0017] According to another aspect of the present invention, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the building physics fuzzy query method based on a large language model as described in any embodiment of the present invention.

[0018] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the building physics fuzzy query method based on a large language model as described in any embodiment of the present invention.

[0019] The technical solution of this invention, through a building physics fuzzy query system, acquires the current user's query question in real time, parses the current user's query question to obtain the parsing result; compares the parsing result of the current user's query question using a multi-type hierarchical fuzzy matching method to generate key information for the current matching query; inputs the key information for the current matching query into a pre-constructed building data dictionary knowledge graph and building physics vector database to generate current information fusion candidate answers; and verifies the current information fusion candidate answers using a large language model answer matching degree verification method. If the verification passes, the target fuzzy query answer is obtained. This solves the problem of low query efficiency and incompleteness caused by the lack of a fuzzy query mechanism tailored to the characteristics of building physics systems, improves the efficiency and accuracy of building physics fuzzy queries, enhances the comprehensiveness and flexibility of building physics fuzzy queries, and improves the user experience of fuzzy queries.

[0020] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a flowchart of a building physics fuzzy query method based on a large language model according to Embodiment 1 of the present invention;

[0023] Figure 2 This is a schematic diagram of the structure of a building physics fuzzy query device based on a large language model according to Embodiment 2 of the present invention;

[0024] Figure 3 This is a schematic diagram of the structure of an electronic device provided according to Embodiment 3 of the present invention. Detailed Implementation

[0025] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0026] It should be noted that the terms "target," "current," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0027] It is worth noting that the information collected in the technical solution of this application is information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data all comply with the relevant laws, regulations and standards of the relevant countries and regions, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entry points for users to choose to authorize or refuse; if the user chooses to refuse, the process will proceed to the expert decision-making process.

[0028] Example 1

[0029] Figure 1 The flowchart of the building physics fuzzy query method based on a large language model is provided in Embodiment 1 of the present invention. This embodiment is applicable to the situation where a fuzzy query matching process is performed on user questions in the field of building physics using a large language model. The method can be executed by a building physics fuzzy query device based on a large language model, which can be implemented in hardware and / or software.

[0030] Correspondingly, such as Figure 1 As shown, the method includes:

[0031] S110. In the building physics fuzzy query system, the current user question to be queried is obtained in real time, and the current user question to be queried is parsed to obtain the parsing result of the current user question to be queried.

[0032] In this embodiment, the building physics fuzzy query system can receive queries from different users in real time, that is, obtain the current query question in real time. For example, the current query question could be "What is the maximum power parameter of the wind turbine?". Furthermore, this current query question can be parsed to obtain the corresponding parsing result.

[0033] In detail, the results of the current query for a user's question can include at least one query keyword. Continuing the previous example, the results of the current query for a user's question can include wind turbine and maximum power.

[0034] S120. By using a preset multi-type hierarchical fuzzy matching method, the parsing results of the current user question to be queried are compared to generate the current matching query key information.

[0035] The multi-type hierarchical fuzzy matching method includes: keyword precise matching sub-method, synonym extended matching sub-method, hyponym / hypernym extended matching sub-method, and relation extended matching sub-method.

[0036] Among them, the multi-type hierarchical fuzzy matching method is a hierarchical matching method based on multi-round reasoning within the ReAct framework. The keyword exact matching sub-method can directly retrieve queries using the query keywords in the parsed results. The synonym expansion matching sub-method can retrieve queries using synonyms associated with the query keywords in the parsed results. The hyponym / hypernym expansion matching sub-method can retrieve queries using hyponyms / hypernyms associated with the query keywords in the parsed results. The relation expansion matching sub-method can retrieve queries using inclusion or subordination relationships associated with the query keywords in the parsed results.

