Machine Learning Algorithm for Protected Predictive Maintenance in Smart Factory Systems

作者：Montater MuhsnHasan, Debarghya Biswas, M.Suganya, Gurram Vijendar Reddy, K. Sriram Kumar, Zahraa Eisa, Murtadha Yehya · 发表于：International Conference Control and Robots · 年份：2025 · DOI：10.1109/iccr67387.2025.11291646

Predictive maintenance is transforming industrial processes in smart manufacturing systems. This transformation enhances production processes, extends equipment life, and lowers downtime. SmartPredict-M was found to be a new platform for real-time predictive maintenance in Industry 4.0 environments. Industry 4.0 inspired the framework. This machine learning-based system was developed to reach this turning point. The system predicts equipment failure using intelligent analysis of sensor readings, operational records, and contextual components. This enables the machine to predict correctly. SmartPredict-M prioritizes machine degradation signals using a hybrid ensemble approach combining Random Forest, Gradient Boosting, and Deep Neural Networks with a dynamic feature selection process. This approach goes well with a hybrid ensemble. Automatic feature selection makes use of this approach. This approach benefits from the hybrid ensemble method. A real-time anomaly detection system has another advantage: it makes machines more responsive to changes in unexpected behavior. The module’s ability to identify irregularities in real-time. Early failure detection, false positive elimination, and maintenance schedule optimization are all strengths of Intelligent Predictive Machine (SmartPredict-M). A synthetic smart factory dataset accomplishes this. Smart industrial settings increase operational resilience, lower maintenance expenses, and allow data-driven decision-making. Many illustrat...