Engineering construction risk knowledge base construction method and device based on large language model

Abstract

The invention belongs to the technical field of engineering construction data processing, and discloses an engineering construction risk knowledge base construction method and device based on a large language model. The construction method of the engineering construction risk knowledge base comprises the steps of obtaining engineering construction risk data; summarizing and obtaining a construction risk grading query manual and an engineering construction risk prevention and control plan template; perfecting default risk assessment results and risk prevention and control measure contents in the engineering construction risk case record data, extracting effective information according to a preset rule, and forming construction risk data with a standard format; and performing block processing and vectorization processing on the construction risk data with the standard format by utilizing the selected text vector retrieval model to form an engineering construction risk knowledge base. According to the technical scheme, the understanding and processing capability of a large model in the engineering construction field on the engineering construction risk text content can be improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of engineering construction data processing, and in particular relates to a method and device for constructing an engineering construction risk knowledge base based on a large language model. Background Art

[0002] With the development of artificial intelligence technology, it has become a development trend to use large language models to complete various natural language processing tasks in traditional industries, and there are already quite a few successful application cases. Explanatory, the application implementation process of using large language models to complete various natural language processing tasks in traditional industries generally includes: first establish a professional domain corpus, and then introduce professional knowledge into the pre-trained language model based on the retrieval-augmented generation (RAG) technology, so that the language model performs better on natural language processing related tasks in a specific field, and completes the construction of a large domain model. Further explanatory, the implementation steps of establishing a large domain model based on the retrieval-augmented generation technology in the existing technical solution generally include: based on a high-quality professional domain corpus, various types of text data are divided into blocks and vectorized and stored as a domain knowledge base; the large language model vectorizes the received question text and transmits the question text vector to the text retrieval model; the text retrieval model searches in the domain knowledge base based on the question text vector, and returns several knowledge fragments that are most similar to the question text; the large language model combines the retrieved knowledge fragments with the question text for analysis and answering. In the above-mentioned existing technical solutions, when the amount of domain corpus data is huge and the data quality is guaranteed to a certain extent, the large domain model established based on retrieval enhancement generation technology can usually achieve better results.

[0003] Due to the lack of professional knowledge, general large language models are weak in their ability to complete highly professional natural language processing tasks. At present, there are few applications of domain-specific large models in the field of engineering construction. The main reason is the lack of high-quality knowledge bases related to engineering construction and their establishment methods. This is mainly due to the complex data types of engineering construction, uneven data quality, and high degree of text specialization. Explanatory, the data of existing engineering domain knowledge bases are all derived from texts such as specifications, standards, and instruction manuals in the field. There is a lack of actual engineering construction data such as engineering records and construction daily reports. There is a lack of standard and efficient processing methods for different types of construction risk data. To further explain, the texts of specifications and standards are highly professional and concise, which are quite different from the language style of the content of construction daily reports in engineering practice. The knowledge base directly established on this basis and the domain big model further established have weak understanding and processing capabilities for the content of actual engineering text data and cannot meet the needs of engineering data processing; in addition, engineering records, construction daily reports, etc. are mostly data containing structured information. The common engineering construction domain knowledge base construction method is only applicable to text data such as specifications and standards, and cannot make good use of the characteristics of structured data to improve the quality of the knowledge base, resulting in the established domain big model having weak understanding and processing capabilities for the actual engineering structured data; in addition, the information provided by general engineering construction data is relatively complex, and there is currently a lack of a unified method for extracting key construction risk information, resulting in the low quality of the knowledge base built on this basis (for example, general engineering construction data contains various aspects of information on specific projects, not only there are many contents irrelevant to construction risks, but also some key construction risk feature contents may be omitted. Directly using them as knowledge base contents may reduce the domain big model's ability to understand and process the text content of engineering construction risks). Summary of the invention

