A novel approach model design for signal data using 1DCNN combing with LSTM and ResNet for damaged detection problem

In the current era, time-series data is widely used in different domains. In Structural Health Monitoring (SHM), the utilization of time-series data is also extensive. However, traditional Machine Learning (ML) methods, such as Artificial Neural Networks, applied in SHM, are no longer effective enough to process and accurately diagnose structural damages based on time-dependent data. To address this issue, this study proposes a novel deep learning approach that integrates 1D Convolutional Neural Networks (1DCNN), Long Short-Term Memory Networks (LSTM), and Residual Networks (ResNet) to enhance the efficiency of damage detection in constructions. The proposed method amalgamates 1DCNN for feature extraction, LSTM for recognizing long-term dependencies, and ResNet to counteract the vanishing gradient problem during deep network training. The effectiveness of the proposed method is assessed through the Z24 bridge dataset, one of the most comprehensive datasets in the field of SHM. The results indicate that the proposed approach achieves an accuracy of 81.5%, significantly outperforming traditional 1DCNN (78.7%), LSTM (79.3%) networks, and the combined 1DCNN-LSTM network (80.6%). This underscores the effectiveness of integrating deep learning techniques. The proposed 1DCNN-LSTM-ResNet method demonstrates considerable potential in practical applications for SHM, with superior accuracy and efficiency.