Abstract:Advances in Structural Engineering, Ahead of Print. This paper proposes a Structural Health Monitoring (SHM) approach to detect localized damage by employing a Multi-Label Radial Basis Function Neural Network (ML-RBFNN). The proposed research methodology aims to identify structural damage more efficiently. This involves dividing the main structure into smaller sub-structures known as zones instead of employing a global search for damage across the entire structure. Each zone is further divided into smaller groups, including sub-zones, nodes, or elements. To reduce the complexities associated with feature extraction, auto-regressive (AR) model parameters extracted from raw time series are considered damage-sensitive features. Considering the simultaneous effects of multiple sensors can increase the number of inputs to the networks, making it challenging to reduce the complexity of the dataset. To address this problem, the principal component analysis (PCA) method is utilized to decrease the dimension of feature space. The effectiveness of the proposed approach is evaluated through a comprehensive comparative analysis against conventional single-stage damage detection methods, with a focus on computational costs and accuracy. Furthermore, the noise-sensitivity characteristics of the proposed method are explored, providing insights into their robustness under various conditions. The validation of the approach is carried out through numerical modeling using the ASCE benchmark structure and experimental data collected from the Qatar University Grandstand Simulator. The application of ML-RBFNN with AR parameters as features yielded a remarkable accuracy exceeding 98% in diagnosing damage occurrence and location for both multi-stage and single-stage SHM methods. Moreover, our proposed approach demonstrates enhanced computational efficiency compared to the single-stage method.

ANN-based damage detection with two-stage approach and model reduction

The Application of a Damage Detection Method Using Artificial Neural Network and Train-Induced Vibrations on a Simplified Railway Bridge Model

Effectiveness Analysis of Structural Anomaly Diagnosis Based on ANN Model

Structural Damage Monitoring Based on Vibration Modal Analysis and Neural Network Technique

COMPOSITE STRUCTURAL DAMAGE MONITORING BASED ON RATIO OF MODAL STRAIN ENERGY AND NEURAL NETWORK TECHNIQUE

Two-Stage Damage Detection Method Using The Artificial Neural Networks And Genetic Algorithms

Damage Detection in Structural Health Monitoring Using an Integrated ANNIRSA Approach

A hybrid approach for structural health monitoring integrating EMI with ANN

Multi-Level Interval Estimation for Locating Damage in Structures by Using Artificial Neural Networks

Damage Detection in Structures Using Modified Back-Propagation Neural Networks

Damage diagnosis of spatial trusses based on modal strain energy and neural network

Statistical Detection of Structural Damage Based on Model Reduction

A novel unsupervised real‐time damage detection method for structural health monitoring using machine learning

Structural damage detection based on fundamental Bayesian two-stage model considering the modal parameters uncertainty

Multi-stage Structural Damage Diagnosis Method Based on "Energy-Damage" Theory

A combined method of wavelet and ANN for damage detection of structure

A multi-stage sub-structural damage localization approach using multi-label radial basis function neural network and auto-regressive model parameters

Neural Networks-Based Structural Damage Detection Through Modal Parameter Measurements

Structural Damage Identification Based on Genetically Trained Anns in Beams

A Nonlinearity-Sensitive Approach for Early Damage Detection Using NOFRFs and the Hybrid-LSTM Model.

Damage Identification for Large Span Structure Based on Multiscale Inputs to Artificial Neural Networks