Simulation-based Multi-unit Collaborative Optimization in Data Centers

Abstract

With the rapid development of information technology in China, the problem of energy conservation in data centers has become increasingly prominent, and the green data center model is inevitable. The refrigeration equipment which is the main power consumption in the data center is affected by the environment, room temperature, unit output, and other factors. This paper provides a scheme of multi-unit collaborative optimization cooling. It is mainly divided into four stages: first, a model of the data center is built through Modelica, and it is exported as a fmu file. Second, the PSO optimization strategy is used to generate the data sets which contain unit output and cooling effect at different times. Third, LSTM and TCN load forecasting models are trained. Finally, through a comparison of various optimization algorithms, the PSO-LSTM-TCN model is used to simulate the unit output and optimize the output parameters of multiple units.