Energy-Saving Regulation Method of Data Center Cooling System Based on Machine Learning

As a key infrastructure supporting modern information technology, data center's energy consumption has increasingly become a global problem. In order to improve the energy efficiency of data center and reduce the operating cost, this study proposes an energy-saving regulation method of data center cooling system based on machine learning(ML). By integrating the innovative deep learning neural network model, this method can intelligently analyze the operation state of the data center and adjust the cooling system parameters in real time to optimize energy utilization. In the experiment, we use simulated data sets to verify and design a complete experimental process. By comparing the actual energy consumption with the prediction results of the model, it is observed that the model has good generalization ability under different loads and operating conditions. After applying the model, the energy consumption has obviously decreased, which shows that the energy saving effect is remarkable. The Long Short Term Memory (LSTM) model performs significantly better than simple linear regression models in predicting energy consumption. Adopting an improved LSTM as the core model to adapt to the complex temporal characteristics of data center operation status. The improved LSTM model exhibits good convergence performance during training and validation, while avoiding overfitting issues. This ML based energy-saving regulation method for data center cooling systems has high feasibility and effectiveness. This innovative framework provides a new technological approach for data centers to achieve green operations, and is expected to achieve more significant energy-saving results in practical applications.