-
摘要: 非侵入式负荷监测通过对总负荷电表数据进行分析处理, 能够实现对各个用电设备及其工作状态的辨识, 可广泛应用于建筑节能、智慧城市、智能电网等领域. 近年来, 随着智能电表的大规模部署以及各类机器学习算法的广泛应用, 非侵入式负荷监测引起了学术界与工业界的共同关注. 本文对非侵入式负荷监测方面的研究进行综述. 首先提炼非侵入式负荷监测的问题模型及基本框架; 然后分别对非侵入式负荷监测的数据采集与预处理过程、负荷分解模型与方法、常用数据集及评估指标进行归纳总结; 最后, 对目前研究中存在的挑战进行分析, 并对未来的研究方向进行展望.Abstract: Non-intrusive load monitoring can realize the identification of individual electrical equipment and its working state by analyzing and processing the aggregated load data from electricity meters, which can be widely used in building energy conservation, smart cities, smart grids, etc. With the large-scale deployment of smart meters and the widespread application of various machine learning algorithms, non-intrusive load monitoring has aroused the common concern of academia and industry in recent years. This article reviews the research on non-intrusive load monitoring. First, the mathematical model and basic framework of non-intrusive load monitoring are refined, and then we separately summarize the data collection and pre-processing process, load disaggregation models and algorithms, data sets and evaluation metrics used in non-intrusive load monitoring. Finally, some challenges that the current research are confronted with are analyzed and we give some views on the future research.
-
近年来, 物联网、信息物理系统、工业互联网及其应用领域, 例如智慧家居、车联网、智慧工厂等各种智能互联系统得到了迅猛的发展.这种发展得益于通信和计算两种技术的巨大进步, 一方面, 通信技术的进步, 特别是无线通信网络的发展, 使得信息流动更容易, 成本更低, 促进了大规模互联系统的构建; 另一方面, 计算能力的进步, 特别是微型嵌入式计算芯片的发展, 使得更多的微型设备也可以具有计算能力, 为系统的智能化发展提供了物质基础[1-6].无线通信和嵌入式计算等技术在控制系统中的广泛使用为控制系统设计和分析提出了新的挑战, 也促使了各个不同但相关的研究领域的出现和发展.从考虑各种通信网络在控制系统信息传输中的作用的角度, 出现了网络化控制系统的概念, 该领域在过去近20年的大量研究为包含通信网络因素的控制系统的设计和分析奠定了坚实的基础[7-10]; 从考虑多个设备间互联及其拓扑结构和群体行为的角度, 为多自主体系统的研究提供了一个重要的模型基础[11-12]; 从物理系统和信息系统交互影响的角度, 信息物理系统的研究也在近些年蓬勃发展起来[13-14].
与上述角度不同, 本文关注各类智能系统中大量采用的微型处理器及其带来的计算能力限制问题.事实上, 各类智能系统以多个具有有限计算和通信能力的设备的互联互通为基本特征, 势必导致其计算能力成为关注的焦点, 一方面运行控制器设备本身的计算能力有限, 另一方面又要在互联的设备间共享计算资源, 使得对某特定的控制系统而言, 其控制器的可用计算资源成为有限时变且不可预测[15-16].
传统控制系统的可用计算资源很少有不可预测的特点.因为传统控制系统往往不存在大量设备的互联, 从而在系统边界上是封闭的, 即控制系统的各个组件对该系统是私有的.因此, 即便计算资源有限时变, 这一变化特点在控制系统的范畴内也往往是事先预知的.而对已知的时变资源的分配可以采用各种资源调度的算法进行有效处理[17-19].
在本文研究的问题中, 对于提供计算能力的设备, 仍依赖调度算法对该设备的计算资源进行全局分配, 但是, 因为全局调动算法的分配很难实时准确地将控制系统的需要考虑在内, 分配给控制系统的可用计算资源就难免具有有限时变且不可预知的特点.有限时变的特点意味着在某些时刻控制系统可能会因为计算资源太少而无法运行控制器, 使得系统开环, 在极端情况下可造成系统失稳, 而不可预测的特点又使得任何离线设计的方法失效.这使得不考虑这一新特性的传统控制方法很难在这种新情境下有效发挥作用, 从而要求一种全新的在线控制框架来处理这一问题.
本文面向计算受限控制系统提出一种全资源预测控制策略及相关控制器设计方法.该策略注意到了有限时变可用计算资源的另一个可能特点, 即在某些时刻可用计算资源可能极为充足, 在运行控制器外还有大量剩余, 因此就可以使用这些充足剩余的计算资源进行控制量的前向预测, 这些控制预测量进而能够在控制系统可用计算资源不足时闭合系统.这一策略仅使用已分配给控制系统的计算资源, 但通过对其更充分的利用可获得更好的控制系统性能.
本文结构安排如下:第1节严格定义所考虑的问题; 第2节提出全资源预测控制策略; 第3节提供了一种改进的模型预测控制算法作为该策略所要求的控制器的一种具体实现; 第4节使用基于MATLAB的数值仿真和MATLAB/LabVIEW联合仿真对提出的策略和算法进行验证; 第5节总结全文.
1. 问题描述
考虑图 1所示的控制系统, 其受控对象由离散时间多输入多输出线性时不变系统描述.
$ \begin{align}\label{sys} x(k+1)=Ax(k)+B u(k) \end{align} $
(1) 
图 1 具有时变不可预知计算资源的控制系统Fig. 1 Control systems with time-varying and unpredictable computational resources其中, $x(k)\in{\bf R}^n$, $u(k)\in{\bf R}^m$, $A\in{\bf R}^{n\times n}$, $B\in {\bf R}^{n\times m}$.
