Operation Parameter Optimization of Cooling Systems in Data Centers Based on NN-HS

Abstract

The rapid development of Internet services such as cloud computing has accelerated the construction speed and scale of data centers. High energy consumption has become a prominent problem for data centers. Cooling systems account for a considerable proportion in the total energy consumption of data centers. Operation parameter optimization for reducing energy consumption is one of the key technologies to build green data centers. In order to solve this problem, we propose a new method with PUE (Power Usage Effectiveness) as performance index. Setting the operation parameters of cooling systems as the inputs and the corresponding PUE as the output, the neural network is constructed. The PUE function is approximated by this neural network with historical operating data of the data center of a certain bank in Beijing, and it is taken as the optimization objective. Considering other constraints, the relevant mathematical optimization model is established. At the same time, the harmony search algorithm is improved, including the search method of “seeking advantages and avoiding disadvantages” and the adaptive strategy. Therefore, novel improved harmony search algorithm (IHS) is proposed. Finally, the proposed IHS algorithm is applied to solve the above model. Based on the relevant instance data of the bank, the optimal operation parameters of the cooling system under given conditions are obtained. Through the analysis and comparison of optimal results, the effectiveness and feasibility of the proposed method are verified.