[0037] Optionally, the step of comparing the parsing results of the current user's question using a preset multi-type hierarchical fuzzy matching method to generate key information for the current matching query includes: comparing the parsing results of the current user's question using the keyword precise matching sub-method to determine whether there is precise matching information for the target keyword; wherein, the multi-type hierarchical fuzzy matching method is a hierarchical matching method based on multi-round reasoning of the ReAct framework; if it exists, the precise matching information for the target keyword is determined as the key information for the current matching query; if there is no precise matching information for the target keyword, the parsing results of the current user's question are matched using the synonym expansion matching sub-method; determining whether there is target synonym expansion matching information, and if it exists, determining the target synonym expansion matching information as the key information for the current matching query.

[0038] Optionally, after determining whether target synonym expansion matching information exists, the method further includes: if target synonym expansion matching information does not exist, then matching the current query user question parsing result is performed using the hypernym / hypernym concept expansion matching sub-method; determining whether target hypernym / hypernym concept expansion matching information exists, and if it exists, then determining the target hypernym / hypernym concept expansion matching information as the current matching query key information.

[0039] Optionally, after determining whether target hypernym / hypernym concept extension matching information exists, the method further includes: if target hypernym / hypernym concept extension matching information does not exist, then the current query user question parsing result is matched using the relationship extension matching sub-method; if target relationship extension matching information exists, then the target relationship extension matching information is determined as the current matching query key information.

[0040] In this embodiment, the keyword precise matching sub-method is first used to compare the parsing results of the current user's query to determine whether there is precise matching information for the target keyword. For example, the existence of precise matching information for the target keyword is determined by "wind turbine" and "maximum power". If it exists, the matching information is directly identified as the key information for the current matching query, and the key information for the current matching query is input into the pre-built building data dictionary knowledge graph and building physics vector database to generate candidate answers for current information fusion.

[0041] Conversely, if no exact match information for the target keyword exists, the current query result needs to be matched using the synonym expansion matching sub-method. This involves using the dictionary or semantic similarity in the synonym expansion matching sub-method to find synonyms. It is necessary to determine whether target synonym expansion matching information exists. If it does, the target synonym expansion matching information is determined as the key information for the current matching query.

[0042] If no target synonym expansion matching information exists, the current query result needs to be matched using the hypernym / hypernym expansion matching sub-method. Specifically, hierarchical expansion is performed based on the query keywords; for example, synonyms for "fan" include "air blower" or "exhaust fan." It is then determined whether target hypernym / hypernym expansion matching information exists; if so, this information is identified as the key information for the current matching query.

[0043] If no target hierarchical concept expansion matching information exists, further matching processing is needed on the current query user question parsing results through the relationship expansion matching sub-method; that is, the search scope is expanded through inclusion or subordination relationships, and then the corresponding query keywords after the expansion scope are determined, so that the existence of target relationship expansion matching information can be confirmed, and the target relationship expansion matching information is determined as the current matching query key information.

[0044] The advantages of this setup are as follows: by using multi-type hierarchical fuzzy matching methods to determine the key information of the current matching query, a fuzzy query mechanism tailored to the characteristics of the building physics system can be constructed, improving the efficiency and accuracy of building physics fuzzy queries, as well as their comprehensiveness and flexibility; and the query process strictly follows the extension logic, ensuring traceability of results and avoiding illusory answers.

[0045] S130. Input the current matching query key information into the pre-built building data dictionary knowledge graph and building physics vector database to generate current information fusion candidate answers.

[0046] Optionally, the step of inputting the current matching query key information into a pre-built building data dictionary knowledge graph and building physics vector database to generate current information fusion candidate answers includes: inputting the current matching query key information into the pre-built building data dictionary knowledge graph to locate at least one target-related node and target neighborhood extension information results corresponding to each target-related node; performing semantic recall processing operations in the pre-built building physics vector database based on the current matching query key information to obtain target vector recall results; and performing distillation processing on each target neighborhood extension information result and target vector recall result using a pre-set agent distillation method to generate the current information fusion candidate answers.

[0047] In this embodiment, one or more target-related nodes can be determined from a pre-built building data dictionary knowledge graph based on the current matching query key information. Each target-related node can be matched with corresponding target neighborhood extended information results. The target-related nodes can be standard objects of different types. The target neighborhood extended information results can be information describing the standard object information points, standard object relationships, and hierarchical concepts of the standard objects corresponding to the target-related nodes.

