Intelligent question answering method of circuit block diagram structural extraction and knowledge enhancement

By combining YOLOv5 and optical character recognition models to extract circuit component information, and integrating circuit domain knowledge with a multimodal large language model, the problem of intelligent question answering of circuit block diagrams is solved, achieving efficient circuit topology logic understanding and professional knowledge generation.

CN122152880APending Publication Date: 2026-06-05ANHUI UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
ANHUI UNIV
Filing Date
2026-01-15
Publication Date
2026-06-05

AI Technical Summary

Technical Problem

Existing technologies struggle to quickly and accurately extract structured information from circuit block diagrams and perform intelligent question answering, failing to effectively understand circuit topology logic and component functional attributes.

Method used

By combining a YOLOv5-based target detection model and optical character recognition model with an adaptive weight matching algorithm, the geometry, ports, and connection relationships of circuit components are extracted. Furthermore, through deep integration of a multimodal large language model with a circuit domain knowledge base, structured data and professional knowledge fragments are generated.

Benefits of technology

It achieves end-to-end, highly reliable, and intelligent analysis from circuit block diagrams to professional understanding, improving the efficiency of circuit design assistance and the accuracy of knowledge transfer.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122152880A_ABST
    Figure CN122152880A_ABST
Patent Text Reader

Abstract

The application discloses an intelligent question and answer method for circuit block diagram structured extraction and knowledge enhancement, comprising the following steps: obtaining a circuit block diagram image to be analyzed; inputting the circuit block diagram image into a trained circuit component detection model to obtain the geometric shape category and the boundary box coordinates of each circuit component; the geometric shape category comprises element components and connection arrows; performing text region detection and recognition on the circuit components belonging to the element components according to the geometric shape category and the boundary box coordinates to obtain corresponding component name information; and determining the port of each circuit component by using a self-adaptive weight matching port determination algorithm according to the detection result of the connection arrows output by the circuit component detection model to obtain the input and output port quantity and position coordinates of each circuit component. The method realizes end-to-end automatic processing from the input of the circuit block diagram image to the output of the intelligent question and answer, and improves the understanding depth of the circuit principle and the question and answer reliability.
Need to check novelty before this filing date? Find Prior Art