<![CDATA[The reliability of induced draft (ID) fans in cement production is critical to ensuring operational continuity, energy efficiency, and cost-effective maintenance. Excessive vibration in these fans often triggered by unbalance, misalignment, and bearing defects can lead to catastrophic failures, unplanned downtime, and increased operational expenses. This study presents the development of a deep neural network (DNN) model for predicting vibration anomalies in cement mill fans. Vibration signals were collected over a 34-week period from multiple sensors installed on induced draft fans at a cement production mill in Nigeria. Statistical time-domain features, including root mean square (RMS), kurtosis, crest factor, and impulse factor, were extracted and processed through advanced feature engineering and selection techniques. A Multi-Layer Perceptron (MLP)-based deep neural network was then designed, trained, and optimized in MATLAB. The model achieved high classification accuracy and robust generalization across different operational conditions. Furthermore, a real-time monitoring application was developed in MATLAB App Designer, enabling interactive visualization and prediction from Excel-based sensor inputs. The findings underscore the significance of integrating artificial intelligence into predictive maintenance workflows, demonstrating that deep learning-driven vibration prediction can enhance machine reliability, reduce downtime, and support the industry 4.0 agenda in cement manufacturing and other process industries.]]>
