An intelligent question and answer method for structural mechanics test design combining a knowledge base and a large language model

Abstract

The application discloses a structural mechanics test design intelligent question and answer method combining a knowledge base and a large language model. The structural mechanics test design intelligent question and answer system provided by the application significantly improves the efficiency and intelligent level of structural mechanics test design. Through systematic knowledge integration and intelligent retrieval, the repetitive labor and manual review time of the design personnel are greatly reduced, and design errors and rework caused by insufficient experience or missing information are effectively avoided. The system can quickly respond to user questions and accurately output solutions for specific test scenarios, ensuring the standardization and reliability of the design scheme. At the same time, the visual interactive interface reduces the threshold for use, so that non-professionals can also conveniently obtain professional guidance. Overall, the system shortens the test design cycle, improves the design quality and success rate, and provides strong support for the digitalization and intelligent upgrading of structural mechanics test design.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence application technology, specifically to an intelligent question-answering method for structural mechanics experimental design that combines a knowledge base and a large language model. Background Technology

[0002] Traditional structural mechanics test design is typically completed manually by experienced engineers and experimenters. This method is labor-intensive, inefficient, and prone to errors, often resulting in tooling interference or inability to install tooling connections, leading to rework and delays. Structural mechanics test design involves a wide range of areas, a large amount of information, and intricate details, making it a highly experience-based task.

[0003] Traditional experimental design models are no longer sufficient for current scientific research tasks, revealing problems such as information lag, poor utilization efficiency, low process efficiency, and frequent errors. Therefore, the intelligent upgrading of traditional experimental design methods is an inevitable trend. Currently, my country has limited research on intelligent design methods for structural mechanics experiments, and even more so, lacks a complete intelligent design system for structural mechanics experiments.

[0004] In the process of structural mechanics experimental design, many similar problems will arise repeatedly for the same type of structural mechanics experiment, and similar problems will have similar solutions. If we can collect and organize various experiences and knowledge in the process of structural mechanics experimental design, and apply the mature experience and knowledge from the past to the current problem when encountering similar problems, thereby quickly forming a solution, the effect on improving work efficiency and shortening the test cycle is obvious. At present, the research on knowledge bases for structural mechanics experimental design at home and abroad is still incomplete, and an efficient and comprehensive knowledge base architecture system for structural mechanics experimental design has not yet been established. Summary of the Invention

[0005] To address the technical problems mentioned above, this invention provides an intelligent question-answering method for structural mechanics experimental design that combines a knowledge base and a large language model, comprising the following steps: S1: Collect existing online and offline materials and expert experience to form a knowledge text on structural mechanics test design; S2: Based on the knowledge text of structural mechanics test design, determine the framework of the structural mechanics test design knowledge base based on the knowledge graph, and input the knowledge text of structural mechanics test design to establish the structural mechanics test design knowledge base; S3: Based on the knowledge base of structural mechanics test design, deploy a large language model and design a visual user interface to build an intelligent question-and-answer system for structural mechanics test design.

[0006] Preferably, S1 includes: identifying the factors affecting structural mechanics test design and the basic form of the design document content based on collected online and offline data and expert experience, and identifying, expressing, classifying, merging and organizing the knowledge required for structural mechanics test design to form a complete knowledge text of structural mechanics test design.

[0007] Preferably, in S2, the knowledge base framework for structural mechanics test design includes four parts: similarity theory and structural model design, loading and excitation system in structural tests, response parameter measurement methods in structural tests, and digital processing of test data.

[0008] Preferably, in step S2, when building the similarity theory and structural model design section of the structural mechanics test design knowledge base, the output model geometric dimensions, model plate thickness, model material selection, and model processing technology are obtained based on the input basic principles of similarity theory, basic principles of dimensional analysis, similarity criteria for structural strength problems, similarity criteria for vibration problems, and model design requirements.

[0009] Preferably, in step S2, when building the loading and excitation system part of the structural test in the structural mechanics test design knowledge base, the output loading method, loading position, loading equipment, loading size and loading direction are obtained based on the input electro-hydraulic servo power loading system, excitation system and test platform.