[0004] The purpose of the present invention is to provide a method and device for constructing a construction risk knowledge base based on a large language model to solve one or more of the above-mentioned technical problems. The construction risk knowledge base construction scheme disclosed in the present invention overcomes the problem caused by the lack of actual construction data, realizes the construction of a high-quality risk knowledge base related to construction, and helps to improve the understanding and processing capabilities of the large model in the construction field for the text content of construction risk.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions: In a first aspect, the present invention provides a method for constructing a construction risk knowledge base based on a large language model, comprising the following steps: Acquire engineering construction risk data; wherein the engineering construction risk data includes construction specifications, standards, instruction manuals and engineering construction risk case record data of preset engineering types; Based on the obtained engineering construction risk data, a construction risk classification query manual and an engineering construction risk prevention and control plan template are summarized and obtained; Based on the obtained construction risk classification query manual and the engineering construction risk prevention and control plan template, the default risk assessment results and risk prevention and control measures in the engineering construction risk case record data are improved, and then effective information is extracted according to predetermined rules to form construction risk data in a standardized format; The construction risk data in the standard format is processed into blocks and vectorized using the selected text vector retrieval model to form a construction risk knowledge base.

[0006] A further improvement of the present invention is that the step of obtaining engineering construction risk data includes: Acquire original engineering construction risk data and perform preprocessing to obtain the engineering construction risk data; Among them, the preprocessing includes: converting text and table data into editable text format; screening and removing information irrelevant to construction in the data while ensuring the semantic integrity of the text; correcting errors in the engineering construction risk case record data based on the selected engineering construction-related regulations, laws and expert evaluation results.

[0007] A further improvement of the present invention is that the step of summarizing and obtaining a construction risk classification query manual and a construction risk prevention and control plan template based on the acquired engineering construction risk data comprises: Based on the data of construction risk industry standards and enterprise internal rules and regulations in construction specifications, standards and guidance manuals, we searched and obtained the risk sources with a probability of appearing in different construction projects and the degree of influence of each risk source on the construction risk level, and summarized it into a construction risk classification query manual; Based on the data on industry standards for engineering construction risks in construction specifications, standards, and instruction manuals, the company's internal rules and regulations, safety specification types, and data on safety accident report types in engineering construction risk case records, a safety regulations library is established to summarize and obtain prevention and control measures for different risk sources and different types of construction risks, and a template for engineering construction risk prevention and control plans is compiled.

[0008] A further improvement of the present invention is that the step of extracting effective information according to a predetermined rule to form construction risk data in a standardized format includes: Extract the project name, project scale, construction geological and meteorological conditions, current construction progress, current project status, construction work plan, current risk level, risk judgment reasons, and risk prevention and control recommendation fields, combine each field name and field content to form a complete text, and obtain construction risk data in a standardized format.

[0009] A further improvement of the present invention is that in the step of extracting effective information according to predetermined rules to form construction risk data in a standardized format, Separate the field name and field content with a colon, separate each field with the special character #, and separate each construction risk case with a line break.

[0010] A further improvement of the present invention is that in the step of using the selected text vector retrieval model to perform block processing and vectorization processing on the construction risk data in the standard format to form a construction risk knowledge base, Store various types of standardized construction risk data in text file format, and divide the construction risk data into blocks using line breaks as separators, so that each construction risk case is stored in a separate text block; Each text block is vectorized and stored using the selected text vector retrieval model to form a construction risk knowledge base.

[0011] A further improvement of the present invention is that after forming the engineering construction risk knowledge base, it also includes: The engineering construction risk knowledge base is mounted on the selected pre-trained large language model to form a large model dedicated to engineering construction.