在图 1中, 控制器通过计算能力受限的某嵌入式处理器实现, 并且该处理器同时负责其他系统(可能并非控制系统)的计算任务.这一模型的提出受近些年快速发展的无线传感网络和信息物理系统等类系统的启发, 该类系统具有计算能力受限的设备大规模互联使用的基本特征.该模型意味着分配给所考虑的控制系统的计算资源一方面是有限的, 因为处理器的计算能力本身有限, 另一方面又呈现出不可预知的时变特性, 因为该计算能力与所考虑的控制系统外的其他系统和设备共享, 造成该计算能力在所考虑控制系统的边界之外而无法有效管理.
如果用$R(k)$表示$k$时刻分配给所考虑控制系统的可用计算资源, 用$R^*$表示处理器的总计算资源, 则有
$ \begin{align} R(k)\le R^* \end{align} $
(2) 将计算控制信号$u(k)$所需的最小计算资源记为$R_u^*$, 则应有$R^*_u<R^*$, 否则该处理器的最大计算能力都不足以运行控制器, 也就是对控制系统而言完全不可用.在时刻$k$控制量可被计算的条件是在此时刻控制系统可使用的计算资源不小于所要求的最小计算资源, 即
$ \begin{align} R_u^*\le R(k) \end{align} $
(3) 如式(3)不满足, 则控制信号$u(k)$无法计算.若式(3)太长时间不被满足, 则系统将很可能因为保持开环太长时间而不稳定, 这在实际系统中是不允许的.
式(2)和式(3)可合并为式(4), 这是传统控制方法能够使用的前提条件.
$ \begin{align} R_u^*\le R(k) \le R^*, ~~~ \forall k \end{align} $
(4) 然而, 可用计算资源的时变特性意味着在某些时刻$R(k)$会太过有限而不能计算控制信号, 即
$ \begin{align} \exists k, ~~~ {\rm s.t.} ~~~R(k)< R_u^* \le R^* \end{align} $
(5) 为了解决上述问题, 一种可能的解决方案是对计算资源进行全局调度.然而, 调度算法在该问题中难以见效, 原因有二: 1)在处理器的计算资源是有限的并且共享的系统和设备对处理器计算资源要求较高的前提下(这是本文考虑问题的出发点), 式(4)中的计算资源要求不可能对所有时刻都满足(否则传统控制方法即可满足要求), 意味着式(5)中的情形总会出现(在这些情形时传统控制方法失效); 2)调度算法通常无法将控制系统的实时计算资源要求考虑在内, 因此, 或者会分配计算资源过少造成式(5)中的情形, 或者太过保守将非必需的宝贵的计算资源分配给控制系统而造成浪费, 也就是会出现如下情形:
$ \begin{align} \exists k, ~~~ {\rm s.t.} ~~~R_u^*\ll R(k) \le R^* \end{align} $
(6) 综合上述讨论, 本文考虑如下问题:
问题1.考虑如图 1所示的控制系统, 其中控制系统可用计算资源是有限时变而不可预知的, 即在某些时刻会太少而不能执行控制器的计算(式(5)), 而在其他一些时刻在执行完控制器计算后还有大量剩余(式(6)).设计控制策略以有效使用计算资源, 提升控制系统性能.
注1. 图 1中嵌入式处理器的有限计算资源在控制系统的控制器和其他可能的共享设备中通过某种调度算法进行分配.本文并不研究这些计算资源的调度算法, 而是在一定的计算资源调度算法下, 研究如何对分配给控制系统的有限时变而不可预知的计算资源进行高效的使用.在这个意义上, 本文所做的研究并非是计算资源调度算法的替代, 而是其在控制系统应用场景下的补充.
2. 全资源预测控制策略
本节首先讨论控制系统可用计算资源与控制预测步长的一个基本关系, 随之给出全资源预测控制策略的描述, 最后讨论使用该策略的算法性能和闭环系统的稳定性.
2.1 可用计算资源与控制预测步长
本节的讨论主要基于如下事实[20]:控制器所需的计算资源随着控制预测步长的增加而严格增加.注意到计算资源一般由分配给的处理器时间来衡量, 在不引起误解的前提下, 用$R(k)$自身表示在$k$时刻分配给控制系统的处理器时间, 从而$R(k)\in {\bf R}_0^+$, 这里${\bf R}_0^+$是非负实数集.记计算控制预测步长$i$所需最小计算资源为$R_i^*$ (从而$R_0^*=0$且$R_1^*=R_u^*$), 有如下的从可用计算资源到控制预测步长的映射, $f:$ ${\bf R}_0^+$ $\rightarrow {\bf N}_0^+$, 其中${\bf N}_0^+$是非负整数集.
$ \begin{align} f(R(k))=N_k, ~ ~ ~ R_{N_k}^*\le R(k)<R_{N_k+1}^* \end{align} $
(7) 式(7)说明, 对时刻$k$的任意可用计算资源, 将有唯一的控制预测步长$N_k$与之对应.可以注意到, 函数$f(\cdot)$是分段常值且右连续的.
式(7)中的映射提供了看待控制系统可用计算资源$R(k)$的另一个视角, 即它可以被等价看作计算资源可以处理的控制预测步长$N_k$.另外, 从式(5)和式(6)可以得出: 1)若式(5)成立, 则$N_k=0$; 2)若式(6)成立, 则$N_k\gg 1$.
