Modeling and Optimization of Data Center Energy Consumption

In this era of high-speed data development, the demand for computing that data centers(DC) need to process is rapidly increasing, and its high energy consumption problem is also becoming increasingly prominent. Currently, the total energy consumption of DC has occupied about 2.7% of the whole energy consumption of society, and optimizing the energy consumption of DC has been a research hotspot of concern in last several years. However, the current study ignores the start-stop relationship and climbing constraints of information technology (IT) equipment between different time periods, and the optimization strategy given by it may lead to frequent start-up and shutdown of IT equipment in actual operation when the arithmetic power demand changes greatly in a short period of time, resulting in excessive IT equipment loss. Based on this, the optimization model of data center energy consumption is established by considering the energy consumption characteristics of IT equipment and the heat exchange relationship of air conditioning equipment, as well as the start-stop relationship and the climbing constraint of IT equipment. For the dynamic and nonlinear characteristics of the temperature exchange process between IT equipment and the air conditioner, the DDPG algorithm is proposed for solving. Finally, the practicality and efficiency of the method are verified by a large-scale data center simulation example with 100,000 IT equipment.