Study on Real-time Leak Position Prediction based on Machine Learning in Phase-Chage Liquid-Cooled Data Center

Due to the advantages of high efficiency, energy saving, and safety, the application of phase change liquid cooling technology in data centers has great prospects. But there are few leak detection methods for the specific coolant used in liquid cooling system. So in this paper, first build a surrogate model based on machine learning, and then the real-time leak position method was studied based real-time data from gas detection sensors and proxy models. After testing, the effectiveness of the method has been verified and it can be applied in phase change liquid cooling data center in future.