Akademik Veri Yönetim Sistemi
Computational Particle Mechanics, 2025 (SCI-Expanded)
This study presents a novel approach for detecting grain breakages in industrial production environments using a hybrid deep learning architecture combining the Xception convolutional neural network (CNN) with a long short-term memory (LSTM) network. The proposed method leverages the feature extraction capabilities of Xception to process data and integrates temporal sequence modelling via LSTM to enhance predictive accuracy. The model is trained and tested on a specialized dataset comprising various stages of breakage. Results demonstrate that the Xception-LSTM model significantly outperforms traditional CNN architectures in terms of accuracy, precision, and recall. Notably, the hybrid model achieved an accuracy of 98.7%, with a marked improvement in sensitivity to minor and early stage breakages. The study underscores the model’s robustness in real-time applications, indicating its potential for deployment in automated monitoring systems to prevent costly downtimes and ensure production continuity. The findings contribute to the field by offering a scalable and efficient solution that bridges advanced modelling with time-series analysis for industrial fault detection.