[0048] Optionally, before obtaining the current user's query question in real time and parsing the current user's query question to obtain the parsing result in the building physics fuzzy query system, the method further includes: obtaining standard building data description document information, generating multiple standard related nodes through a preset multi-dimensional building data dictionary node extraction method, and obtaining standard neighborhood extension information results based on the standard building data description document information of each standard related node; combining the standard related nodes and the standard neighborhood extension information results to generate the building data dictionary knowledge graph; wherein, the standard related nodes are standard objects of different types, and the standard objects include at least one of the following: fans, pumps, and chillers; the standard neighborhood extension information results include standard object information points, standard object relationships, and standard object hierarchical concepts; the standard object information points include at least one of the following: power and flow rate; the standard object relationships include at least one of the following: inclusion object relationship and subordinate object relationship.

[0049] In this embodiment, a building data dictionary knowledge graph also needs to be constructed in advance. Specifically, it is necessary to obtain standard building data description document information and generate multiple standard related nodes using a multi-dimensional building data dictionary node extraction method. Objects such as fans, water pumps, or chillers can be extracted to construct related nodes. Furthermore, it is necessary to obtain standard object information points, standard object relationships, and hierarchical concepts of standard objects. Specifically, this may include power and flow rate, or information such as contained object relationships and subordinate object relationships, to construct standard neighborhood extension information results. This allows the building data dictionary knowledge graph to be fully constructed.

[0050] In addition, it is necessary to build a building physics vector database. Specifically, it is necessary to obtain object description documents, operation and maintenance documents, and risk statement documents. The above documents need to be used for entity identification. Based on the identified entity objects, the relevant descriptions of each entity object can be expanded, and the expanded entity descriptions can be vectorized and stored. This completes the construction of the building physics vector database.

[0051] Furthermore, based on the current matching query key information, semantic recall processing needs to be performed in the building physics vector database to obtain the target vector recall result. This is the process of performing vector semantic recall processing on the query key information, which can obtain the corresponding relevant text information. Correspondingly, an intelligent agent distillation method can be used to distill the results of each target neighborhood expansion information and the target vector recall result to generate the current information fusion candidate answer.

[0052] The advantage of this setup is that by pre-constructing a building data dictionary knowledge graph and a building physics vector database, the current information can be fused to generate accurate candidate answers based on the key information of the current matching query. The building data dictionary knowledge graph can be used first to ensure the logical structure, and then the building physics vector database can be used to supplement the semantic content, balancing accuracy and coverage, and achieving a better and more accurate determination of fuzzy query answers.

[0053] S140. The current information fusion candidate answer is verified by a preset large language model answer matching degree verification method. If the verification is successful, the target fuzzy query answer is obtained.

[0054] Among them, the large language model answer matching degree verification method can be a method of real-time verification based on the current information fusion candidate answer generated in each round.

[0055] Optionally, the step of verifying the current information fusion candidate answer using a preset large language model answer matching degree verification method, and obtaining the target fuzzy query answer if the verification passes, includes: calculating the matching degree between the current user question to be queried and the current information fusion candidate answer using the large language model answer matching degree verification method to obtain the current answer matching degree; obtaining a preset answer matching degree threshold, and verifying whether the current answer matching degree meets the requirements of the answer matching degree threshold; if it meets the requirements, the current information fusion candidate answer is determined as the target fuzzy query answer; if it does not meet the requirements, the operation of comparing the parsing results of the current user question to be queried using a preset multi-type hierarchical fuzzy matching method to generate the current matching query key information is returned, until the target fuzzy query answer is generated or all multi-type hierarchical fuzzy matching methods are traversed.

[0056] In this embodiment, a large language model answer matching degree verification method is used to automatically calculate the matching degree between the generated current information fusion candidate answer and the current query user question in each round, thus obtaining the corresponding current answer matching degree. After calculating the current answer matching degree, it needs to be compared with the answer matching degree threshold. If the threshold is met, it means that the current candidate answer meets the requirements and can be determined as the target fuzzy query answer.