前一个条件$N_k=0$使得传统控制算法在此处失效, 因为可用的计算资源连一步控制信号都无法算出(要求$N_k\ge 1$, $\forall k$), 而后一个条件$N_k\gg 1$则提供了更为有效利用可用计算资源的可能性.注意到在这种情况下, 利用可用计算资源可以计算多步控制信号.
注2.式(7)中的映射是依赖具体的控制算法的.若控制算法本身复杂度高, 则由$R(k)$映射到的$N_k$值较小, 否则$N_k$值较大.但这并不影响这一映射的基本形式和$N_k$随$R(k)$递增且函数$f(\cdot)$是分段常值且右连续的基本性质, 因此也并不影响本文后续的讨论.
2.2 全资源预测控制策略及算法
本文提出的针对计算受限控制系统的全资源预测控制策略的基本架构见图 2.
图 2 面向计算受限控制系统的全资源预测控制策略Fig. 2 The full resource predictive control strategy for control systems subject to computational constraints传统控制方法仅计算当前时刻的控制量, 而在本文的全资源预测控制策略中, 可用的计算资源被充分使用计算尽可能多的前向预测控制量.这些预测控制量进而存储在执行器端以备使用.依照式(7)中的映射, 并记基于$k$时刻信息计算的$k+i$时刻的控制预测量为$u(k+i|k)$, $i=0, 1, \cdots, N_k$, 则$k$时刻的前向预测控制序列可构建如下:
$ \begin{align}\label{FCSOnlyCompCons} U(k|k)=[u(k|k) ~ u(k+1|k) ~ \cdots ~ u(k+N_k-1|k)] \end{align} $
(8) 图 2中的控制信号选择器包含两个部件: 1个寄存器, 能够存储最大可能长度的前向预测控制序列; 1个简单的逻辑电路, 可以用来选择合适的控制信号.控制信号选择器按如下方式工作:
1) 在新的前向控制预测序列到达时更新寄存器, 保证寄存器中的前向控制预测序列总是最新的.由于式(5), 寄存器中$k$时刻的前向控制预测序列并不一定是在当前时刻计算的.记$k$时刻寄存器中的前向控制预测序列已经在其中存储了$\chi_k^*$步, $k$时刻的前向控制预测序列则是在$k-\chi_k^*$时刻计算, 可写为$U(k-\chi_k^*|k-\chi_k^*)$.
2) 在当前计算资源不足无法计算控制信号时, 从前向控制预测序列中选取合适的替代控制量.使用上面符号, $k$时刻从$U(k-\chi_k^*|k-\chi_k^*)$中选取的控制量为
$ \begin{align}\label{conSigOnlyCompCons} u(k)=u(k|k-\chi_k^*) \end{align} $
(9) 尽管选取的控制量$u(k|k-\chi_k^*)$是基于滞后的系统状态(时刻$k-\chi_k^*$)计算的, 但该控制量是专为$k$时刻计算的, 因此若合理设计前向控制预测序列, 仍可在无法获得当前控制量时提供比传统的零控制更好的控制性能.
为使得在任何时刻从前向控制预测序列中都有控制量可用, 可用计算资源需满足一定条件, 如推论1所述.这也是此后所提算法1能够工作的前提条件.
推论1.若下式成立:
$ \begin{align}\label{conditionPBWork} N_{k-\chi_k^*}\ge \chi_k^*+1, ~~~\forall k>\chi_k^* \end{align} $
(10) 则式(9)中的控制量总在控制信号选择器中可供选择.
证明.注意到$k$时刻的前向控制预测序列$U(k$ $-$ $\chi_k^*|k-\chi_k^*)$是在$k-\chi_k^*$时刻计算的.为了使得按照式(9)选取的控制量在$U(k-\chi_k^*|k-\chi_k^*)$中, 前向控制预测序列的长度, 即$N_{k-\chi_k^*}$, 需大于或等于$k$ $-$ $(k-\chi_k^*)+1$, 这正是推论中给出的条件.
注3.考虑如下两种推论1的特例情况.
1) $\chi_k^*\equiv 0$.这一条件意味着控制系统的可用计算资源$R(k)$总能至少计算一步控制量, 即$N_k\ge 1$, $\forall k$.在这种情形下, 传统控制方法也可使用, 因此, 在此意义下本文方法也将传统控制方法作为特例.
2) $\exists k$, $\chi_k^*>0$.这一条件意味着控制系统的可用计算资源$R(k)$在时刻$k-\chi_k^*+1$到$k$内总不能计算任何控制量, 即$N_i\equiv0$, $k-\chi_k^*+1\le i \le k$.在这种情况下, 传统控制方法将无法工作, 而本文方法却总可以有预测控制量$u(k|k-\chi_k^*)$可用.
所提出的策略总结为算法1, 采用该算法的系统整体架构见图 2.
算法1.全资源预测控制策略
步骤1. 传感器采样系统状态并将采集数据传输至控制器;
步骤2. 由可用计算资源$R(k)$按照式(7)获得可计算的控制预测步长$N_k$, 进而按照式(8)计算前向控制预测序列并将之传至执行器;
步骤3. 控制信号选择器更新寄存器, 按式(9)选择控制量并令执行器执行控制量.
注4. 算法1并未考虑计算时间导致的时延.因为与控制系统的时间尺度相比, 该计算时延往往是足够小且可以忽略的.不过, 需要指出的是, 如果绝对必要, 这一时延可以放入到执行器的等待时延$\chi_k^*$中, 因此可以在不更改系统架构下进行有效处理.