[0012] In a second aspect, the present invention provides a system for building a construction risk knowledge base based on a large language model, comprising: A data acquisition module is used to acquire engineering construction risk data; wherein the engineering construction risk data includes construction specifications, standards, instruction manuals of preset engineering types, and engineering construction risk case record data; A data summary module, for summarizing and obtaining a construction risk classification query manual and a construction risk prevention and control plan template based on the acquired engineering construction risk data; The data improvement and format specification module is used to improve the default risk assessment results and risk prevention and control measures in the engineering construction risk case record data based on the obtained construction risk classification query manual and the engineering construction risk prevention and control plan template, and then extract effective information according to predetermined rules to form construction risk data in a standardized format; The block segmentation and vectorization processing module is used to utilize the selected text vector retrieval model to perform block segmentation and vectorization processing on the construction risk data in the standard format to form a construction risk knowledge base.

[0013] In a third aspect of the present invention, there is provided an electronic device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for constructing an engineering construction risk knowledge base based on a large language model as described in any one of the first aspect of the present invention is implemented.

[0014] In a fourth aspect of the present invention, a non-transitory computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the method for constructing an engineering construction risk knowledge base based on a large language model as described in any one of the first aspect of the present invention is implemented.

[0015] Compared with the prior art, the present invention has the following beneficial effects: The present invention discloses a method for constructing a construction risk knowledge base based on a large language model. The method targets data of the construction industry standards and internal enterprise rules and regulations, searches and retrieves the possible risk sources in different construction projects and the influence of each risk source on the construction risk level, and summarizes the construction risk classification query manual; targets data of the construction risk industry standards, safety specifications, internal enterprise rules and regulations, and safety accident report types, establishes a safety procedure library, summarizes the prevention and control measures of different risk sources and different construction risk types, and writes a construction risk prevention and control plan template; compiles and improves the risk assessment results and risk prevention and control measures of each construction case information according to the above documents, avoids information default, and extracts effective information from the construction risk case according to certain rules to form a complete text with a standardized format, making full use of the characteristics of various types of data, and being able to form a high-quality construction risk corpus. In summary, in the technical solution disclosed by the present invention, a processing method for structured data such as engineering records and construction daily reports is proposed, which overcomes the problem caused by the lack of actual construction data, realizes the construction of a high-quality risk knowledge base related to construction, and helps to improve the understanding and processing ability of the large model in the field of construction for the text content of construction risk.

[0016] In the preferred technical solution of the present invention, a strategy for extracting and improving effective information of construction risk case data is proposed to ensure the establishment of a high-quality construction risk knowledge base; among them, the key risk feature fields such as project progress, operation risk, personnel management, resource guarantee, meteorological environment, project status, construction plan, construction risk level and risk prevention and control measures of each construction case are extracted, and the names of each field and the content of the field are combined in a certain standard format to form a complete text and input into the knowledge base. In addition, for construction cases with some missing information, the default risk level fields in some construction cases are supplemented and improved based on the engineering risk classification query manual and expert analysis; the risk prevention and control plan for each construction case is supplemented and improved with reference to the safety regulations library and the engineering construction risk prevention and control plan template. The principle that the preferred solution of the present invention can further improve the quality of the construction risk knowledge base is that, from the perspective of the principle of the retrieval enhancement generation technology of the large language model, the daily construction records and daily data are processed in a certain format, which significantly improves the text retrieval accuracy, that is, improves the ability of the model to find the risk analysis and processing plan for the construction project that is most similar to the user-input construction project in the knowledge base. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below; obviously, the drawings described below are some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1 It is a flow chart of a method for constructing an engineering construction risk knowledge base based on a large language model in an embodiment of the present invention.

[0019] Figure 2 It is a schematic diagram of a system for building an engineering construction risk knowledge base based on a large language model in an embodiment of the present invention. DETAILED DESCRIPTION

[0020] In order to make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention; it is obvious that the described embodiments and technical solutions are only part of the embodiments of the present invention, not all of the embodiments.

