Structural Health Monitoring and Damage Detection Using Namib-Beetle Optimization Algorithm with 1D Convolutional Neural Network

Structural Health Monitoring and damage detection in structures, buildings and roadways are an eminently vital field due to the widespread utility of the method, ranging from public security to civil application. Numerous architectures are used for extraction, selection and classification to detect the location, extent of the damage and monitor the structure. However, traditional dimensionality reduction methods exhibit poor-accuracy in dealing with non-linear distributions which manifest at higher Degrees of Freedom in vehicle-bridge interactions. This work proposes a Namib-Beetle Optimization Algorithm (NBOA) with 1D-Convolutional Neural Network (CNN) which helps to minimize the feature cost function and effectively capture sequential time series data in classification in structural health monitoring and damage detection. In the preprocessing step, frequency transformation, signal averaging and normalization procedures are considered to eliminate the stochastic nature of time domain processes and then, an Adversarial Auto-Encoder (AAE) architecture is used to perform feature extraction and understand the latent space in the architecture. NBOA is used to perform feature selection and 1D-CNN performs classification with reduced computation and an increased attention to features. The proposed NBOA with 1D-CNN achieves an Area Under the Curve (AUC) value of 0.987 compared to the existing methods like Convolutional Auto- Encoder module and U-Net based Auto-Encoder module.