Learning Enabled Data Center Operation Scheme for Participating in Power Market

As the key industrial infrastructure in the digitalera, the energy consumption of data center is increasing rapidly. which leads to a significant increase in the energy cost for data centers. Meanwhile, the nature of job scheduling can provide time-flexible loads for data centers. Therefore, Data centers can participate in the power market bidding as large energy customers and sense the market electricity price signal for job scheduling. This can not only improve the operating profit of data centers but also further coordinate with the power grid to maximize social welfare. In this paper, we establish a decision model for data centers that perceives market prices. Then, we provide the joint optimal strategy of job scheduling and bidding for the data center via a deep reinforcement learning approach. Finally, we validate our method through numerical experiments. The proposed strategy allows for greater flexibility and higher profits for data centers compared to other strategies.