Levenberg–Marquardt DenseNet with hybrid distance-based Feature Fusion model for Earthquake Detection

Earthquake detection systems utilize sophisticated algorithms and seismic data to monitor and analyze seismic activity in real-time, which provides early warnings that reduce potential damage. However, these systems face several challenges, such as accurately forecasting seismic events, filtering out background noise from actual seismic signals and maintaining consistent performance in areas with limited infrastructure or sparse sensor networks. These limitations highlight the need for more robust and efficient detection models that can effectively process seismic data and improve detection reliability. To address these challenges, a new model called Levenberg–Marquardt DenseNet (LM_DenseNet) is proposed for detecting earthquakes. Initially, the input data is passed to the data transformation phase, which is done by the Box-Cox transformation. Next, a feature fusion technique is applied using a Siamese Convolutional Neural Network (SCNN) with hybrid distance measures that integrate City Block and Bhattacharyya distances. Finally, an earthquake is detected using the LM_DenseNet model, which is the incorporation of Levenberg–Marquardt (LM_Net) with DenseNet. The effectiveness of LM_DenseNet is examined by considering the metrics, like accuracy, sensitivity, specificity and F-measure with superior values of 96.87%, 96.17%, 96.72%, and 96.22%.