Method and System for Predicting Load and Strain of Steel Structures Using Deep Learning-Based Image Processing

Steel Structure Load and Strain Prediction System Using Deep Learning-Based Image Processing

This technology is a method and system that receives and preprocesses images of steel structures taken on-site, extracts image features using a neural network, and predicts the load and strain of the steel structure based on matched load and strain data.

Conventional safety inspections for steel structures are costly and time-consuming due to the need for sensor installation and measurement, making it difficult to quickly identify loads and deformations.

By analyzing captured images with deep learning to predict load and strain, this technology can be applied to steel structure safety inspections to rapidly estimate load and deformation without the need for sensors.

‍

‍

Key Features:

• A preprocessing unit that receives and preprocesses on-site images generated by photographing steel structures in the field
• A feature extraction unit that extracts load and strain-related features from the preprocessed on-site images using a neural network
• A prediction unit that estimates load and strain based on the load and strain values matched to the features of the on-site images within the neural network
• A neural network trained using default shape images of steel structures and location-specific load and strain data as training sets

‍