A novel method for damage identification in cantilever beams using Generalized energy index and deep learning model

• A novel method identifies beam damage using Generalized Energy Index and deep learning . • GEI as input boosts classification accuracy to 99.55%, outperforming traditional CNNs. • The method shows strong noise resistance and high accuracy under low-noise conditions. • An efficient and stable solution for structural health monitoring is experimentally validated. This study proposes a damage identification method for cantilever beams based on the Generalized Energy Index (GEI) and a deep learning framework that integrates Convolutional Neural Networks (CNN), Multilayer Perceptron (MLP), and Long Short-Term Memory (LSTM). The method uses GEI as the input feature, with CNN for spatial feature extraction, MLP for nonlinear mapping, and LSTM for enhancing the feature extraction capabilities. The proposed method was verified through numerical simulations and experimental tests. In the simulation studies, the classification accuracy reached 99.55 %, significantly outperforming traditional methods. To evaluate the robustness of the model, Gaussian noise was introduced into ten randomly selected test samples. The experimental results show that the model has strong noise resistance while maintaining a high prediction accuracy. To verify its practical applicability, a modal test was conducted on an aluminum cantilever beam. Subsequently, an MLP model was used for damage identification based on the GEI, achieving an experimental prediction accuracy of 93.10 %, and the accuracy of Leave-One-Out Cross-Validation (LOO-CV) was 88%, further confirming the robustness and practical viability of the GEI-based deep learning method in structural health monitoring .