[0010] Preferably, in step S2, when constructing the response parameter measurement method section of the structural test in the structural mechanics test design knowledge base, the output sensor structure and characteristics, sensor selection, sensor calibration, measurement circuit, and sensor arrangement and measurement are obtained based on the input strain measurement, displacement measurement, velocity measurement, acceleration measurement, force measurement and acquisition system.

[0011] Preferably, in step S2, when building the digital processing part of the test data in the structural mechanics test design knowledge base, the output test data processing, data pattern mining, basic data visualization, and data error analysis are obtained based on the input data preprocessing, data conversion and standardization, and data analysis and modeling.

[0012] Preferably, in step S3, based on the structural mechanics test design knowledge base, information related to the question is retrieved from the structural mechanics test design knowledge base using retrieval enhancement generation technology, and the retrieved information is used as context input for the large language model to obtain an answer that meets the user's needs; and based on the answer, a visual user interface is designed, which includes customized dialog boxes and input boxes.

[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: The intelligent question-and-answer system for structural mechanics experimental design provided by this invention, which combines a knowledge base and a large language model, significantly improves the efficiency and intelligence level of structural mechanics experimental design. Through systematic knowledge integration and intelligent retrieval, it greatly reduces repetitive work and manual review time for designers, effectively avoiding design errors and rework caused by insufficient experience or information omissions. The system can quickly respond to user questions and accurately output solutions for specific experimental scenarios, ensuring the standardization and reliability of the design solutions. At the same time, the visual interactive interface lowers the barrier to entry, allowing non-professionals to easily obtain professional guidance. Attached Figure Description

[0014] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments are briefly introduced below. Obviously, the 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.

[0015] Figure 1 This is a flowchart illustrating the method framework of an embodiment of the present invention; Figure 2 This is a diagram illustrating the knowledge base module architecture of an embodiment of the present invention. Figure 3 This is a diagram illustrating the architecture of the large language model module in an embodiment of the present invention. Figure 4 This is a schematic diagram illustrating the similar theoretical and structural model design of an embodiment of the present invention; Figure 5 This is a schematic diagram of the loading and excitation system in the structural test of an embodiment of the present invention; Figure 6 This is a schematic diagram of the response parameter measurement method in the structural test according to an embodiment of the present invention; Figure 7 This is a schematic diagram illustrating the digital processing of test data in an embodiment of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, 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 are within the scope of protection of the present invention.

[0017] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0018] Example 1 This embodiment provides an intelligent question-answering method for structural mechanics experimental design that combines a knowledge base and a large language model. Its flowchart is as follows: Figure 1 As shown, the steps include: S1. Collect existing online and offline materials and expert experience to form a knowledge text on structural mechanics test design.

[0019] This process involves collecting existing online and offline materials, as well as expert experience, to create a knowledge document on structural mechanics experimental design. Data sources in the field of structural mechanics experimental design include online and offline textual materials and expert experience. Offline materials include, but are not limited to, textbooks and lecture notes; online materials include, but are not limited to, equipment product webpages, papers, and patents; and expert experience is obtained through questionnaires or interviews. The study investigates the factors influencing structural mechanics experimental design and the basic formats of design documents, such as model scale, boundary conditions, loading methods, and measurement types. The necessary knowledge for structural mechanics experimental design is identified, effectively expressed, categorized, merged, and organized to form a complete knowledge document.

[0020] S2. Based on the knowledge text of structural mechanics test design, determine the framework of the structural mechanics test design knowledge base based on the knowledge graph, and input the knowledge text of structural mechanics test design to establish the structural mechanics test design knowledge base.

[0021] The framework of the structural mechanics test design knowledge base is determined based on knowledge graphs, and the knowledge text is input to build the knowledge base. Specifically, the knowledge text is input into the Neo4j database for analysis to construct a knowledge graph for intelligent question answering and information retrieval.

