Autoencoder Secures River Tunnels with 90% Anomaly Detection
A Novel Autoencoder for Structural Anomalies Detection in River Tunnel Operation
Tags: Universiti Malaya, Malaysia, Computing Technology, Construction & Smart Cities
This technology introduces an adaptive loss function autoencoder (AdaAE) model for detecting structural anomalies in river tunnels, even from poor-quality data. It enhances anomaly detection accuracy by over 90% through the use of simulated data, including structural damage and disturbances with Gaussian noise. The AdaAE model is particularly effective in diagnosing anomalies caused by environmental pollution, making it suitable for real-world tunnel operations. Applications include intelligent management and disaster prevention in tunnel operations. A case study on the Wuhan Yangtze River tunnel demonstrated the model's effectiveness in identifying structural strain anomalies.
IP Type or Form Factor: Discovery & Research; Software & Algorithm
TRL: 5 - prototype ready for testing in intended environment
Industry or Tech Area: Big Data Analytics & Simulations; Civil Engineering
