A Power Consumption Prediction Method for Data Center Refrigeration System Based on LSTM

The energy consumption problem caused by data centers is becoming more and more serious, and air conditioning systems rank second only to IT equipment in terms of power consumption. Accurate prediction of air conditioning power consumption in data centers not only helps enterprises grasp the cost level of data centers, but also helps data centers to save energy and build green data centers. This paper presents an energy consumption prediction method for air-conditioning in data center based on LSTM network. This paper also utilizes simulation software to model data center facilities and simulate the power consumption necessary to maintain the air conditioning system under various conditions. The experimental results demonstrate that utilizing the LSTM neural network model results in more accurate predictions for air conditioning in data center facilities compared to the BP neural network model.