Energy Consumption Forecast Of Data Center Fresh Air System Based On LSSVM Optimization

At present,the energy consumption of the data centers in our country is huge,but the energy consumption efficiency is not high,and the fresh air system which accounted for a high proportion of energy consumption is one of the important directions of data center energy saving.The online prediction of the energy consumption of the data center fresh air system can help to analyze the energy consumption of the data center in real time and improve the energy efficiency.Least squares support vector machine(LSSVM)has a good performance of generalization and promotion and fast calculation speed at the same time,which is very suitable for the energy consumption characteristics of the fresh air system in data center.The focus of this research is how to study the characteristics of data effectively,as much as possible to improve the quality of the data,and how to find the most suitable penalty factor and kernel function width parameter when the data samples are limited,so as to improve the prediction performance of the model.This paper first studies the current situation of building energy consumption prediction at home and abroad,and the application of intelligent algorithm in the field of energy consumption prediction.Then the main related technologies of energy consumption prediction are introduced,and the shortcomings of LSSVM load prediction model in data processing and parameter selection are analyzed.Finally,according to the energy consumption characteristics of the data center fresh air system,the energy consumption prediction model is optimized from the aspects of model and algorithm: the data processing module is added to the basic LSSVM model,and the LSSVM model is optimized by improving the quality of the data from the abnormal data processing algorithm,the selection of similar days and the dimensionality reduction of the input variables;based on the intelligent optimization performance of particle swarm optimization(PSO)algorithm,the IPSO-LSSVM model is proposed to overcome theshortcomings of PSO algorithm,and improved by introducing the mechanism of selection,crossover and mutation mechanism,optimizing the LSSVM model from the perspective of the algorithm.Finally,using the real load data as the experimental sample,the CCP-IPSO-LSSVM prediction model designed in this paper is used to predict the real-time energy consumption of the fresh air system in the data center.The validity of the model has been verified by the prediction results.