Work

Practical Feasibility of AI Image Analysis and Digital Twins for Remote Management of Drainage Pump Stations and Agricultural Canal Gates

Year: 2025

Type: article

Abstract: Gates and drainage pumps at river or coastal interfaces are vital for flood control and protecting residential zones. Many small sluice gates and drainage pump stations, which are managed locally, depend on manual on-site operations, posing challenges due to aging staff and safety risks. We developed a "Sluice Gate Monitoring System" utilizing a CNN-based semantic segmentation model (DeepLabV3+) to monitor water levels and gate openings from surveillance images remotely. Field tests achieved a Mean Absolute Error of 1.7 cm (water levels) and 7.7 cm (gate openings). Additionally, a "Drainage Pump Station Monitoring System" was created, which involved 3D modelling with UAV and LiDAR enables remote monitoring of inundation status and equipment operability. Integrating these digital twins with simulated flood scenarios facilitates the identification of critical water levels, supports rapid disaster recovery, and enhances informed flood control decisions.

Source: e-Journal of Nondestructive Testing

DOI: 10.58286/31723

Authors: Toru Nakada, Hiroto Kichise, Hirohide Kiri

Institutions: National Agriculture and Food Research Organization, Institute for Rural Engineering

Language: English

Topic: Infrastructure Maintenance and Monitoring

Subfield: Civil and Structural Engineering

Field: Engineering

Domain: Physical Sciences

SDGs: Industry, innovation and infrastructure, Clean water and sanitation

Open Access status: diamond

Funder: Council for Science, Technology and Innovation