|
چکیده
|
Ensuring the safety and longevity of infrastructure such as bridges is crucial and heavily depends on effective structural health monitoring. This study introduces a novel method for detecting anomalies in bridge structures using vibration data processed through scalogram generation and classified with the ResNet-152 model. The proposed approach involves preprocessing raw vibration signals through filtering and normalization to reduce noise and standardize the data. The processed signals are then converted into scalograms, which provide a visual representation of frequency content over time, aiding in anomaly detection. The ResNet-152 model, pretrained on ImageNet, is used to classify the scalograms, ensuring efficient and accurate anomaly detection. Our method was validated using the Z24 and KW51 datasets, achieving an accuracy of 97.97% for the Z24 dataset and 90.17% for the KW51 dataset. These results significantly surpass those obtained with traditional methods, demonstrating the proposed method’s effectiveness for automatic monitoring within Structural Health Monitoring (SHM) systems. This study explores how this method can enhance bridge maintenance and safety by providing accurate and dependable detection of anomalies. The high performance of our model highlights its potential for real-time, efficient, and reliable applications in SHM systems
|