IT·SCIENCE

AI cuts aging bridge safety checks from 3 days to 10 minutes with 90% accuracy

by
Koo Bon-hyuk
Published : Oct. 1, 2026 - 11:05:39
    • Copy Completed!

View Korean Original

Korea Institute of Civil Engineering and Building Technology develops automated evaluation system combining sensor data and simulation

A vehicle load test is conducted on Haengju Bridge, a cable-stayed bridge, as part of the AI-based bridge safety evaluation research. [Korea Institute of Civil Engineering and Building Technology]
A vehicle load test is conducted on Haengju Bridge, a cable-stayed bridge, as part of the AI-based bridge safety evaluation research. [Korea Institute of Civil Engineering and Building Technology]

A new AI-powered system can assess the structural safety of aging bridges in about 10 minutes — a process that previously took up to three days.

The Korea Institute of Civil Engineering and Building Technology announced Thursday that its structural research division has developed an "AI-based real-time automated bridge safety evaluation system" using data-simulation fusion, or DSF, technology.

According to the Ministry of Land, Infrastructure and Transport, 9,668 of the country's 40,597 bridges — about 23.8 percent — had been in service for 30 years or more as of the end of 2025. Another 13,784 bridges were between 20 and 30 years old. As the number of aging bridges grows rapidly, the importance of safety inspections and maintenance is increasing.

When damage or abnormal signs are detected on a bridge, traffic must be halted and a structural safety assessment conducted quickly to determine whether the bridge can reopen. Under the existing approach, engineers had to run analyses using commercial software and then manually evaluate structural safety — a process that required considerable time and cost.

To address this, the research team combined real-time sensor data collected from bridges with computer simulation. Sensors installed at key points on a bridge measure deck deflection and forces acting on structural members in real time; the AI system then fuses this data with simulation results to automatically generate a safety assessment that reflects the bridge's actual behavior.

The system also visualizes the bridge's deformation as a curve, allowing engineers to compare expected performance at the time of design with actual movement. If abnormal deformation appears, the system can quickly flag the possibility of cracking or a decline in structural stiffness.

In a pilot test on a domestic suspension bridge, the research team achieved a safety evaluation accuracy of more than 90 percent compared with existing commercial software.

The conventional approach required about 15 million won ($11,100) per year in commercial software licensing fees, plus an additional two to three days for expert structural assessment. The new system delivers safety evaluation results in about 10 minutes with no separate software maintenance costs.

The system can also be linked to facility management systems already operated by central government agencies and local governments. The institute plans to expand its application to Han River bridges managed by the Seoul Metropolitan Government, expressway bridges under the Korea Expressway Corporation, and special-purpose bridges overseen by the Korea Infrastructure Safety Corporation. The institute is also pursuing overseas pilot partnerships, targeting Southeast Asian countries with large numbers of maritime bridges.

The technology has been transferred to A-Tech Solution and The Salt for commercialization.

Park Seon-gyu, president of the institute, said the system provides "foundational technology for evaluating bridge safety more quickly and efficiently," adding that the institute would "do its utmost to ensure the safety and maintenance of national infrastructure."


nbgkoo@heraldcorp.com
This content was produced with the assistance of AI translation services.

MOST READ