注5. 算法1仅对已经分配给控制系统的计算资源做有利于控制系统性能的合理使用, 其权限局限于控制系统本身.对整个嵌入式处理器的计算资源的分配需由上一级机构, 即计算资源的调度算法做出.本文的全资源预测控制策略是在此意义下对计算资源调度的一个有效补充.
2.3 算法1性能讨论
传统控制方法在无法获得当前控制信号时, 往往或者使用零阶保持(即保持先前的控制输入), 或者采用零控制(使系统临时开环).这样, 面对本文考虑的计算资源有限的控制系统, 使用传统控制方法实际所采用的控制信号如下:
$ \begin{align}\label{controller-without-strategy} \tilde{u}(k)=\begin{cases} u(k-\chi_k^*|k-\chi_k^*) ~\mbox{或}~ 0, & \mbox{若式}~(5)~\mbox{成立} \\[1mm] u(k|k), & \mbox{其他} \end{cases} \end{align} $
(11) 与之相对应的, 采用全资源预测控制策略的实际控制信号由式(9)给出, 或等价地写为
$ \begin{align}\label{controller-with-strategy} u(k)=\begin{cases} u(k|k-\chi_k^*), & \mbox{若式}~(5)~\mbox{成立} \\[1mm] u(k|k), & \mbox{其他} \end{cases} \end{align} $
(12) 从式(11)和式(12)可以看出, 在控制系统被分配的计算资源充足时, 就实际使用的控制信号而言, 使用或不使用本文提出的全资源预测控制策略并无区别; 但是, 在当前计算资源不足以进行控制计算时(即式(5)成立), 本文策略提供了额外的自由度, 使得可以针对具体情形设计式(8)中的前向预测控制序列对本来缺少的控制信号进行替代补偿.
从这一角度来说, 本文提出的全资源预测控制策略通过充分使用分配给的计算资源(而并不要求更多), 可以实现至少不差于传统控制所带来的控制性能(使$u(k|k-\chi_k^*)=u(k-\chi_k^*|k-\chi_k^*) ~\mbox{或} ~0~$), 但通过合理设计式(8)中的前向预测控制序列$U(k|k)$, 提供了提升控制系统性能的自由度和可能性.
另外, 式(8)中的前向预测控制序列$U(k|k)$仅是一种一般形式, 针对具体形式和目标的系统还需要具体的设计.事实上, 只要可以产生所需的前向控制预测序列, 它允许所有可能的控制器设计方法.例如, 在第3节给出了一种基于改进的模型预测控制的设计方法.进一步, 从$U(k|k)$的一般性上说, 所提出的全资源预测控制策略可用于任意受控对象, 而并不局限于线性系统模型, 本文采用线性对象仅为了后续系统分析和具体前向预测控制序列实现的方便.
2.4 闭环系统稳定性
采用算法1的控制系统所使用的控制信号由式(8)和式(9)给出.但是, 式(8)中预测控制序列$U(k|k)$的设计是一个开放问题, 并没有事先规定任何特定形式, 因此闭环系统的具体形式和相应的闭环稳定性分析就难以一般性的给出.另一方面, 控制系统的可用计算资源由上层调度机构给出, 很多情况下可能无法事先估计, 这导致了对可用计算资源(进而$\chi_k^*$)定量描述的缺乏, 也使得一般性的闭环系统稳定性分析较难进行.
为了能够探讨闭环系统的稳定性, 考虑一种简单情况.首先假设预测控制序列$U(k|k)$采用状态反馈的情况.此时实际使用的控制信号$u(k|k-\chi_k^*)$也具有状态反馈的形式, 即
$ \begin{align} u(k|k-\chi_k^*) = K_{\chi_k^*} x(k-\chi_k^*) \end{align} $
(13) 其中, 状态反馈增益$K_{\chi_k^*}$显式依赖$\chi_k^*$.
闭环系统进而具有如下形式:
$ \begin{align} x(k+1) = Ax(k) + BK_{\chi_k^*} x(k-\chi_k^*) \end{align} $
(14) 其次, 有理由假设分配给控制系统的计算资源不是一直缺乏的, 那么存储在寄存器中的前向预测控制序列就总可以在有限时间内得到更新.即存在某上界$N_R$, 使得$0\le \chi_k^* \le N_R$.进而, 可构建增广系统状态$X(k)$为
$ \begin{align} X(k) = [x(k) ~ x(k+1) ~\cdots ~x(k+N_R)]^{\rm T} \end{align} $
(15) 闭环系统(14)的增广形式可写为
$ \begin{align}\label{augumented-closed-loop} X(k+1) = \Lambda_{\chi_k^*} X(k) \end{align} $
(16) 其中, $\Lambda_{\chi_k^*}$由$A$, $B$, $\chi_k^*$和$K_{\chi_k^*}$等参数确定.
将上述系统视为一般的切换系统, 则可以利用标准切换系统方法在李雅普诺夫理论和线性矩阵不等式框架下得到其稳定性条件.该理论方法已经较为成熟, 在相关文献中有更多的技术细节.
需要指出的是, 上述稳定性分析基于对控制律的状态反馈形式假设, 并不具有普适性, 所得结论也很难对前向控制预测序列的设计提供好的指导意义, 实际应用中闭环系统的稳定性尚需在具体的控制律设计方法给出后另行具体分析.不过, 尽管在此并未给出闭环系统稳定的一般结论, 但由前面两小节的分析可以看出, $U(k|k)$设计所带来的自由度可保证采用算法1的控制系统性能(也包括闭环稳定性)永远不差于常规方法.如何针对具体系统进行具体设计将是后续的主要工作.
