predictive maintenance

Quality is the driving factor of the manufacturing industry. With the advent of Industry 4.0, the integration of advanced sensors, edge computing, artificial intelligence, digital twins, and 3D printing has great potential to achieve resilient manufacturing. The first part of this study addresses real-time quality control with edge computing. A digital imaging system is introduced, equipped with an edge device for real-time defect detection in additive manufacturing. AI promises great potential for achieving zero-defect manufacturing, which was barely possible in the past. Then, as the main structure provider for Industry 4.0, edge computing is described. The computer is used in quality control features of edge computing by real-time cloud monitoring and real-time elasticity for cloud services. In the second part, a predictive maintenance framework is proposed by integrating the digital twin, advanced data processing, and AI algorithms. AI finally determines the degradation status of products with high confidence to maintain the resilience of the product life cycle. Expensive fault events on products can be avoided with the help of AI-based advanced planning of maintenance at an appropriate time. The approach could mitigate potential risks of shortening lifespan, increasing maintenance costs, triggering catastrophic events, or even causing social impact if combined with real-time edge computing and advanced sensors. Such an approach advances the deployment of the manufacturing industry in the era of Industry 4.0.

This study aimed to investigate the role of artificial intelligence (AI) in predictive data center maintenance practices and strategies. Timely detection of different types of equipment faults and subsequent predictive maintenance can enhance data center availability dramatically and minimize costly outages. The rationale for the study came from the rapid growth of digital services and users, as well as the costly data centers supporting this growth. The costs of a minute (or more) of an unplanned service disruption range from $10,000 to $65,000. The three main Data Center Infrastructure Management (DCIM) components—the common sensors for continuous data acquisition, prevention to power off, and ensuing prediction and solution techniques—are investigated. Multiple other machine learning classification models are suggested to be tested on a similar dataset, in addition to classification, and to quantitatively test a real-life installation in future research. In addition, how these components interact in real-world environments is still not clear and would require advanced statistical analyses, which have not been done in this research. Yet, under the conditions of this study, the results demonstrate how AI elements can provide a reliable solution for enhanced data center DCIM and be applied in deliverable form.