Search

CN-122021827-A - Automatic structure monitoring data analysis report generation method based on workflow and industry knowledge base

CN122021827ACN 122021827 ACN122021827 ACN 122021827ACN-122021827-A

Abstract

The invention discloses an automatic generation method of a structural monitoring data analysis report based on a workflow and an industry knowledge base, which is based on a task arrangement mechanism of FastGPT workflow, divides the whole report generation process into data acquisition, processing, analysis, visualization and generation modules, and automatically controls the execution sequence through a configurable workflow to realize the automatic operation of a low code or a zero code. Industry knowledge enhanced generation control, and domain knowledge base embedded LLM. The large language model driven natural language generating engine converts the structural analysis result into text content with consistent semantics, professional credibility by using LLM (such as deepseek-reasoner), and the generated content has logic, context consistency and industry term suitability. The extensible module plug-in and scene multiplexing mechanism can be used for enabling a user to add or combine the analysis module and the language generation module according to service requirements, supporting cross-project/industry multiplexing and improving system flexibility and life cycle value.

Inventors

  • PENG LING
  • LIU WENFENG
  • LIU XINYI
  • JIN LIANG

Assignees

  • 江西飞尚科技有限公司

Dates

Publication Date
20260512
Application Date
20260127

Claims (5)

  1. 1. The automatic structure monitoring data analysis report generation method based on the workflow and industry knowledge base is characterized by comprising the following specific steps of: step 1, defining an analysis range by a user, triggering task starting, inputting a measuring point data range to be analyzed by the user through a system front-end interface, wherein the measuring point data range comprises a structure name, monitoring factors and inquiry time period information, and automatically starting a preset workflow flow by the system according to the structure name, the monitoring factors and the inquiry time period information; step 2, extracting inquiry parameters to finish preprocessing, wherein the workflow process firstly extracts the inquiry parameters of the measuring point data in the input of a user, wherein the inquiry parameters comprise a structure name, a monitoring factor name and start-stop time, and the system performs preliminary analysis on the parameters and standardizes the parameters into a data inquiry format; Step 3, checking the integrity of parameters and performing fuzzy matching, wherein the system performs the integrity check on the parameters to ensure that no missing items or semantic ambiguity exists, and for the names of monitoring factors, the system compares the predefined monitoring factors through a fuzzy matching mechanism to improve the fault tolerance and the user experience; Step 4, requesting measurement point data, generating a trend graph display, calling HTTP request nodes by the system, connecting a rear-end monitoring database interface, finishing data acquisition, and immediately drawing the trend graph by the system after the data acquisition, wherein the trend graph is used for visually displaying the evolution condition of the data and is used for subsequent analysis and calling; Step 5, extracting data characteristics and generating analysis codes, wherein a system calls a code running node to execute characteristic extraction operation on the measuring point data, including but not limited to a structure type, a monitoring item name, a unit and the measuring point data, and the system automatically generates Python analysis codes for the data by combining a preset industry knowledge base, wherein the code content comprises an analysis algorithm and an explanatory annotation; Step 6, executing analysis codes and checking results, submitting the automatically generated Python codes to a back-end code execution service by the system through another HTTP request node, checking the code executability by the system after the execution is finished, and automatically triggering code regeneration logic by the system if the running abnormality is found; And 7, calling the large language model to generate an analysis report, and calling the large language model by the system to complete the semantic analysis and text generation process on the basis of effective analysis results.
  2. 2. The method for automatically generating a structural monitoring data analysis report based on a workflow and industry knowledge base according to claim 1, wherein in step S1, the monitoring factors comprise sedimentation, displacement and inclination.
  3. 3. The method for automatically generating a structural monitoring data analysis report based on a workflow and industry knowledge base according to claim 1, wherein in step 4, the system instantly draws a trend chart as a time series chart.
  4. 4. The method for automatically generating the structural monitoring data analysis report based on the workflow and the industry knowledge base according to claim 1, wherein in step 5, the preset industry knowledge base comprises structural engineering specifications and expert cases.
  5. 5. The method for automatically generating a structural monitoring data analysis report based on a workflow and industry knowledge base according to claim 1, wherein in step 7, the text generation logic comprises the following substeps: 1) Inputting the analysis result as a user problem; 2) The related content in the joint knowledge base, namely industry specifications, project schemes, technical data and expert lectures; 3) And generating a data analysis report containing abstract, analysis process, conclusion and suggestion content according to a preset structure.