[0022] Based on the complex relationships in the data, the knowledge base module framework for structural mechanics experimental design is constructed (its structure is as follows). Figure 2 As shown, it is divided into similarity theory and structural model design (such as...). Figure 4 As shown), the loading and excitation system in structural tests (such as...) Figure 5 (as shown), response parameter measurement methods in structural tests (such as...) Figure 6 (as shown), digital processing of experimental data (such as...) Figure 7 (As shown) These are the four parts. The knowledge base takes as input a large amount of basic knowledge required for structural mechanics test design, and outputs specific knowledge or solutions required by the user for a particular test scenario.

[0023] S3. Based on the knowledge base of structural mechanics test design, deploy a large language model and design a visual user interface to build an intelligent question-and-answer system for structural mechanics test design.

[0024] This section, "Similarity Theory and Structural Model Design," is part of a knowledge base for structural mechanics experimental design. Input data includes: basic principles of similarity theory, basic principles of dimensional analysis, similarity criteria for structural strength problems, similarity criteria for vibration problems, and model design requirements. Output data includes: model geometric dimensions, model plate thickness, model material selection, and model manufacturing process. Basic principles of similarity theory include: positive similarity theorem, π theorem, and inverse similarity theorem. Basic principles of dimensional analysis include: definition of dimensions, transformation of dimensional systems, dimensional equations and their effects. Similarity criteria for structural strength problems include: geometric similarity, load similarity, and displacement similarity. Similarity criteria for vibration problems include: physical condition similarity, boundary condition similarity, and initial condition similarity. Model design requirements include: realistic model design, abnormal model design, model materials, and model manufacturing.

[0025] Furthermore, a section on "Loading and Excitation Systems in Structural Testing" is constructed within the structural mechanics experimental design knowledge base. The input data for this knowledge base includes: electro-hydraulic servo-powered loading systems, excitation systems, and test platforms. The output data includes: loading method, loading location, loading equipment, loading magnitude, and loading direction. The electro-hydraulic servo-powered loading system includes: hydraulic actuators, hydraulic source systems, and control systems. The excitation system includes: vibrator, charge amplifier, and control system. The test platform includes: anchor bolt-type test platform, horizontal reaction wall, and reaction frame.

[0026] Furthermore, a section on "Measurement Methods of Response Parameters in Structural Testing" is constructed within the knowledge base for structural mechanics experimental design. The input data for this knowledge base includes: strain measurement, displacement measurement, velocity measurement, acceleration measurement, force measurement, and data acquisition systems. The output data includes: sensor structure and characteristics, sensor selection, sensor calibration, measurement circuits, and sensor placement and measurement. Strain measurement includes: resistance strain gauges, fiber optic strain gauges, and DIC measurement. Displacement measurement includes: pin-type displacement gauges, laser displacement gauges, and DIC measurement. Velocity measurement includes: photoelectric velocimeters and laser Doppler velocimeters. Acceleration measurement includes: piezoelectric accelerometers and capacitive accelerometers. Force measurement includes: resistance strain gauge force sensors and piezoelectric force sensors. Data acquisition systems include: static data acquisition devices and dynamic data acquisition devices.

[0027] Furthermore, a "Digital Processing of Experimental Data" section is constructed within the structural mechanics experimental design knowledge base. The input materials for this knowledge base include: data preprocessing, data transformation and standardization, and data analysis and modeling. The output materials include: experimental data processing, data pattern mining, basic data visualization, and data error analysis. Data preprocessing includes: noise removal, missing value handling, outlier detection, and data alignment. Data transformation and standardization includes: data standardization / normalization and data feature extraction. Data analysis and modeling includes: statistical data analysis, data modeling and prediction, and data uncertainty assessment.

[0028] Furthermore, a large language model module framework is constructed (structure as follows). Figure 3 As shown, the large language model module is based on RAG (Retrieval Enhanced Generation) technology. It retrieves information related to the question from a pre-built knowledge base, uses the retrieved information as contextual input to the large language model, and combines the large language model to generate an answer that meets the user's needs.

