基于数据挖掘的通信基站能耗分析研究
发布时间：2018-07-17 02:35
【摘要】：随着通信基站用电量的不断增加,基站耗能的管控变得越来越重要,基于对公共建筑建立能耗标杆的方法的研究,进行数据挖掘的三种方法引用,从而建立通信基站能耗标杆。首先利用多元线性回归建立基站能耗标杆,确立影响基站耗电量的重要影响因素,通过标杆的分析结果给出节能和能耗管理建议;然后引用聚类算法将大量基站能耗数据分成8类,代表8种典型的基站能耗模式,通过分析能耗标杆得到通信基站的耗能特点及现行基站能耗管理的不足与建议;接着引用人工神经网络方法来预测基站全年能耗,预测精度达到最大相对误差为7.55%的水平,同时,根据人工神经网络的重要性分析得到影响基站能耗的最重要因素分别为:主设备功率,空调能效比和气温,从而给出节能应从这三方面抓起的管理建议。最后,对比多元线性回归、聚类分析和人工神经网络三种方法的优势和劣势,给出应用场景建议,多元线性回归为最经济简便的数据挖掘方法,聚类分析为最能获取深度知识的数据挖掘方法,人工神经网络为最适合应用于预测的数据挖掘方法,并综合三种方法的分析结果,给出通信基站的能耗管理建议。
[Abstract]:With the increasing of the power consumption of the communication base station, the control of the base station energy consumption becomes more and more important. Based on the research on the method of establishing the energy consumption benchmark in the public building, three methods of data mining are used to establish the energy consumption benchmark of the communication base station. Firstly, the energy consumption benchmark of base station is established by multivariate linear regression, and the important influencing factors of base station power consumption are established. The energy saving and energy consumption management suggestions are given through the analysis results of benchmark. Then a large number of base station energy consumption data are divided into 8 categories, representing 8 typical base station energy consumption modes. The characteristics of energy consumption of communication base station and the shortcomings and suggestions of current base station energy consumption management are obtained by analyzing the energy consumption benchmark. Then the artificial neural network method is used to predict the annual energy consumption of the base station, and the prediction accuracy reaches the maximum relative error level of 7.55%, at the same time, According to the analysis of the importance of artificial neural network, the most important factors affecting the energy consumption of base station are the power of main equipment, the energy efficiency ratio of air conditioning and the air temperature, respectively, and the management suggestions for energy saving should be made from these three aspects. Finally, compared with the advantages and disadvantages of the three methods of multivariate linear regression, clustering analysis and artificial neural network, the paper gives the suggestion of the application scenario that multivariate linear regression is the most economical and convenient method for data mining. Clustering analysis is the data mining method that can obtain the most deep knowledge, and artificial neural network is the most suitable data mining method for prediction. The analysis results of the three methods are synthesized, and some suggestions for energy consumption management of communication base stations are given.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP311.13;TN929.5
[Abstract]:With the increasing of the power consumption of the communication base station, the control of the base station energy consumption becomes more and more important. Based on the research on the method of establishing the energy consumption benchmark in the public building, three methods of data mining are used to establish the energy consumption benchmark of the communication base station. Firstly, the energy consumption benchmark of base station is established by multivariate linear regression, and the important influencing factors of base station power consumption are established. The energy saving and energy consumption management suggestions are given through the analysis results of benchmark. Then a large number of base station energy consumption data are divided into 8 categories, representing 8 typical base station energy consumption modes. The characteristics of energy consumption of communication base station and the shortcomings and suggestions of current base station energy consumption management are obtained by analyzing the energy consumption benchmark. Then the artificial neural network method is used to predict the annual energy consumption of the base station, and the prediction accuracy reaches the maximum relative error level of 7.55%, at the same time, According to the analysis of the importance of artificial neural network, the most important factors affecting the energy consumption of base station are the power of main equipment, the energy efficiency ratio of air conditioning and the air temperature, respectively, and the management suggestions for energy saving should be made from these three aspects. Finally, compared with the advantages and disadvantages of the three methods of multivariate linear regression, clustering analysis and artificial neural network, the paper gives the suggestion of the application scenario that multivariate linear regression is the most economical and convenient method for data mining. Clustering analysis is the data mining method that can obtain the most deep knowledge, and artificial neural network is the most suitable data mining method for prediction. The analysis results of the three methods are synthesized, and some suggestions for energy consumption management of communication base stations are given.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP311.13;TN929.5
【参考文献】
相关期刊论文 前10条
1 薛善斌;马波;姜峰;;基于能耗PUE标杆进行通信基站能耗管控的思路和办法[J];通信电源技术;2014年S1期
2 张辉辉;胡记文;李婷君;;移动通信发展史综述[J];电子制作;2013年24期
3 李江坤;张春海;;通信行业能耗现状分析与节能战略思考[J];信息通信;2013年06期
4 高波;;基站主设备节能技术的典型应用[J];内蒙古科技与经济;2012年18期
5 姜宁;;基站电费“标杆站”的研究[J];江苏通信;2012年02期
6 张英杰;章兢;刘文强;姚s，
本文编号：2128632
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2128632.html