[0057] Conversely, a further query operation is required, which means returning to execute the preset multi-type hierarchical fuzzy matching method to compare the parsing results of the current user question and generate the current matching query key information, until the target fuzzy query answer is generated or all multi-type hierarchical fuzzy matching methods have been traversed. Additionally, an iteration round threshold can be set, allowing the process to exit once the requirements are met.

[0058] Correspondingly, after obtaining the answer to the target fuzzy query, the answer can be provided to the user in real time to improve the user experience.

[0059] The technical solution of this invention, through a building physics fuzzy query system, acquires the current user's query question in real time, parses the current user's query question to obtain the parsing result; compares the parsing result of the current user's query question using a multi-type hierarchical fuzzy matching method to generate key information for the current matching query; inputs the key information for the current matching query into a pre-constructed building data dictionary knowledge graph and building physics vector database to generate current information fusion candidate answers; and verifies the current information fusion candidate answers using a large language model answer matching degree verification method. If the verification passes, the target fuzzy query answer is obtained. This solves the problem of low query efficiency and incompleteness caused by the lack of a fuzzy query mechanism tailored to the characteristics of building physics systems, improves the efficiency and accuracy of building physics fuzzy queries, enhances the comprehensiveness and flexibility of building physics fuzzy queries, and improves the user experience of fuzzy queries.

[0060] Example 2

[0061] Figure 2 This is a schematic diagram of a building physics fuzzy query device based on a large language model, provided in Embodiment 2 of the present invention. The building physics fuzzy query device based on a large language model provided in this embodiment can be implemented by software and / or hardware, and can be configured in a terminal device or server to implement a building physics fuzzy query method based on a large language model according to the present invention. Figure 2 As shown, the device includes: a current query user question parsing result determination module 210, a current matching query key information generation module 220, a current information fusion candidate answer generation module 230, and a target fuzzy query answer determination module 240.

[0062] The current query user question parsing result determination module 210 is used to obtain the current query user question in real time in the building physics fuzzy query system, and parse the current query user question to obtain the current query user question parsing result.

[0063] The current matching query key information generation module 220 is used to compare the parsing results of the current user question to be queried with a preset multi-type hierarchical fuzzy matching method to generate current matching query key information;

[0064] The multi-type hierarchical fuzzy matching method includes: keyword precise matching sub-method, synonym extended matching sub-method, hyponym / hypernym extended matching sub-method, and relation extended matching sub-method;

[0065] The current information fusion candidate answer generation module 230 is used to input the current matching query key information into the pre-built building data dictionary knowledge graph and building physics vector database to generate the current information fusion candidate answer;

[0066] The target fuzzy query answer determination module 240 is used to verify the current information fusion candidate answer through a preset large language model answer matching degree verification method. If the verification is successful, the target fuzzy query answer is obtained.

[0067] The technical solution of this invention, through a building physics fuzzy query system, acquires the current user's query question in real time, parses the current user's query question to obtain the parsing result; compares the parsing result of the current user's query question using a multi-type hierarchical fuzzy matching method to generate key information for the current matching query; inputs the key information for the current matching query into a pre-constructed building data dictionary knowledge graph and building physics vector database to generate current information fusion candidate answers; and verifies the current information fusion candidate answers using a large language model answer matching degree verification method. If the verification passes, the target fuzzy query answer is obtained. This solves the problem of low query efficiency and incompleteness caused by the lack of a fuzzy query mechanism tailored to the characteristics of building physics systems, improves the efficiency and accuracy of building physics fuzzy queries, enhances the comprehensiveness and flexibility of building physics fuzzy queries, and improves the user experience of fuzzy queries.

