Chemical pipeline intelligent calculation method and system based on multi-modal large model

By constructing a chemical pipeline topology using a multimodal large model and combining it with an engineering rule base for consistency verification, the problems of labeling and graphic association errors and recognition failures in the parsing of single-line diagrams of chemical pipelines were solved, achieving highly accurate and reliable quantity calculation results.

CN122290164APending Publication Date: 2026-06-26THE SIXTH CONSTR CO LTD OF CHINA NAT CHEM ENG
View PDF 1 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
THE SIXTH CONSTR CO LTD OF CHINA NAT CHEM ENG
Filing Date
2026-04-02
Publication Date
2026-06-26

AI Technical Summary

Technical Problem

Existing technologies for analyzing single-line diagrams of chemical pipelines suffer from problems such as incorrect association between text annotations and graphic elements, failure to recognize non-standard characters, and lack of engineering logic verification, leading to unreliable quantity calculation results.

Method used

A multimodal large model is used to construct the pipeline topology, and text, visual and structural modal information are combined to form a graphic-text association. An embedded engineering rule base is used for consistency verification to form an intelligent quantity calculation closed loop.

Benefits of technology

It improves the accuracy of annotation judgment and the robustness of parsing, realizes fault-tolerant identification and logical auditing of engineering semantics, and significantly improves the reliability and usability of the calculation results.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122290164A_ABST
    Figure CN122290164A_ABST
Patent Text Reader

Abstract

This application relates to an intelligent quantity calculation method for chemical pipelines based on a multimodal large model, comprising the following steps: parsing chemical pipeline drawings to identify pipeline regions and engineering graphic symbols; constructing a pipeline topology representing physical connectivity based on the identified pipeline regions; and performing graphic-text association and multimodal semantic understanding steps based on the engineering graphic symbols: determining the text annotation corresponding to the engineering graphic symbol based on the pipeline topology and the lead-line connectivity in the chemical pipeline drawings; and recognizing the text annotation to obtain text modality information. By constructing a pipeline topology based on pixel-level connected paths and using real physical connectivity instead of geometric distance as the basis for graphic-text association, this method solves the technical problem of easily misattributing text annotations and graphic symbols in environments with dense pipelines and overlapping annotations, significantly improving the certainty and accuracy of annotation determination and providing a reliable data foundation for subsequent quantity calculation.
Need to check novelty before this filing date? Find Prior Art