Workload Scheduling of Data Center Based on Electricity Price and Load Forecasting

With the rapid development of artificial intelligence and information technology, the scale of data center has increased rapidly. As a high-energy-consuming industry, data centers consume huge amounts of electricity, and their electricity expenses account for about 70% of the operating costs. Therefore, there is an urgent need for data centers to change their electricity usage behavior to reduce electricity bills. In addition, the workload of the data center is transferable in the time dimension. Based on this, this paper firstly constructs an energy consumption model based on workload transfer to characterize the energy consumption of the data center. Then, the LSTM-based prediction model for electricity price and data center's workload is built. Lastly, the load optimization scheduling model of data center based on electricity price and load forecasting is applied to minimize electricity cost.