[0068] Based on the above embodiments, the current matching query key information generation module 220 can be specifically used to: compare the current query user question parsing results with the keyword precise matching sub-method to determine whether there is target keyword precise matching information; wherein, the multi-type hierarchical fuzzy matching method is a hierarchical matching method based on multi-round reasoning of the ReAct framework; if it exists, the target keyword precise matching information is determined as the current matching query key information; if there is no target keyword precise matching information, the current query user question parsing results are matched with the synonym expansion matching sub-method; determine whether target synonym expansion matching information exists, and if it exists, determine the target synonym expansion matching information as the current matching query key information.

[0069] Based on the above embodiments, the current matching query key information generation module 220 can also be specifically used for: after determining whether target synonym extended matching information exists, if target synonym extended matching information does not exist, then matching the current query user question parsing result is performed through the hypernym and hyponym concept extended matching sub-method; determining whether target hypernym and hyponym concept extended matching information exists, and if it exists, then determining the target hypernym and hyponym concept extended matching information as the current matching query key information.

[0070] Based on the above embodiments, the current matching query key information generation module 220 can also be specifically used to: after determining whether there is target superordinate concept extended matching information, if there is no target superordinate concept extended matching information, then the current query user question parsing result is matched by the relationship extended matching sub-method; if there is target relationship extended matching information, then the target relationship extended matching information is determined as the current matching query key information.

[0071] Based on the above embodiments, the current information fusion candidate answer generation module 230 can be specifically used to: input the current matching query key information into a pre-built building data dictionary knowledge graph, locate at least one target-related node, and target neighborhood extension information results corresponding to each target-related node; perform semantic recall processing operation in a pre-built building physical vector database according to the current matching query key information to obtain target vector recall results; and perform distillation processing on each target neighborhood extension information result and target vector recall result through a pre-set intelligent agent distillation method to generate the current information fusion candidate answer.

[0072] Based on the above embodiments, a building data dictionary knowledge graph generation module is also included. Specifically, this module is used to: obtain standard building data description document information before, in the building physics fuzzy query system, acquire the current user query question in real time, parse the current user query question, and obtain the parsing result of the current user query question; generate multiple standard related nodes using a preset multi-dimensional building data dictionary node extraction method; and obtain standard neighborhood extension information results based on the standard building data description document information for each standard related node; combine and construct the standard related nodes and the standard neighborhood extension information results to generate the building data dictionary knowledge graph; wherein, the standard related nodes are standard objects of different types, and the standard objects include at least one of the following: fans, pumps, and chillers; the standard neighborhood extension information results include standard object information points, standard object relationships, and standard object hierarchical concepts; the standard object information points include at least one of the following: power and flow rate; the standard object relationships include at least one of the following: inclusion object relationships and subordinate object relationships.

[0073] Based on the above embodiments, the target fuzzy query answer determination module 240 can be specifically used to: calculate the matching degree between the current user question to be queried and the current information fusion candidate answer through the large language model answer matching degree verification method, and obtain the current answer matching degree; obtain a preset answer matching degree threshold, and verify whether the current answer matching degree meets the requirements of the answer matching degree threshold. If it meets the requirements, the current information fusion candidate answer is determined as the target fuzzy query answer; if it does not meet the requirements, the operation of comparing the parsing results of the current user question to be queried through the preset multi-type hierarchical fuzzy matching method and generating the current matching query key information is returned, until the target fuzzy query answer is generated or all multi-type hierarchical fuzzy matching methods are traversed.

[0074] The building physics fuzzy query device based on a large language model provided in this embodiment of the invention can execute the building physics fuzzy query method based on a large language model provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0075] Example 3

[0076] Figure 3A schematic diagram of an electronic device 10, which can be used to implement Embodiment 3 of the present invention, is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0077] like Figure 3 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0078] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0079] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as the building physics fuzzy query method based on a large language model.

[0080] In some embodiments, the building physics fuzzy query method based on a large language model can be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the building physics fuzzy query method based on a large language model described above can be performed. Alternatively, in other embodiments, processor 11 can be configured to perform the building physics fuzzy query method based on a large language model by any other suitable means (e.g., by means of firmware).

