A digital test-based extreme working condition ship welding quality prediction method

By using digital experimentation methods and dynamic adversarial network training, a digital model of ship welding was constructed, which solved the problem of efficient and accurate prediction of welding quality under extreme working conditions, reduced costs and improved prediction accuracy.

CN122134173APending Publication Date: 2026-06-02BEIHANG UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
BEIHANG UNIV
Filing Date
2026-02-05
Publication Date
2026-06-02

AI Technical Summary

Technical Problem

Existing technologies are insufficient for efficiently and accurately predicting the quality of welded joints in ships under extreme conditions. Traditional physical experiments are costly and risky, while data-driven methods suffer from low data quality and insufficient domain adaptability.

Method used

A digital model of ship welding was constructed using a digital experimental approach. The model was trained using a dynamic adversarial network (DANN) and its accuracy and generalization ability were optimized by combining virtual working condition reproduction with physical experimental data verification.

Benefits of technology

It enables efficient and accurate prediction of ship welding quality under extreme working conditions, reduces sample acquisition costs, and provides technical support for quality control in the welding process.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122134173A_ABST
    Figure CN122134173A_ABST
Patent Text Reader

Abstract

This invention discloses a method for predicting the quality of ship welding under extreme conditions based on digital experiments, belonging to the fields of electronic engineering and computer science. It includes: constructing a digital model of ship welding, including: defining quantitative representation methods for three key influencing factors: ship welding process parameters, material properties, and equipment status; processing historical ship welding data including these three key influencing factors; training the DANN (Dynamic Adversarial Network) using transfer learning based on the processed historical data; iterating and validating the trained DANN using relevant features of ship welding and ship welding data under extreme conditions; and verifying the prediction accuracy and generalization ability of the digital model by reproducing virtual conditions and comparing with physical experimental data. This invention can ensure the quality of ship welding, reduce sample acquisition costs, achieve efficient and accurate prediction of weld quality, and provide technical support for quality control of ship welding processes under extreme conditions.
Need to check novelty before this filing date? Find Prior Art