PROGNOS-AI: AI-Driven Predictive Maintenance System Using Time-Series Sensor Data

Abstract

Predictive maintenance plays a critical role in modern industrial systems by enabling early detection of equipment degradation and reducing unexpected failures. This paper presents PrognosAI, an Attention-Enhanced Bidirectional Long Short-Term Memory (BiLSTM) based predictive maintenance system designed to estimate the Remaining Useful Life (RUL) of turbofan engines using the NASA CMAPSS dataset. The proposed system integrates a complete data processing pipeline including RUL computation, sensor feature engineering, normalization, and sliding window sequence generation. A stacked BiLSTM architecture captures long-term temporal dependencies, while a self-attention mechanism highlights the most informative time steps contributing to equipment failure. The trained model is deployed using a Streamlit-based dashboard that visualizes predicted RUL trends and categorizes engines into Critical, Warning, and Safe zones. Experimental results demonstrate accurate RUL prediction and effective degradation modeling, making the system suitable for real-time industrial predictive maintenance applications.