[0081] The method includes: in a building physics fuzzy query system, acquiring the current user's query question in real time, parsing the current user's query question to obtain the current user's query question parsing result; comparing the current user's query question parsing result with a preset multi-type hierarchical fuzzy matching method to generate current matching query key information; wherein, the multi-type hierarchical fuzzy matching method includes: keyword precise matching sub-method, synonym extended matching sub-method, hyponym / hypernym extended matching sub-method, and relation extended matching sub-method; inputting the current matching query key information into a pre-constructed building data dictionary knowledge graph and building physics vector database to generate current information fusion candidate answers; verifying the current information fusion candidate answers with a preset large language model answer matching degree verification method, and if the verification passes, obtaining the target fuzzy query answer.

[0082] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0083] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0084] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0085] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0086] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0087] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0088] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0089] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

[0090] Example 4

[0091] Embodiment 4 of the present invention also provides a computer-readable storage medium, wherein the computer-readable instructions, when executed by a computer processor, are used to execute a building physics fuzzy query method based on a large language model. The method includes: in a building physics fuzzy query system, acquiring the current user question to be queried in real time, parsing the current user question to be queried, and obtaining the parsing result of the current user question to be queried; comparing the parsing result of the current user question to be queried using a preset multi-type hierarchical fuzzy matching method to generate current matching query key information; wherein the multi-type hierarchical fuzzy matching method includes: a keyword precise matching sub-method, a synonym extended matching sub-method, a hypothetical concept extended matching sub-method, and a relation extended matching sub-method; inputting the current matching query key information into a pre-constructed building data dictionary knowledge graph and a building physics vector database to generate current information fusion candidate answers; and verifying the current information fusion candidate answers using a preset large language model answer matching degree verification method. If the verification passes, the target fuzzy query answer is obtained.

[0092] Of course, the computer-executable instructions provided in the embodiments of the present invention, which include a computer-readable storage medium, are not limited to the method operations described above, but can also perform related operations in the building physics fuzzy query based on a large language model provided in any embodiment of the present invention.

[0093] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0094] It is worth noting that in the above embodiments of building physics fuzzy query based on large language model, the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy differentiation and are not used to limit the scope of protection of the present invention.

[0095] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A building physics fuzzy query method based on a large language model, characterized in that, The application relates to a building physics fuzzy query system. Real-time acquisition of a current user query problem, and analysis of the current user query problem to obtain a current user query problem analysis result; Comparison of the current user query problem analysis result through a preset multi-type hierarchical fuzzy matching method to generate current matching query key information; The multi-type hierarchical fuzzy matching method comprises a keyword accurate matching sub-method, a synonym expansion matching sub-method, an upper-lower concept expansion matching sub-method and a relationship expansion matching sub-method; Input of the current matching query key information into a pre-constructed building data dictionary knowledge graph and building physics vector database to generate a current information fusion candidate answer; Verification of the current information fusion candidate answer through a preset large language model answer matching degree verification method, and if the verification is passed, a target fuzzy query answer is obtained.

2. The method of claim 1, wherein, The comparison of the current user query problem analysis result through the preset multi-type hierarchical fuzzy matching method to generate the current matching query key information comprises: Comparison of the current user query problem analysis result through the keyword accurate matching sub-method to determine whether there is target keyword accurate matching information; The multi-type hierarchical fuzzy matching method is a hierarchical matching method based on a ReAct framework multi-round reasoning; If the target keyword accurate matching information exists, the target keyword accurate matching information is determined as the current matching query key information; If there is no target keyword accurate matching information, the current user query problem analysis result is matched through the synonym expansion matching sub-method; If there is target synonym expansion matching information, the target synonym expansion matching information is determined as the current matching query key information.

3. The method of claim 2, wherein, After the determination of whether there is target synonym expansion matching information, the method further comprises: If there is no target synonym expansion matching information, the current user query problem analysis result is matched through the upper-lower concept expansion matching sub-method; If there is target upper-lower concept expansion matching information, the target upper-lower concept expansion matching information is determined as the current matching query key information.