3. 基于改进模型预测控制的前向控制预测序列计算
如前所述, 前向预测控制序列$U(k|k)$需针对所考虑系统具体化.针对本文所考虑系统形式(1), 本节提出一种改进的模型预测控制设计方法作为$U(k|k)$的一种具体实现方法.经典模型预测控制是一种有限时域优化控制方法, 通过优化有限步长的系统轨迹来求取一串控制序列, 然后将其中的第1步用于实际控制.通过在每一步的滚动优化, 模型预测控制方法可以处理噪声、不确定性和其他限制[21].
模型预测控制的目标函数一般定义如下:
$ \begin{align}\label{obj} J_{k}(N) = &\ X^{\rm T}(k|k)Q(N)X(k|k)\, + \nonumber\\[1mm] &\ U^{\rm T}(k|k)R(N) U(k|k) \end{align} $
(17) 其中, $J_{k}(N)$是$k$时刻的目标函数, 定义于式(8)的$U(k|k)$ $=$ $[u(k|k)~ \cdots~ u(k+N-1|k)]^{\rm T}$是需确定的控制预测序列, $X(k|k)=[x(k+1|k) ~\cdots~ x(k+N|k)]^{\rm T}$是预测状态轨迹, $Q(N)$和$R(N)$是权重矩阵, $N$为预测时域.
上述优化问题的计算复杂性主要决定于预测时域$N$.记解决预测时域为$N$的优化问题所需的最小计算资源为$R_{J_N}$, 式(7)中所定义的函数$f(\cdot)$可对目标函数(17)进一步具体化如下:
$ \begin{align}\label{mapSpecifiedMPC} f(R(k))=N_k, ~~~\mbox{若}~ R_{J_{N_k}}\le R(k)<R_{J_{N_k+1}} \end{align} $
(18) 依照式(18), 在如下讨论中将以$N_k$指代控制系统的可用计算资源$R(k)$.
从式(18)得到的具体形式的$f(\cdot)$意味着带目标函数(17)和$N=N_k$的优化问题可以在时刻$k$使用可用计算资源求解.为解决优化问题, $k$时刻的预测状态可递归求解如下:
$ \begin{align*} x(k+j|k)= A^{j}x(k) + \sum\limits_{l=0}^{j-1}A^{j-l-1}B u(k+l|k) \end{align*} $
定义$E(N_k)=[A^{\rm T} \cdots (A^{N_k})^{\rm T}]^{\rm T}$, $F(N_k)$是$N_k$ $\times$ $N_k$下三角矩阵块, 其非零项定义为$F(N_k)_{ij}=A^{i-j}B$, $j\le i $.则上式的向量形式可表示为
$ \begin{equation} X(k|k)=E(N_k)x(k)+F(N_k)U(k|k) \label{Yk} \end{equation} $
(19) 最优的前向控制预测序列可以通过将式(19)代入式(17)并最小化$J_k(N_k)$得到, 其解具有如下的状态反馈形式:
$ \begin{align}\label{FCS-MPC} U(k|k)=K(N_k)x(k) \end{align} $
(20) 其中,
$ \begin{align*} K(N_k)=&-(F^{\rm T}(N_k)Q(N_k)F(N_k)\, +\\[1mm] &\ R(N_k))^{-1}F^{\rm T}(N_k)Q(N_k)E(N_k) \end{align*} $
使用如上的改进模型预测控制方法进行全资源预测控制的步骤总结为算法2.
算法2.基于改进模型预测控制的全资源预测控制算法
步骤1. 传感器采样系统输出并将数据传输至控制器;
步骤2. 基于分配的计算资源$R(k)$, 依照式(18)决定前向控制预测序列的步长$N_k$, 通过式(20)计算前向控制预测序列并将之传至执行器;
步骤3. 控制选择器更新其寄存器并依照式(9)选择合适的控制量并令执行器应用至系统.
注6. 本节中的改进模型预测控制算法在两个方面不同于传统模型预测控制.首先, 算法2中的预测时域是时变的并决定于可用计算资源$R(k)$.其次, 传统模型预测控制只使用所计算的控制序列中的第一个, 而算法2却利用了所有可能的前向控制预测.
4. 仿真算例
本节通过如图 3所示的小车-倒立摆系统对所提出的全资源预测控制策略和算法进行验证.该系统可建模为式(1)中的线性系统, 其系统矩阵如下:
$ \begin{align*} &A=\left( \begin{array}{cccc} 0 & 1 &0 &0 \\ 0 & 0 & -m\times\dfrac{{\rm g}}{M} & 0\\[2mm] 0 & 0 & 0 &1 \\ 0 & 0 & (M+m)\times \dfrac{{\rm g}}{Ml} & 0 \end{array} \right) \\[2mm] &B=\left( \begin{array}{c} 1 \\ \dfrac{1}{M}\\[2mm] 0\\ -\dfrac{1}{Ml} \end{array} \right) \end{align*} $
其中, 状态变量$x(k)=[x_1(k) ~ x_2(k) ~ x_3(k) ~ x_4(k)]^{\rm T}$且$x_1(k)$为小车的位移, $x_2(k)$为小车的速度, $x_3(k)$为倒立摆与垂直方向的夹角或偏离角度, $x_4(k)$为刚性摆杆的角速度.各物理量都具有相应的单位, 其值如下:小球的质量$m=1$, 小车的质量$M=10$, 刚性摆长度$l=10$, 重力加速度${\rm g}=9.8$.