[0021] All other embodiments obtained by those of ordinary skill in the art without creative work based on the technical solutions disclosed in the embodiments of the present invention belong to the scope of protection of the present invention. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device including a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0022] See also Figure 1 , a method for constructing a construction risk knowledge base based on a large language model provided by an embodiment of the present invention comprises the following steps: Step 1, obtaining engineering construction risk data; wherein the engineering construction risk data includes construction specifications, standards, instruction manuals and engineering construction risk case record data of preset engineering types; in a specific exemplary technical solution, the construction specifications, standards and instruction manuals of the required engineering types can be obtained by online browsing or official website downloading, and the engineering construction risk case record data can be obtained by enterprise and department database query; Step 2: Based on the engineering construction risk data obtained in step 1, summarize and obtain a construction risk classification query manual and an engineering construction risk prevention and control plan template; Step 3: Based on the construction risk classification query manual and the engineering construction risk prevention and control plan template obtained in step 2, the default risk assessment results and risk prevention and control measures in the engineering construction risk case record data are improved, and then effective information is extracted according to predetermined rules to form construction risk data in a standardized format; Step 4: Use the selected text vector retrieval model to perform block processing and vectorization on the construction risk data in the standard format to form a construction risk knowledge base.

[0023] The method for constructing a construction risk knowledge base disclosed in the embodiment of the present invention overcomes the problems caused by the lack of actual construction data, realizes the construction of a high-quality risk knowledge base related to construction, and helps to improve the understanding and processing capabilities of the large model in the field of construction for the text content of construction risks. Specifically and explanatoryally, the method of the present invention obtains construction risk data from multiple sources (such as online inquiries, official website downloads, enterprise and department databases), thereby ensuring the comprehensiveness and diversity of the data. This not only includes theoretical materials such as construction specifications, standards and instruction manuals, but also covers actual construction risk case record data, making the content of the knowledge base richer and more practical; the present invention summarizes and obtains construction risk classification query manuals and construction risk prevention and control plan templates, providing a systematic framework for the assessment and management of construction risks, which helps to unify the standards and processes for risk assessment and improve the efficiency and accuracy of risk management; in the process of perfecting the construction risk case record data and extracting effective information, the method emphasizes the format The importance of standardization ensures the uniformity and comparability of construction risk data, which is convenient for subsequent data analysis and application; the present invention uses the selected text vector retrieval model for vectorization processing, which is a significant improvement. This can not only convert unstructured text data into structured vector data, which is convenient for computer processing and retrieval, but also improve the intelligence level of the knowledge base. Through vectorization processing, relevant risk information and prevention and control measures can be located more quickly, providing timely decision support for engineering construction; the method of the present invention has good scalability. With the continuous accumulation and updating of engineering construction risk data, new data can be continuously incorporated into the knowledge base, and updated and optimized through vectorization processing, which enables the knowledge base to continue to maintain its timeliness and accuracy.

[0024] In a preferred embodiment of the present invention, the method for constructing a project construction risk knowledge base based on a large language model further comprises the following steps after step 4: Step 5: Select a suitable large language model, mount the engineering construction risk knowledge base built in step 4, form a domain large model, and test the model's construction risk analysis capability.

[0025] In the preferred scheme of the embodiment of the present invention, by selecting a suitable large language model and mounting the constructed knowledge base, a specific large model for the field of engineering construction risks can be formed. This field large model can better understand and process data and information related to engineering construction risks, and improve the accuracy and pertinence of the analysis. The mounted constructed knowledge base can enable the large language model to make full use of the engineering construction risk data, risk assessment results, risk prevention and control measures and other contents therein, which helps the model to retrieve and extract relevant information faster in practical applications, and provide strong support for engineering construction risk analysis. Through the field large model, a more intelligent construction risk analysis can be achieved. The model can automatically infer and judge potential construction risks based on the input data and context, and provide corresponding prevention and control measures. This intelligent analysis method can greatly improve the efficiency and accuracy of risk management.

