Hybrid AI Framework for Real -Time Monitoring of Corrosion in Steel Bridges

Corrosion poses a significant threat to the structural integrity and service life of steel bridges, demanding advanced monitoring solutions that go beyond traditional inspection techniques. This study proposes a hybrid artificial intelligence (AI) framework that integrates machine learning (ML), deep learning (DL), and signal processing methods for real-time corrosion detection and monitoring. The system combines vibration analysis, acoustic emission, and electrochemical sensor data to capture early-stage corrosion signatures under varying environmental and loading conditions. A two-stage architecture is developed: the first stage uses a convolutional neural network (CNN) to extract spatial features from sensor data, while the second stage applies a long short-term memory (LSTM) network to track temporal evolution and predict corrosion progression. To enhance decision reliability, fuzzy logic is incorporated to fuse multi-sensor outputs and quantify corrosion severity. The framework is trained and validated using a dataset collected from laboratory-scale steel bridge components subjected to accelerated corrosion, showing an accuracy of ۹۶.۴% in identifying active corrosion states and predicting future damage trends. The results demonstrate the system’s potential for scalable, autonomous, and continuous health monitoring of aging steel infrastructure, contributing to safer and more cost-effective maintenance strategies.

کلیدواژه ها:

Corrosion Monitoring ، Steel Bridges ، Hybrid Artificial Intelligence ، Convolutional Neural Networks (CNN) ، Long Short-Term Memory (LSTM) ، Fuzzy Logic ، Real-Time Structural Health Monitoring ، Multi-Sensor Fusion ، Infrastructure Resilience ، Predictive Maintenance