Description

Automatic structure monitoring data analysis report generation method based on workflow and industry knowledge base Technical Field The invention relates to the technical field of data analysis, in particular to an automatic generation method of a structure monitoring data analysis report based on a workflow and industry knowledge base. Background Currently, the following implementations are mainly used in the industry in terms of automatically generating data analysis reports: (1) Template-based report generation tools some Business Intelligence (BI) platforms (e.g., tableau, power BI, fineBI, etc.) provide chart generation and fixed template report export functions. The user may preset a template and then derive an analysis document containing charts and fixed text by dragging the charts, selecting data fields. This approach focuses on visual presentation, rather than semantic analysis and automatic text composition. (2) Statement stitching methods based on rule engines part of the industry (e.g., finance, manufacturing) uses predefined rules (e.g., "when sales drop more than 10%, generate alert statements") to map data to fixed statement fragments and stitch them into reports. Such methods are typically driven by code rules or business engines, without language flexibility and deep understanding capabilities. (3) The automatic abstract generation scheme of the NLP technology is primarily used, wherein some scientific research or report class applications introduce traditional Natural Language Processing (NLP) technology, such as textRank, LDA, word bag models and the like, and simple abstracts are generated on analysis results. However, these techniques are difficult to understand in context, domain terms, or to make structural expressions, and have limited applicability. (4) Part of AI products integrate LLM but do not have workflow capability, namely, at present, part of AI document products (such as Notion AI, chatGPT plug-ins, tencent mixing elements and the like) try to use a large language model for text generation, but the tools are generally oriented to a general scene, lack of complete task flow from original data to text report, and are more incapable of realizing the arrangement and landing of flow such as structured data access, analysis module call, semantic style control and the like. Despite the availability of existing solutions in specific dimensions, the following major problems remain in the face of data reporting requirements requiring high automation, strong analysis capabilities and flexible language generation: (1) The method is lack of full-flow automation, is time-consuming to manually arrange data, is low in efficiency, and is difficult to deal with the high-frequency analysis requirement. (2) The generated content lacks intelligent understanding and flexibility, the template mode is fixed, the rule splicing mode is dead and can not adapt to dynamic requirements, the generated content lacks semantic understanding capability, and the result is stiff. (3) The generation quality is limited, the manual rule is relied on, the maintenance cost is high, the traditional NLP method can not generate natural language content with logic flow, conclusion support and coherent structure, and industry knowledge optimization can not be combined. (4) The large language model application on the current market lacks the orchestration capability facing the actual business process, and a user cannot define the process control capability of key steps such as data source access logic, analysis logic selection, reporting structure and the like " Disclosure of Invention The invention provides a structure monitoring data analysis report generation method based on combination of a workflow and an industry knowledge base, which realizes automatic execution of analysis tasks by defining an input-processing-output full flow through the workflow, invokes an analysis module to complete generation and execution of statistical analysis codes of data to obtain a calculation result by combining the industry knowledge base, and converts the calculation result into a natural language by utilizing a large language model to form a segmented and specialized analysis conclusion. In order to achieve the above purpose, the present invention adopts the following technical scheme: the automatic structure monitoring data analysis report generation method based on the workflow and industry knowledge base comprises the following specific steps: step 1, defining an analysis range by a user, triggering task starting, inputting a measuring point data range to be analyzed by the user through a system front-end interface, wherein the measuring point data range comprises a structure name, monitoring factors and inquiry time period information, and automatically starting a preset workflow flow by the system according to the structure name, the monitoring factors and the inquiry time period information; step 2, extracting inquiry param