کنفرانس مهندسی برق ایران

صفحه اصلی / سی و چهارمین کنفرانس بین المللی مهندسی برق

A Comparative Study of Efficient Power Distribution in Data Centers Based on a Novel Decision-Making Predictive Model

نویسندگان :

Zahra Mojtahedin¹ Omid Farsi² Siavash Bayat³

1- دانشگاه صنعتی شریف 2- دانشگاه صنعتی شریف 3- دانشگاه صنعتی شریف

کلمات کلیدی :

Data centers،DC power distribution،Distributed generation،Energy efficiency،Sequential decision making

چکیده :

The high energy consumption and strict reliability demands of modern data centers make efficient power distribution and intelligent energy management critical research challenges. In this study, a comprehensive framework for DC power distribution in data centers based on a 380 V DC architecture is proposed. The system integrates a predictive control strategy for boost converter voltage regulation, coordinated operation of distributed energy resources in a DC microgrid, and neural-network-based voltage stability validation. A sequential decision-making (SDM) approach is employed to determine the optimal duty cycle of the boost converter in real time using locally available measurements. The data center’s microgrid is simulated in two modes: operating independently from the main grid and operating while connected to it. Simulation results demonstrate a significant improvement in DC bus voltage stability and a reduction of source current total harmonic distortion from 29.51% to 2.51% after applying the proposed learning based control strategy. Moreover, in the grid-connected scenario, optimized energy management improves the performance of the microgrid. A neural network model trained using continuation power flow data confirms the high accuracy of voltage stability prediction for both AC and DC structures. The results verify the effectiveness of the proposed 380 V DC power distribution framework for next-generation data centers.