为了验证提出的控制策略的有效性, 考虑两种情况的比较, 两种情况都使用第3节提出的模型预测控制器, 但一种使用第2节提出的全资源预测控制策略, 另一种则不使用.也就是说, 前者可利用存储在前向控制预测序列中的预测控制量, 而后者在可用计算资源无法进行控制计算的时候, 即$R(k)$ $<$ $R^*_u$时, 只能使用零控制, 而在其他时刻则如传统模型预测控制一样使用预测序列的第1个控制量.
首先进行基于MATLAB的数值仿真, 进而采用MATLAB/LabVIEW相结合的联合仿真方法进行验证.
例1.基于MATLAB的数值仿真
将计算资源用整数表示, 并设处理器的最大计算资源$R^*=100$.设控制系统进行1~7步控制预测所需的最小计算资源如下:
$ \begin{align}\label{mapCompExmp} [R_1^* ~ R_2^* ~ R_3^* ~ R_4^* ~ R_5^* ~ R_6^* ~ R_7^*] = [25 ~ 31~ 40 ~ 51~ 68~ 84 ~ 105] \end{align} $
(21) 按定义, $R_u^*=R_1^*$, 并注意到$R_6^*<R^*<R_7^*$, 即处理器的最大计算资源可进行6步控制预测序列的计算.式(18)中的函数$f(\cdot)$可按上式随之定义.
控制系统的可用计算资源$R(k)$假设为某一马尔科夫过程, 并具有较高的概率小于$R^*_u$ (见式(5))或接近$R^*$ (见式(6)).在实际中, 已分配计算资源的突然大幅度切换较为少见, 因此在下一时刻控制系统的可用计算资源有较高概率仍保持在原有或相近水平.在$R(k)$的转移概率矩阵中本文也将此特性考虑在内.
在上述系统设置下, 控制系统可用计算资源$R(k)$的典型时间演化可见图 4 (a).由式(21)可知, 前向控制预测序列的长度$N_k$可由计算资源$R(k)$来确定, 如图 4 (b)所示.另外, 从图 4 (c)中$\chi^*_k$的演化可看出, 可用计算资源经常会太少不足以计算任何控制量, 这使得传统控制方法无法在这种情况下工作.
图 4 MATLAB仿真的$R(k)$, $N_k$和$\chi_k$随时间的典型演化曲线Fig. 4 Typical evolution curves of $R(k)$, $N_k$ and $\chi_k$ in MATLAB simulation图 5 (a)的系统响应支持本文期望的结论:在不增加任何计算消耗的前提下, 使用我们的控制策略的系统状态响应更快的到达稳定, 并较少波动. 图 5 (b)也显示采用我们的控制策略的实际控制量也更为平滑.
图 5 MATLAB仿真的系统响应和控制量曲线Fig. 5 The system responses and control signals in MATLAB simulation例2. MATLAB/LabVIEW联合仿真
本例仍考虑图 3中的小车-倒立摆系统, 但使用MATLAB/LabVIEW联合仿真的方法来验证我们提出的全资源预测控制策略和算法的有效性.系统实现的架构见图 6.其中受控对象即小车-倒立摆系统在LabVIEW中实现, 而控制部分则在MATLAB中实现, 二者通过LabVIEW中的MATLAB脚本调用实现联合仿真. LabVIEW中的小车-倒立摆系统添加了一个很小的外部扰动以模拟真实情况.计算资源的刻画则以MATLAB的控制器程序运行时间为标准.因此, 与仅基于MATLAB的纯数值仿真相比, 这一联合仿真方案具有如下优点: 1) LabVIEW中实现的小车-倒立摆系统更为真实的模拟了受控对象; 2)计算资源的具体定量描述也更为真实, 因此可更有效地验证本文中提出的控制策略和算法.
控制器执行所需要的计算资源通过MATLAB重复实验并归一化, 结果如下:
$ \begin{align*} [R_1^* ~ R_2^* ~ R_3^* ~ R_4^* ~ R_5^* ~ R_6^* ~ R_7^*] = [23 ~ 30~ 38 ~ 49~ 65~ 80 ~ 100] \end{align*} $
与例1类似, 同样使用两种方法进行比较, 一种使用我们的全资源预测策略, 另一种不使用.系统响应证明了全资源预测控制策略可以保证倒立摆在垂直方向的稳定性(图 7 (a)), 而不使用我们的策略, 即便采用同样的控制器设计, 也无法保证倒立摆的稳定性(图 7 (b)).
图 7 MATLAB和LabVIEW联合仿真的系统响应Fig. 7 The system responses using MATLAB and LabVIEW co-simulation5. 结论
通过对可用计算资源的充分利用, 在不增加额外计算资源要求的前提下实现了计算受限控制系统的性能提升.这是一种控制和计算的联合设计方法, 对该方法的进一步改进将依赖于前向预测控制序列的更为有效的设计, 另外对闭环系统稳定性的保证也是后续的研究工作之一.
