Research On Energy Saving Method Of Data Center Computer Room

Along with the advent of big data and cloud computing era,especially the continuous advancement of smart city construction,the number of all kinds of entrepreneurial big data has soared which has driven the rapid growth of data centers.Its large amount of energy consumption and the ensuing environmental problems have drawn increasing attention.Therefore,the research on data center energy-saving methods has become one of the hot issues in the current data center construction and operation.There are many factors that affect the energy consumption of the data center Among them,the energy consumption of the server and air-conditioning system accounts for more than 70% of the total energy consumption of the data center.Therefore,optimizing the operation mode and control mode of servers and air conditioners is of great significance to data center energy conservation.In order to ensure stable operation of the server and no local hot spots,the room temperature is usually set at a low temperature of 20 to 21 ?C,resulting in great waste of energy.The traditional temperature control method relies on the collector to obtain the temperature which has a large time delay.At the same time,the temperature characteristics and the complexity of the environment make it difficult to meet the demand.Therefore,this paper studies the optimization control of the engine room temperature from the following aspects:(1)Analysis of energy saving factors.The energy saving factors of data center are mainly divided into three directions: server,air conditioning system and temperature control.The relationship between the three and the data center equipment energy consumption is described separately,and the energy consumption models of the three are analyzed.In this paper,the data center energy-saving method is studied from the perspective of temperature control.By comparing the advantages and disadvantages of several types of computer room temperature control methods,the model prediction control is selected as the control method of this paper.(2)Computer room temperature prediction model.Accurate prediction of the temperature of the equipment room is the premise and key to temperature control.The data center temperature acquisition interval is large,resulting in a small sample size of temperature data.Support vector machine is an ideal method for nonlinear modeling.The selection of its nuclear parameters and penalty factors has a critical impact on prediction accuracy.Therefore,particle swarm optimization and particle swarmoptimization are used to optimize parameter selection and improve prediction accuracy.Air conditioning supply air temperature,return air temperature,air volume and rack temperature and humidity are selected to predict future rack temperatures.(3)Temperature prediction control method.Model predictive control comes down to three aspects: model prediction,rolling optimization,and online correction.Temperature prediction control method.Model predictive control comes down to three aspects: model prediction,rolling optimization,and online correction.The accuracy of the model prediction directly affects the effect of the control.Using the ACOPSO-SVM prediction model already proposed above,combined with the genetic algorithm rolling optimization strategy,dynamic adjustment to achieve optimal control of the room temperature.Through the simulation analysis and comparison,the improved temperature prediction model has better prediction accuracy and smaller error than the support vector machine,and can better perform the task of predictive model in predictive control.At the same time,the prediction control simulation results show that the control method based on ACOPSO-SVM prediction model has better response speed and higher control precision.By comparing the controlled air conditioning temperature with the actual air supply temperature,the predictive control optimizes the output of the air conditioner temperature and shows its good energy saving effect.