[0026] Further explaining, the present invention combines the large language model with the constructed knowledge base to form a domain large model, which is an important embodiment of technological integration and innovation. This integration not only gives full play to the advantages of each, but also achieves overall performance improvement through interaction and collaboration. By forming a domain large model, the application scenarios of the large language model can be expanded from general fields to the field of engineering construction risks, which provides more accurate and efficient risk management tools for enterprises and institutions in related fields, and helps to promote the intelligent development of the industry. The domain large model can provide more accurate, timely and comprehensive decision-making support for engineering construction risk analysis. Through the analysis and prediction of the model, enterprises can formulate risk prevention and control measures and emergency plans more scientifically to reduce the probability of construction risks and the degree of loss.

[0027] In the embodiment of the present invention, the steps of testing the performance of the construction risk analysis of the domain large model specifically include: Two knowledge bases were established respectively. The first knowledge base only input the text content of the construction risk classification query manual and the construction risk prevention and control plan template. In addition to the above text content, the second knowledge base also inputs data such as processed engineering construction case records. Typical engineering construction risk records were used as test data to compare the construction risk analysis capabilities of the large language model when different knowledge bases were mounted.

[0028] The embodiment of the present invention can significantly improve the construction risk analysis capability of the domain big model by establishing a high-quality engineering construction risk knowledge base. The above implementation method establishes an actual knowledge base, and the results of the domain big model calculation test also support the above conclusion. Specifically, based on the method proposed in the embodiment of the present invention, a total of 2,500 engineering case information were collected and improved, and their risk levels were divided into three levels: low, medium, and high. Among them, the number of low, medium, and high risk cases were 1,934, 431, and 135, respectively. In order to make the comparison results more intuitive, the accuracy of the domain big model's judgment on the engineering construction risk level is used as an indicator to evaluate the construction risk analysis capability of the model.

[0029] In the specific test phase, the large language model is input with detailed information about the project construction, including the project name, project scale, construction geological and meteorological conditions, current construction progress, current project status, construction work plan, etc. The large model is required to output the risk level of the project construction and the reason for the risk judgment. When the result is consistent with the actual risk level and reason for the judgment of the project construction, the output result of the large model is considered accurate. A total of 644 project construction case data were prepared for the test, of which 494, 110 and 39 were low-, medium- and high-risk cases, respectively. All test data were input one by one, and the calculation formula for the accuracy of the output result is as follows: ; In the formula, is the accuracy of the model output results, is the number of correctly classified samples, is the total sample size.

[0030] The final test results show that when the local knowledge base mounted on the large language model only contains the text content of the construction risk classification query manual, the model's construction risk classification rate on the test data is 46.66%. When the local knowledge base mounted on the large language model contains both the construction risk classification query manual and the collected 2,500 engineering case information, the model's construction risk classification rate on the test data is 63.61%, an increase of 16.95%. The above results show that the knowledge base constructed using the technical solution of the embodiment of the present invention has significantly improved the construction risk classification ability of the large model in the field of engineering construction.

[0031] In one embodiment of the present invention, in the step of obtaining engineering construction risk data, it is necessary to collect original engineering construction risk data and perform preprocessing; wherein, the preprocessing step includes: converting data in the form of text, tables, etc. of engineering construction into an editable text format; under the premise of ensuring the semantic integrity of the text, filtering out information in the engineering data that is not related to construction; and correcting errors in the engineering construction case data based on the norms, laws and expert evaluation results related to engineering construction.

[0032] In the preferred technical solution of the embodiment of the present invention, the correctness of the data can be ensured through the above-mentioned preprocessing steps, laying a good foundation for subsequent steps.

[0033] In one embodiment of the present invention, the step of inputting construction risk data into a text vector retrieval model to construct a knowledge base specifically includes: Store various types of construction risk data in text file format, use line breaks as separators, and divide the construction risk data into blocks, so that each construction risk case is stored in a separate text block; In a specific exemplary technical solution, when the collected engineering case data are mostly Chinese texts, bge-large-zh-v1.5 can be selected as the text vector retrieval model to vectorize each text block and store it in the local knowledge base.