-
图 1 WoS数据库中相关期刊论文数量分布(2010 ~ 2019)
Fig. 1 Number distribution of related journal article publications indexed by WoS (2010 ~ 2019)
图 10 基于自动编码器的非侵入式负荷监测网络结构图
Fig. 10 Network structure diagram of NILM based on automatic encoder
表 1 NILM方法优缺点对比
Table 1 Comparison of NILM methods
方法 优点 缺点 HMM 模型直观 运算量大 精确推理困难 GSP 表征能力强 现有算法相对较少 训练周期短 运算复杂度较低 ML 自动提取特征 模型参数多 分解准确率高 训练所需数据量大 泛化性能好 可解释性差 
下载: 导出CSV
表 2 非侵入式负荷监测公开数据集
Table 2 Publicly available datasets for NILM
数据集 地点 采集持续时间 房屋
数量传感器数量/
房屋采集频率 采集参数1 其他数据 REDD[9] 美国 几天 ~ 数月 6 10 ~ 24 15 kHz (Agg);
0.5 Hz, 1 Hz (Sub)V, P (Agg); P (Sub) BERDS[149] 美国 1 年 1 NA 20 s P, Q, S 气候数据 BLUED[150] 美国 8 天 1 1 12 kHz (Agg);
20 Hz (Sub)I, V (P通过计算
得出@60 Hz)各设备的状态转换标签 Smart* Home[151] 美国 3 个月 3 A : 26;
B, C : 211 Hz P, S (Agg); P (Sub) 太阳能和风力发电数据, 气候,
室内温湿度数据DRED[152] 荷兰 6 个月 3 12 1 Hz E, P (Agg & Sub) 室内外温度, 风速, 降水, 入住率,
房屋布局, 设备配置, 无线信号Tracebase[153] 德国 1883 天
(累计)15 NA 1 s, 8 s (Sub) P (Sub) 用于设备识别, 未采集总表数据 AMPds2[21, 154] 加拿大 2 年 1 21 1 min V, I, F, P, Q, S,
F, E 等 10 项水表、天然气表数据,
电费账单数据UK-DALE[155] 英国 2 个月 ~ 4.3 年 5 5 ~ 54 16 kHz (I, V of Agg); 6 s (Agg & Sub);
1 s (Agg)P, I, V 设备状态切换信息,
住户人员构成
及能源构成信息iAWE[156] 印度 73 天 1 33 1 Hz (Agg);
1 s, 6 s (Sub)V, I, F, P, ph 用水量和环境数据 (温度, 人员
活动, 声音及无线信号强度)REFIT[157] 英国 2 年 20 10 8 s P 天然气表和环境数据 GREEND[159] 意大利/
奥地利1 年 9 9 1 Hz P 用电负荷配置, 住户情况描述 ECO[34] 瑞士 8 个月 6 6 ~ 10 1 Hz P, Q 住户情况描述 PLAID[160] 美国 — 56 共 11 类, 大于 200 个设备 30 kHz I, V EMBED[161] 美国 14 ~ 27 天 3 共 21 类, 约
40 个设备12 kHz (Agg);
1 ~ 2 Hz (Sub)I, V, P, Q, F 各设备的状态转换标签, 入住率 1 Agg: 总表; Sub: 分表; P: 有功功率; Q: 无功功率; S: 视在功率; E: 电量; F: 频率; V: 电压; I: 电流; ph: 相位.
下载: 导出CSV
-
[1] Hart G W. Nonintrusive appliance load monitoring. Proceedings of the IEEE, 1992, 80(12): 1870-1891 doi: 10.1109/5.192069 [2] Batra N, Singh A, Whitehouse K. If you measure it, can you improve it? Exploring the value of energy disaggregation. In: Proceedings of the 2nd ACM International Conference on Embedded Systems for Energy-Efficient Built Environments. Seoul, South Korea: ACM, 2015. 191−200 [3] Bergés M, Goldman E, Matthews H S, Soibelman L. Enhancing electricity audits in residential buildings with nonintrusive load monitoring. Journal of Industrial Ecology, 2010, 14(5): 844-858 doi: 10.1111/j.1530-9290.2010.00280.x [4] Papadopoulos P M, Reppa V, Polycarpou M M, Panayiotou C G. Scalable distributed sensor fault diagnosis for smart buildings. IEEE/CAA Journal of Automatica Sinica, 2020, 7(3): 638-655 doi: 10.1109/JAS.2020.1003123 [5] Qin J H, Wan Y N, Yu X H, Li F Y, Li C J. Consensus-based distributed coordination between economic dispatch and demand response. IEEE Transactions on Smart Grid, 2019, 10(4): 3709-3719 doi: 10.1109/TSG.2018.2834368 [6] Wang Y, Chen Q X, Hong T, Kang C Q. Review of smart meter data analytics: Applications, methodologies, and challenges. IEEE Transactions on Smart Grid, 2019, 10(3): 3125-3148 doi: 10.1109/TSG.2018.2818167 [7] Mohassel R R, Fung A, Mohammadi F, Raahemifar K. A survey on advanced metering infrastructure. International Journal of Electrical Power and Energy Systems, 2014, 63: 473-484 [8] Armel K C, Gupta A, Shrimali G, Albert A. Is disaggregation the holy grail of energy efficiency? The case of electricity. Energy Policy, 2013, 52: 213-234 doi: 10.1016/j.enpol.2012.08.062 [9] Kolter J Z, Johnson M J. Redd: a public data set for energy disaggregation research. In: Proceedings of the 1st SustKDD Workshop on Data Mining Applications in Sustainability. San Diego, USA: ACM, 2011. 1−6 [10] Batra N, Parson O, Bergés M, Singh A, Rogers A. A comparison of non-intrusive load monitoring methods for commercial and residential buildings. arXiv: 1408.6595, 2014. [11] Zeifman M, Roth K. Nonintrusive appliance load monitoring: Review and outlook. IEEE Transactions on Consumer Electronics, 2011, 57(1): 76-84 doi: 10.1109/TCE.2011.5735484 [12] Kelly J, Knottenbelt W. Neural NILM: Deep neural networks applied to energy disaggregation. In: Proceedings of the 2nd ACM International Conference on Embedded Systems for Energy-Efficient Built Environments. Seoul, South Korea: ACM, 2015. 