Article
Predictive Maintenance Framework for Smart Manufacturing using Machine Learning Models: A Case Study
Predictive maintenance is a critical component of smart manufacturing in the industry 4.0, as it employs machine learning algorithms and data from IoT sensor networks to anticipate equipment malfunctions in advance. The paper examines how artificial intelligence (AI)-controlled predictive maintenance (PdM) can be applied to a smart manufacturing workplace that uses Internet of Things (IoT) technology. It covers data gathering, sensor technologies, and communication protocols as well as distributed storage schemes employed to derive and handle big industrial data. An automotive parts manufacturing plant case study used sensor data in real-time, such as temperature, vibration, acoustics and electrical data, and also used maintenance records to build predictive models. Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) models were designed and tested after a thorough data preprocessing and feature engineering. As shown by experimental results, LSTM was the most effective of CNN and other baseline models, with the highest accuracy (ACC) (88.1%), precision (PRE) (92.5%), recall (REC) (89.3%), and F1-score (F1) (90.9%). The implementation of the offered AI-based PdM system has led to the 24% of reduction of unscheduled downtime, 18% growth in equipment use, and 30% decrease in emergency maintenance prices. The results indicate that the incorporation of the Internet of Things-enabled sensing and sophisticated machine learning algorithms considerably boosts the reliability, operational performance, and decision-making concerning intelligent production systems.
Full Text Attachment