[0034] In one embodiment of the present invention, the step of selecting a suitable large language model specifically includes: Select a suitable base language model based on data characteristics, data volume, and hardware performance; In the specific exemplary technical solution, when the collected engineering case data are mostly Chinese texts and the data volume is small, the open source Qwen2-instruct-7B can be selected as the base large language model. This model has a small number of parameters and good reasoning ability, and has low hardware requirements.

[0035] In a specific embodiment of the present invention, taking the field of electric power engineering as an example, a knowledge base of electric power engineering construction risks is established by using the method disclosed in the above embodiment of the present invention, including the following steps: Step S1, collecting relevant standards, specifications and internal engineering construction case data of power engineering, and cleaning and preprocessing the data; Step S2, referring to the standards and specifications for power engineering construction, summarizing the power engineering construction risk classification query manual, referring to typical safety accident reports and expert analysis, and compiling a template for engineering construction risk prevention and control plan; Step S3, for the power engineering construction risk case data, refer to the power engineering risk classification query manual and the power engineering construction risk plan template, improve the risk level and prevention and control plan of each construction case, extract the effective information in the construction case data, and form a complete editable text; for example, respectively extract the project name, project scale, construction geological and meteorological conditions, current construction progress, current project status, construction work plan, current risk level, risk judgment reason, and risk prevention and control suggestion fields in the construction project record, combine the name of each field with the field content, and form a complete text; wherein, the following rules can be set to separate the field name and the field content with a colon, separate each field with a special character #, and separate each construction risk case with a line break; Step S4, input all construction risk data into the text vector retrieval model, divide the text paragraphs into blocks according to line breaks, and vectorize them to establish a power engineering construction risk knowledge base.

[0036] In the technical solution of the embodiment of the present invention, a processing method for different types of construction risk data is proposed to ensure the efficient establishment of a construction risk corpus; among them, for data of engineering construction industry standards and internal enterprise rules and regulations, the engineering risk classification query manual is consulted and summarized; for data of engineering construction risk industry standards, safety specifications, internal enterprise rules and regulations, and safety accident report types, a safety regulations library is established, and a template for engineering construction risk prevention and control plan is summarized and compiled; for data of construction risk engineering records, the risk level and prevention and control measures are improved to form a complete case. In addition, a strategy for extracting and improving effective information of construction risk case data is proposed to ensure the establishment of a high-quality construction risk knowledge base. Taking the engineering risk classification query manual as the standard, the key risk characteristics of each construction case, such as engineering progress, operation risk, personnel management, resource guarantee, meteorological environment, etc., are extracted to assess the risk level of the construction case; the risk prevention and control plan for each construction case is improved by referring to the safety regulations library and the engineering construction risk prevention and control plan template.

[0037] Explanatory, most open source text vector retrieval models can achieve text segmentation and vectorization according to certain rules, such as segmenting text paragraphs according to special symbols such as periods and line breaks, or segmenting text paragraphs according to certain length limits to build a high-quality knowledge base. The present invention proposes a processing method for various types of engineering construction risk data, and proposes key risk characteristics of engineering construction. When processing engineering construction data, a large amount of text content that is irrelevant or less related to construction risks can be first screened out to improve the efficiency of construction risk data processing. The present invention proposes a strategy for improving construction risk case data, and proposes a method for establishing a high-quality construction risk knowledge base to support the establishment of a large model in the field of construction risk analysis. Since relevant domain knowledge can be obtained from a high-quality knowledge base, the performance of the domain large model in tasks such as construction risk classification and construction risk prevention and control plan generation can be significantly improved compared to the general pre-trained large language model.

