随着互联网社会高速发展，包括数据中心在内的新型基础设施建设规模不断扩大，也带来巨大的能耗。在碳中和背景下，数据中心节能也成为企业和学术界的重点关注问题。数据中心节能可从提高IT设备能效，选取高效供配电系统，优化空调系统能耗管理等方面进行。本文基于对数据中心能耗特征的分析，在不改变数据中心已有设备的条件下，聚焦于数据中心冷源系统的节能优化控制策略研究。通过对不同地区数据中心实际运行数据调研，以及使用DeST模拟软件对数据中心进行能耗模拟分析，探索数据中心供冷能耗及PUE的特征规律，明确了数据中心冷源系统，尤其是冷水机组季节性运行策略对数据中心整体能耗影响较大的特点，确定了充分利用自然冷源以减少冷水机组开启时间和优化冷源系统运行参数两条节能优化路径。为了实现上述空调系统的节能优化控制路径，提出了基于物理模型的运行参数优化方法。首先根据运行数据建立冷源系统及末端空调的能耗模型和传热模型，使用基于最小二乘法的待定系数法求解能耗模型函数式的系数，建立冷水机组、泵与风机的能耗模型和板式换热器、末端表冷器换热模型，并使用NTU法建立冷却塔换热模型计算以冷却水出塔温度，最终获得精确度90%以上的冷源系统功率预测模型，并可以在误差0.57%范围内预测冷却水出塔温度。模型确立之后，建立以空调系统总能耗最小为优化目标的优化问题，在给定的冷冻水、冷却水流量范围、冷冻水供水温度范围等约束条件下，利用遍历法、粒子群优化法及退火算法等多种优化方法求解优化问题，结果表明遍历法一定可以找到全局最优点，且耗费时间在工程上可以接受，是一种稳定的方法。最后，以某实际数据中心为例进行了案例研究，验证模型的准确性与算法的有效性。案例研究结果显示，冷水机组使用时间减少了7.5%，完全利用自然冷源时间增加了36.87%，整体节能20.4%。本文针对数据中心空调系统的用能特征，提出了基于物理模型的空调系统运行参数优化方法，基于空调设备的能耗模型和传热模型，建立空调系统总能耗最小为优化目标的优化问题，通过求解优化问题，得出最优冷机运行台数、最优季节运行模式、最优冷冻水、冷却水流量、最优冷冻水供水温度。将所提出的空调系统运行参数优化方法在某实际数据中心进行了案例研究，验证了所提出方法有效性和节能效果，对于指导数据中心空调系统的优化节能运行具有重大实用价值。

With the rapid development of the Internet society, the scale of new infrastructure construction, including data centers, continues to expand, which also brings huge energy consumption. In the context of carbon neutrality, data center energy efficiency has also become a key concern for companies and researchers. Data center energy efficiency can be increased by increasing the utilization rate of IT equipment, selecting high-efficiency power supply and distribution systems, and optimizing the energy consumption management of air conditioning systems. Based on the analysis of the energy consumption characteristics of the data center, this research focuses on the energy-saving optimization control strategy of the cold-source system of the data center without changing the existing equipment and operation mode of the data center. Through building the energy consumption model of the data center through DeST to analyze the operation data of data centers in different regions and find the relationship between the outdoor air temperature and the cooling load and PUE of the data center, it is finally clear that the operation situation of cooling source system of the data center, especially the chiller, varies with seasons The operation has a great impact on the overall energy consumption of the data center. Therefore, two optimizing paths are confirmed. It is of great importance that the natural cooling source is fully utilized to reduce the turn-on time of the chiller, and optimizing the operating parameters of the cooling source system. In order to realize the energy-saving optimization control method, an optimization method of operating parameters based on physical model is proposed. Firstly, the energy consumption model and heat transfer model of the cooling source system and terminal air conditioner are established according to the operation data, and the least square method is used to solve the energy consumption formula to find the undetermined coefficient. The energy consumption model of the chiller, the pump and the fan, the plate heat exchanger and the heat exchange model of the terminal surface cooler are established, and the "air-water" heat exchange model of the cooling tower is established by the NTU method to calculate the cooling water temperature. Finally, the cooling source system power prediction model with an accuracy of more than 90% is obtained, and the cooling water outlet temperature can be predicted within the error range of 0.57%. After the model is established, an optimization problem with the minimum total energy consumption of the air-conditioning system as the optimization objective is established. Under the given constraints such as chilled water and cooling water flow range, and chilled water supply temperature range, the traversing method, particle swarm optimization method and simulated annealing algorithm are used to solve the optimization problem. The results show that the traversing method can definitely find the global optimal point, and the consuming time of the stable method is acceptable in engineering. Finally, a case study is conducted with an actual data center as an example to verify the accuracy of the model and the effectiveness of the algorithm. The results of the case study showed that the chiller usage time was reduced by 7.5%, the time to fully utilize the natural cooling source was increased by 36.87%, and the overall energy saving was 20.4%.In view of the energy consumption characteristics of the data center air conditioning system, this research proposes an optimization method for finding the operating parameters of the air conditioning system based on the established physical model. By solving the optimization problem, optimal operation parameters such as the number of chillers in operation, the seasonal operation mode, the chilled water supply temperature, the chilled water and cooling water flow rate are obtained. A case study of the proposed method for optimizing the operating parameters of the air-conditioning system was carried out in an actual data center, and the energy-saving effect of the proposed method were verified, which has great practical value for guiding the optimal energy-saving operation of the data center air-conditioning system.