4. The method of claim 3, wherein, After the determination of whether there is target upper-lower concept expansion matching information, the method further comprises: If there is no target upper-lower concept expansion matching information, the current user query problem analysis result is matched through the relationship expansion matching sub-method; If there is target relationship expansion matching information, the target relationship expansion matching information is determined as the current matching query key information.

5. The method of claim 4, wherein, The input of the current matching query key information into the pre-constructed building data dictionary knowledge graph to locate at least one target related node and target neighborhood expansion information corresponding to each target related node to generate the current information fusion candidate answer comprises: Input of the current matching query key information into the pre-constructed building data dictionary knowledge graph to locate at least one target related node and target neighborhood expansion information corresponding to each target related node to generate the current information fusion candidate answer. According to the current matching query key information, a semantic recall processing operation is performed in a pre-constructed building physical vector database to obtain a target vector recall result; By means of a pre-set agent distillation method, the target neighborhood expansion information result and the target vector recall result are subjected to distillation processing to generate the current information fusion candidate answer.

6. The method of claim 5, wherein, Before the real-time acquisition of the current to-be-queried user question and the analysis of the current to-be-queried user question to obtain the current to-be-queried user question analysis result in the building physical fuzzy query system, the method further comprises: Standard building data description document information is acquired, and a plurality of standard related nodes are generated by means of a pre-set multi-dimensional building data dictionary node extraction method, and each standard related node is used to acquire each standard neighborhood expansion information result in the standard building data description document information; The standard related nodes and the standard neighborhood expansion information results are combined and constructed to generate the building data dictionary knowledge graph; The standard related nodes are different types of standard objects, and the standard objects include at least one of the following: a fan, a water pump, and a cold machine; the standard neighborhood expansion information result includes standard object information points, standard object relationships, and standard object upper and lower concept hierarchies; the standard object information points include at least one of the following: power and flow; and the standard object relationships include at least one of the following: a containing object relationship and a subordinate object relationship.

7. The method of claim 6, wherein, The current information fusion candidate answer is verified by means of the pre-set large language model answer matching degree verification method, and if the verification is passed, a target fuzzy query answer is obtained, comprising: The matching degree between the current to-be-queried user question and the current information fusion candidate answer is calculated by means of the large language model answer matching degree verification method to obtain a current answer matching degree; A pre-set answer matching degree threshold value is acquired, and the current answer matching degree is verified to see whether it meets the requirements of the answer matching degree threshold value, and if it does, the current information fusion candidate answer is determined as the target fuzzy query answer; If it does not, the operation of comparing the current to-be-queried user question analysis result by means of the pre-set multi-type hierarchical fuzzy matching method to generate the current matching query key information is returned to be executed until a target fuzzy query answer is generated or all multi-type hierarchical fuzzy matching methods are traversed. 8.A building physics fuzzy query device based on a large language model, characterized in that, Comprise: A current to-be-queried user question analysis result determination module is configured to acquire a current to-be-queried user question in real time in a building physical fuzzy query system, and analyze the current to-be-queried user question to obtain a current to-be-queried user question analysis result; A current matching query key information generation module is configured to compare the current to-be-queried user question analysis result by means of a pre-set multi-type hierarchical fuzzy matching method to generate a current matching query key information; The multi-type hierarchical fuzzy matching method comprises a keyword exact matching sub-method, a synonym expansion matching sub-method, an upper and lower concept expansion matching sub-method, and a relationship expansion matching sub-method. The current information fusion candidate answer generation module is configured to input the current matching query key information into a pre-constructed building data dictionary knowledge graph and a building physics vector database to generate a current information fusion candidate answer. The target fuzzy query answer determination module is configured to check the current information fusion candidate answer by using a pre-set large language model answer matching degree checking method, and obtain a target fuzzy query answer if the checking is passed.

9. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the building physics fuzzy query method based on the large language model.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer instructions, and the computer instructions are configured to enable the processor to implement the building physics fuzzy query method based on the large language model.

Images