[0029] This embodiment also includes a visual user interface with customized dialog boxes and input fields. Users can enter an opening statement, such as: "Hello, I am a structural mechanics experiment design assistant. You can ask me questions you encounter in structural mechanics experiment design." The interface also summarizes keywords for questions, providing a more natural and fluid interactive experience. Examples of questions are provided, such as: "What is a strain gauge, and how do I use it? What is the maximum loading capacity of a ship and marine structural mechanics laboratory?"

[0030] Example 2 Taking the Ship and Marine Structure Mechanics Laboratory of the School of Naval Architecture and Ocean Engineering at Shanghai Jiao Tong University as an example, this paper presents the establishment and testing of an intelligent question-and-answer system for structural mechanics experimental design.

[0031] Using keywords such as "structural mechanics experiment," "experimental mechanics," "structural testing technology," and "data processing," we collected textbooks, papers, patents, and other materials, and used web crawlers to obtain text resources. We recorded the models and testing capabilities of specific equipment and instruments already in the laboratory, such as reaction frames, hydraulic cylinders, and data acquisition instruments, and recorded the parameters of various sensors, such as strain gauges, displacement gauges, and force sensors. We obtained influencing factors in structural mechanics experimental design through questionnaires or interviews, supplementing this with expert experience. We completed the expression and identification of knowledge, and compiled the above materials into a complete electronic document.

[0032] Using the Neo4j database and its built-in Neo4j Browser visualization tool, entity nodes and relationships are created by writing Cypher statements, and then query statements are executed to load the node and relationship data. The built-in force-directed layout automatically generates a graphical display, ultimately achieving an interactive knowledge graph visualization effect. This visually presents the network of relationships between entities, such as... Figure 2 .

[0033] By utilizing the node and relationship data of the knowledge graph, a knowledge base framework for an intelligent design system for structural mechanics experiments is summarized. The knowledge base module framework is divided into four parts: similarity theory and structural model design, loading and excitation system in structural experiments, response parameter measurement methods in structural experiments, and digital processing of experimental data.

[0034] Based on knowledge graphs, the knowledge base framework is further refined. The "Similarity Theory and Structural Model Design" section is subdivided into: basic principles of similarity theory, basic principles of dimensional analysis, similarity criteria for structural strength problems, similarity criteria for vibration problems, and model design requirements. The "Loading and Excitation Systems in Structural Testing" section is subdivided into: electro-hydraulic servo-driven loading systems, excitation systems, and test platforms. The "Response Parameter Measurement Methods in Structural Testing" section is subdivided into: strain measurement, displacement measurement, velocity measurement, acceleration measurement, force measurement, and acquisition systems. The "Digital Processing of Experimental Data" section is subdivided into: data preprocessing, data conversion and standardization, and data analysis and modeling. After completing the knowledge classification, each subdivided area is formed into an independent knowledge document.

[0035] Based on the MaxKB platform, the deployment of the large language model is set up. According to the server conditions, a suitable AI model is selected. In this embodiment, deepseek-rl:1.5b is selected. The AI ​​parameters are set, the temperature is set to 0.3 (a higher value will make the output more random, while a lower value will make it more concentrated and deterministic), and the maximum number of tokens output is set to 10240.

[0036] Link to the knowledge base, upload knowledge documents to the knowledge base, set parameters, select vector search as the search mode, the similarity must be higher than 0.6, the maximum number of reference segments is 3, and the maximum number of referenced characters is 5000.

[0037] Design a visual user interface that allows users to ask questions and generate answers, and preview and debug it. Customize the dialog boxes and input fields. Enter the opening line: "Hello, I'm the structural mechanics experiment design assistant. You can ask me questions you encounter in structural mechanics experiment design." Provide example questions: "What is a strain gauge, and how do I use it? What is the maximum loading capacity of the ship and ocean structural mechanics laboratory? How can I accurately measure dynamic forces and displacements?" Click on the example question: "What is the maximum loading capacity of the ship and ocean structural mechanics laboratory?"

