贝叶斯电流分解:利用单个传感器感知家用电器电流  

Bayesian Current Disaggregation:Sensing the Current Waveforms of Household Appliances Using One Sensor

作　　者：刘晶杰[1] 聂磊 Liu Jingjie;Nie Lei(State Key Laboratory of Computer Architecture(Institute of Computing Technology,Chinese Academy of Sciences),Beijing 100190;University of Chinese Academy of Sciences,Beijing 100049)

机构地区：[1]计算机体系结构国家重点实验室(中国科学院计算技术研究所),北京100190 [2]中国科学院大学,北京100049

出　　处：《计算机研究与发展》2018年第3期662-672,共11页Journal of Computer Research and Development

基　　金：中国科学院战略性先导科技专项基金项目(XDA06010403);国家自然科学基金重点项目(61532016)

摘　　要：通过单个传感器对家庭中各个电器的用电行为进行感知是普适计算中的一个重要应用,其关键问题是电流分解,即在给定总电流波形情况下计算各个电器的实际电流.此问题现有2类求解方法:稳态估计方法和线性分解方法.前一类方法基于电器稳态耗电假设,使用稳态波形估计电器的工作电流.虽然该类方法能避免电器间的相互干扰,但是其结果不能反映总电流的实时变化.后一类方法通过模型约束或数据约束对电流波形进行线性降维,之后将总电流分解到各个低维线性空间中.虽然其分解结果能够反映总电流的实时变化,但是相似电器会降低分解结果的精度.从贝叶斯统计的角度将上述方法的关键假设松弛为位置向量先验分布与噪音先验分布,并提出了基于这2个分布的贝叶斯电流分解方法.利用真实用电数据,构造了多组仿真实验对此方法进行评测.实验结果表明:提出方法分解精度高于原有2类方法,其感知结果既能够反映总电流及各个电器电流的实时变化,又能够降低相似电器对分解结果的干扰.An important application of pervasive computing is to obtain the electricity usage information for each appliance in a household using one sensor.The key problem of this application is current disaggregating,which is to estimate the currents of individual appliances from the total current waveform.Existing methods to solve this problem can be classified into two classes:steady-state estimation methods and linear disaggregating methods.Based on the steady-state load assumption,the methods in the first class estimate the current for a running appliance using its steady-state current waveform.These methods can avoid the interference between appliances.But the results of these methods cannot reflect the real-time changes of the total current.The methods in the second class reduce the dimensions of the current waveforms for a specific appliance using model constraints or data constraints,and disaggregate the total current into the linear spaces with low dimensions.The results of these methods can reflect the real-time change of the total current,but similar appliances reduce the accuracy of disaggregating results.From the perspective of the Bayesian statistics,this paper relaxes the key assumptions of the above methods as the prior of position vectors and the prior of noises,and proposes a Bayesian current disaggregating method based these two priors.Using the electricity usage data generated by actual appliances,we conduct several simulation experiments to evaluate our method.The experiment results show that the accuracy of the proposed method is higher than previous methods.Our method not only reflects the real-time change of the total current,but also reduces the effects of similar appliances on the disaggregating results.

关 键 词：家庭用电 单点感知 电流分解 普适计算 贝叶斯方法 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]