[0038] To further explain, the knowledge text fragments in the knowledge base referenced by the large language model are actually a text vector retrieval model that compares the vectorized data of the user input text with the vectorized data of the text in the knowledge base, and selects the most similar knowledge fragments. Most construction records and daily data do not strictly introduce construction projects in the above field order, and the names of each field are not labeled. There is also a lot of information that is not related to construction risks. If directly input into the knowledge base, when the user inputs structured data, there are fewer similar fragments in the knowledge base, and the difficulty of model retrieval increases, that is, the model has difficulty understanding the content of each part of the data. The technical solution of the embodiment of the present invention performs structured preprocessing on the data input into the knowledge base, which can fundamentally alleviate the above problems. Experimental calculations also show that the method of the present invention can significantly improve the quality of the construction risk knowledge base. Based on the method of the present invention, more than 10,000 engineering records, construction cases, construction daily reports and other data were collected and preprocessed according to the aforementioned method to obtain complete text format data. Combined with data from dozens of standards, specifications, instruction manuals and other data in the field, text segmentation and vectorization were performed to establish a high-quality engineering construction risk knowledge base. The model calculation experiment results show that using some construction risk analysis cases as test data and the engineering construction risk knowledge base, the large language model has a 10% to 30% improvement in the accuracy of risk grading results for daily construction projects compared to when no knowledge base is used.

[0039] The following are device embodiments of the present invention, which can be used to implement the method embodiments of the present invention. For details not disclosed in the device embodiments, please refer to the method embodiments of the present invention.

[0040] See also Figure 2 In an embodiment of the present invention, a system for building a construction risk knowledge base based on a large language model is provided, comprising: A data acquisition module is used to acquire engineering construction risk data; wherein the engineering construction risk data includes construction specifications, standards, instruction manuals of preset engineering types, and engineering construction risk case record data; A data summary module, for summarizing and obtaining a construction risk classification query manual and a construction risk prevention and control plan template based on the acquired engineering construction risk data; The data improvement and format specification module is used to improve the default risk assessment results and risk prevention and control measures in the engineering construction risk case record data based on the obtained construction risk classification query manual and the engineering construction risk prevention and control plan template, and then extract effective information according to predetermined rules to form construction risk data in a standardized format; The block segmentation and vectorization processing module is used to utilize the selected text vector retrieval model to perform block segmentation and vectorization processing on the construction risk data in the standard format to form a construction risk knowledge base.

[0041] In one embodiment of the present invention, a computer device is provided, the computer device includes a processor and a memory, the memory is used to store a computer program, the computer program includes program instructions, and the processor is used to execute the program instructions stored in the computer storage medium. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is suitable for implementing one or more instructions, and is specifically suitable for loading and executing one or more instructions in the computer storage medium to implement the corresponding method flow or corresponding function; the processor described in the embodiment of the present invention can be used to perform the operation of the method for constructing a construction risk knowledge base based on a large language model.

[0042] In one embodiment of the present invention, a storage medium is provided, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device for storing programs and data. It is understandable that the computer-readable storage medium here can include both built-in storage media in the computer device and, of course, extended storage media supported by the computer device. The computer-readable storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by the processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here can be a high-speed RAM (Random Access Memory) memory, or a non-volatile memory (non-volatile memory), such as at least one disk memory. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of the method for constructing a construction risk knowledge base based on a large language model in the above embodiment.

[0043] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, optical storage, etc.) containing computer-usable program codes.

[0044] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0045] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0046] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0047] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for constructing a construction risk knowledge base based on a large language model, characterized in that: The following steps are involved: Acquire engineering construction risk data; wherein the engineering construction risk data includes construction specifications, standards, instruction manuals and engineering construction risk case record data of preset engineering types; Based on the obtained engineering construction risk data, a construction risk classification query manual and an engineering construction risk prevention and control plan template are summarized and obtained; Based on the obtained construction risk classification query manual and the engineering construction risk prevention and control plan template, the default risk assessment results and risk prevention and control measures in the engineering construction risk case record data are improved, and then effective information is extracted according to predetermined rules to form construction risk data in a standardized format; The construction risk data in the standard format is processed into blocks and vectorized using the selected text vector retrieval model to form a construction risk knowledge base.