[0038] Output the answer and display the knowledge source. The output answer is: "A multi-channel static loading system is a device that applies different loads simultaneously through multiple test benches, mainly used to compare the effects of different loads on a structure. The MTS system, on the other hand, is used for dynamic loading, potentially performing continuous or real-time load application tests in a laboratory, suitable for analyzing the dynamic response of a structure. The main characteristics of a multi-channel static loading system include: the ability to simultaneously apply multiple loads, suitable for static testing; and the MTS system, which has dynamic loading capabilities, suitable for analyzing the dynamic characteristics of a structure." The knowledge sources are displayed as: "Test Platform," "Electro-hydraulic Servo Dynamic Loading System," and "Vibration System."

[0039] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for intelligent question answering in structural mechanics experimental design that combines a knowledge base and a large language model, characterized in that, Includes the following steps: S1: Collect existing online and offline materials and expert experience to form a knowledge text on structural mechanics test design; S2: Based on the knowledge text of structural mechanics test design, determine the framework of the structural mechanics test design knowledge base based on the knowledge graph, and input the knowledge text of structural mechanics test design to establish the structural mechanics test design knowledge base; S3: Based on the knowledge base of structural mechanics test design, deploy a large language model and design a visual user interface to build an intelligent question-and-answer system for structural mechanics test design.

2. The intelligent question-answering method for structural mechanics experimental design combining a knowledge base and a large language model according to claim 1, characterized in that, S1 includes: based on collected online and offline data and expert experience, identifying the factors affecting structural mechanics test design and the basic form of the design document content, and identifying, expressing, classifying, merging and organizing the knowledge required for structural mechanics test design to form a complete knowledge text of structural mechanics test design.

3. The intelligent question-answering method for structural mechanics experimental design combining a knowledge base and a large language model as described in claim 1, is characterized in that... In S2, the knowledge base framework for structural mechanics test design includes four parts: similarity theory and structural model design, loading and excitation system in structural tests, response parameter measurement methods in structural tests, and digital processing of test data.

4. The intelligent question-answering method for structural mechanics experimental design combining a knowledge base and a large language model as described in claim 3, is characterized in that... In S2, when building the similarity theory and structural model design section of the structural mechanics test design knowledge base, the output model geometric dimensions, model plate thickness, model material selection, and model processing technology are obtained based on the input basic principles of similarity theory, basic principles of dimensional analysis, similarity criteria for structural strength problems, similarity criteria for vibration problems, and model design requirements.

5. The intelligent question-answering method for structural mechanics experimental design combining a knowledge base and a large language model as described in claim 3, is characterized in that... In S2, when building the loading and excitation system part of the structural test in the structural mechanics test design knowledge base, the loading method, loading position, loading equipment, loading size and loading direction are obtained based on the input electro-hydraulic servo power loading system, excitation system and test platform.

6. The intelligent question-answering method for structural mechanics experimental design combining a knowledge base and a large language model as described in claim 3, is characterized in that... In S2, when building the response parameter measurement method section of the structural test in the structural mechanics test design knowledge base, based on the input strain measurement, displacement measurement, velocity measurement, acceleration measurement, force measurement and acquisition system, the output sensor structure and characteristics, sensor selection, sensor calibration, measurement circuit, and sensor arrangement and measurement are obtained.

7. The intelligent question-answering method for structural mechanics experimental design combining a knowledge base and a large language model as described in claim 3, is characterized in that... In S2, when building the digital processing part of the test data in the structural mechanics test design knowledge base, the output test data processing, data pattern mining, basic data visualization and data error analysis are obtained based on the input data preprocessing, data conversion and standardization, and data analysis and modeling.

8. The intelligent question-answering method for structural mechanics experimental design combining a knowledge base and a large language model as described in claim 1, characterized in that, In step S3, based on the structural mechanics test design knowledge base, information related to the question is retrieved from the structural mechanics test design knowledge base using retrieval enhancement generation technology, and the retrieved information is used as the context input of the large language model to obtain an answer that meets the user's needs. Based on the answers, design a visual user interface, which includes customized dialog boxes and input fields.

Images