Multimodal information fusion and artificial intelligence approaches for sustainable computing in data centers

Abstract

With the rapid expansion of cloud computing, artificial intelligence, and big data analytics, data centers have become integral to modern digital infrastructure. However, their escalating energy consumption poses significant challenges for sustainable operations. This paper presents a novel multimodal data fusion algorithm aimed at optimizing energy management in data centers. By integrating environmental sensor data, system logs, and visual information, we constructed a comprehensive framework for analyzing energy consumption patterns. Experiments conducted on publicly available datasets validated the algorithm’s effectiveness in energy prediction, enhancing energy efficiency, and optimizing server loads. Results indicate that the proposed method outperforms traditional baseline algorithms such as Support Vector Machines, Random Forest, and Long Short-Term Memory networks across multiple evaluation metrics. Additionally, the algorithm demonstrates good computational efficiency, making it suitable for deployment in large-scale data centers. Our research provides a significant theoretical foundation and practical guidance for sustainable energy management.