【作者】 张浩；

【导师】 贺明峰；

【作者基本信息】 大连理工大学 ， 应用统计（专业学位）， 2021， 硕士

【摘要】 数据中心是计算机服务器大量存放的建筑场所,他的快速建设保证了信息技术的飞速发展,带来了大量的经济效益。伴随着其建设程度的加深,数据中心带来的能耗问题也愈发严重。数据中心内部的服务器和空调制冷系统耗电量巨大,由于缺乏科学有效的调控措施,造成了很多的能源浪费。本文针对一类大型的数据中心,分析其能耗构成,对其空调制冷系统进行能耗优化。本文针对一类大型数据中心的能耗问题,考虑了数据中心机房的机器产热及空调制冷,以使机房中的热量在一个合理的限度内为基本条件,建立了机房的制冷能耗优化模型。对该模型,利用数据中心内部热平衡关系,空调作用的时间延后性,温度的变化受多时刻影响等特征,采用LSTM神经网络模型进行求解。通过实际案例求解结果表明,算法误差最大值为0.5度,平均误差在0.35度,案例中9个温度全部分布在23.5-24.5度之间,保证了案例的温度达标。同时空调的节能率在26%左右,降低了数据中心的能耗。案例说明本文采取的算法在不对现有数据中心空间布局做任何改变的前提下,仅通过历史数据建立能耗优化模型就可以控制温度达标,同时节省能耗,证明了模型的合理性以及算法的有效性。更多还原

【Abstract】 Data center is a building place where a large number of computer servers are stored.Its rapid construction ensures the rapid development of information technology and brings a lot of economic benefits.With the deepening of its construction,the energy consumption problem of data center is becoming more and more serious.The power consumption of servers and air conditioning refrigeration system in data center is huge.And a lot of energy is wasted because of the lack of scientific and effective control measures.This paper analyzes the energy consumption composition of a large data center,and optimizes the energy consumption of its air conditioning and refrigeration system.Aiming at the problem of energy consumption in data center,this paper considers the machine heat production and air conditioning refrigeration in the data center room,so as to make the heat in the room within a reasonable limit as the basic condition,and establishes the optimization model of refrigeration energy consumption in the room.For this model,the LSTM neural network model is used to solve it by using the characteristics of the internal heat balance relationship of the data center,the time delay of the air conditioning effect,and the multi time influence of the temperature change.Through the actual case,the results show that the maximum error of the algorithm is 0.5centigrade,the average error is 0.35 centigrade,and all the nine temperatures in the case are distributed between 23.5 and 24.5 centigrade,which ensures that the temperature of the case reaches the standard.At the same time,the energy saving rate of air conditioning is about26%,which reduces the energy consumption of data center.The case shows that the algorithm adopted in this paper can control the temperature to reach the standard and save energy consumption without any change to the existing spatial layout of data center,and proves the rationality of the model and the effectiveness of the algorithm.更多还原