2. The method for constructing a construction risk knowledge base based on a large language model according to claim 1, characterized in that: The steps of obtaining engineering construction risk data include: Acquire original engineering construction risk data and perform preprocessing to obtain the engineering construction risk data; Among them, the preprocessing includes: converting text and table data into editable text format; screening and removing information irrelevant to construction in the data while ensuring the semantic integrity of the text; correcting errors in the engineering construction risk case record data based on the selected engineering construction-related regulations, laws and expert evaluation results.

3. The method for constructing a construction risk knowledge base based on a large language model according to claim 1, characterized in that: The step of summarizing and obtaining a construction risk classification query manual and a construction risk prevention and control plan template based on the obtained engineering construction risk data includes: Based on the data of construction risk industry standards and enterprise internal rules and regulations in construction specifications, standards and guidance manuals, we searched and obtained the risk sources with a probability of appearing in different construction projects and the degree of influence of each risk source on the construction risk level, and summarized it into a construction risk classification query manual; Based on the data on industry standards for engineering construction risks in construction specifications, standards, and instruction manuals, the company's internal rules and regulations, safety specification types, and data on safety accident report types in engineering construction risk case records, a safety regulations library is established to summarize and obtain prevention and control measures for different risk sources and different types of construction risks, and a template for engineering construction risk prevention and control plans is compiled.

4. The method for constructing a construction risk knowledge base based on a large language model according to claim 1, characterized in that: The step of extracting effective information according to predetermined rules to form construction risk data in a standardized format includes: Extract the project name, project scale, construction geological and meteorological conditions, current construction progress, current project status, construction work plan, current risk level, risk judgment reasons, and risk prevention and control recommendation fields, combine each field name and field content to form a complete text, and obtain construction risk data in a standardized format.

5. The method for constructing a construction risk knowledge base based on a large language model according to claim 4 is characterized in that: In the step of extracting effective information according to predetermined rules to form construction risk data in a standardized format, Separate the field name and field content with a colon, separate each field with the special character #, and separate each construction risk case with a line break.

6. The method for constructing a construction risk knowledge base based on a large language model according to claim 1, characterized in that: In the step of using the selected text vector retrieval model to process the construction risk data in a standardized format into blocks and vectors to form a construction risk knowledge base, Store various types of standardized construction risk data in text file format, and divide the construction risk data into blocks using line breaks as separators, so that each construction risk case is stored in a separate text block; Each text block is vectorized and stored using the selected text vector retrieval model to form a construction risk knowledge base.

7. The method for constructing a construction risk knowledge base based on a large language model according to claim 1, characterized in that: After the construction risk knowledge base is formed, it also includes: The engineering construction risk knowledge base is mounted on the selected pre-trained large language model to form a large model dedicated to engineering construction.

8. A system for building a construction risk knowledge base based on a large language model, characterized in that: include: A data acquisition module is used to acquire engineering construction risk data; wherein the engineering construction risk data includes construction specifications, standards, instruction manuals of preset engineering types, and engineering construction risk case record data; A data summary module, for summarizing and obtaining a construction risk classification query manual and a construction risk prevention and control plan template based on the acquired engineering construction risk data; The data improvement and format specification module is used to improve the default risk assessment results and risk prevention and control measures in the engineering construction risk case record data based on the obtained construction risk classification query manual and the engineering construction risk prevention and control plan template, and then extract effective information according to predetermined rules to form construction risk data in a standardized format; The block segmentation and vectorization processing module is used to utilize the selected text vector retrieval model to perform block segmentation and vectorization processing on the construction risk data in the standard format to form a construction risk knowledge base.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the method for constructing an engineering construction risk knowledge base based on a large language model as described in any one of claims 1 to 7 is implemented.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for constructing an engineering construction risk knowledge base based on a large language model as described in any one of claims 1 to 7 is implemented.

Images