55−64 [13] Zhang C Y, Zhong M J, Wang Z Z, Goddard N, Sutton C. Sequence-to-point learning with neural networks for non-intrusive load monitoring. In: Proceedings of the 2017 AAAI Conference on Artificial Intelligence. New Orleans, USA: AAAI, 2017. [14] Calamaro N, Donko M, Shmilovitz D. A highly accurate nilm: with an electro-spectral space that best fits algorithm's national deployment requirements. Energies, 2021, 14(21): 7410 [15] Zeifman M, Roth K. Viterbi algorithm with sparse transitions (vast) for nonintrusive load monitoring. In: Proceedings of the 2011 IEEE Symposium on Computational Intelligence Applications in Smart Grid (CIASG). Paris, France: IEEE, 2011. 1−8 [16] Nakajima H, Nagasawa K, Shishido Y, Kagiya Y, Takagi Y. The state estimation of existing home appliances using signal analysis technique. In: Proceedings of the 2014 SICE Annual Conference (SICE). Sapporo, Japan: IEEE, 2014. 1247−1252 [17] Makonin S, Popowich F, Bajić I V, Gill B, Bartram L. Exploiting hmm sparsity to perform online real-time nonintrusive load monitoring. IEEE Transactions on Smart Grid, 2016, 7(6): 2575-2585 doi: 10.1109/TSG.2015.2494592 [18] Nguyen T K, Dekneuvel E, Jacquemod G, Nicolle B, Zammit O, Nguyen V C. Development of a real-time non-intrusive appliance load monitoring system: An application level model. International Journal of Electrical Power and Energy Systems, 2017, 90: 168-180 [19] Gisler C, Ridi A, Zufferey D, Khaled O A, Hennebert J. Appliance consumption signature database and recognition test protocols. In: Proceedings of the 8th International Workshop on Systems, Signal Processing and their Applications (WoSSPA). Algiers, Algeria: IEEE, 2013. 336−341 [20] Mei J, He D W, Harley R G, Habetler T G. Random forest based adaptive non-intrusive load identification. In: Proceedings of the 2014 International Joint Conference on Neural Networks (IJCNN). Beijing, China: IEEE, 2014. 1978−1983 [21] Makonin S, Ellert B, Bajić I V, Popowich F. Electricity, water, and natural gas consumption of a residential house in Canada from 2012 to 2014. Scientific Data, 2016, 3: Article No. 160037 doi: 10.1038/sdata.2016.37 [22] Hart G W. Prototype Nonintrusive Appliance Load Monitor, Technical Report RP. 2568-2, MIT Energy Laboratory and EPRI, USA, 1985, [23] Simpson C D. Principles of Electronics. Upper Saddle River: Prentice Hall, 1996. 85−102 [24] Lee K D, Leeb S B, Norford L K, Armstrong P R, Holloway J, Shaw S R. Estimation of variable-speed-drive power consumption from harmonic content. IEEE Transactions on Energy Conversion, 2005, 20(3): 566-574 doi: 10.1109/TEC.2005.852963 [25] Wichakool W, Avestruz A T, Cox R W, Leeb S B. Modeling and estimating current harmonics of variable electronic loads. IEEE Transactions on Power Electronics, 2009, 24(12): 2803-2811 doi: 10.1109/TPEL.2009.2029231 [26] Gupta S, Reynolds M S, Patel S N. ElectriSense: Single-point sensing using EMI for electrical event detection and classification in the home. In: Proceedings of the 12th ACM International Conference on Ubiquitous Computing. Copenhagen, Denmark: ACM, 2010. 139−148 [27] Tsai M S, Lin Y H. Modern development of an adaptive non-intrusive appliance load monitoring system in electricity energy conservation. Applied Energy, 2012, 96: 55-73 doi: 10.1016/j.apenergy.2011.11.027 [28] Ahmadi H, Martí J R. Load decomposition at smart meters level using eigenloads approach. IEEE Transactions on Power Systems, 2015, 30(6): 3425-3436 doi: 10.1109/TPWRS.2014.2388193 [29] Leeb S B, Shaw S R, Kirtley J L. Transient event detection in spectral envelope estimates for nonintrusive load monitoring. IEEE Transactions on Power Delivery, 1995, 10(3): 1200-1210 doi: 10.1109/61.400897 [30] Baranski M, Voss J. Nonintrusive appliance load monitoring based on an optical sensor. In: Proceedings of the 2003 IEEE Bologna Power Tech Conference. Bologna, Italy: IEEE, 2003. Article No. 8 [31] Klemenjak C, Egarter D, Elmenreich W. YoMo: The Arduino-based smart metering board. Computer Science-Research and Development, 2016, 31(1-2): 97-103 doi: 10.1007/s00450-014-0290-8 [32] Makonin S, Sung W, Cruz R D, Yarrow B, Gill B, Popowich F, et al. Inspiring energy conservation through open source metering hardware and embedded real-time load disaggregation. In: Proceedings of the 2013 IEEE PES Asia-Pacific Power and Energy Engineering Conference (APPEEC). Hong Kong, China: IEEE, 2013. 1−6 [33] Batra N, Dutta H, Singh A. INDiC: Improved non-intrusive load monitoring using load division and calibration. In: Proceedings of the 12th International Conference on Machine Learning and Applications. Miami, USA: IEEE, 2013. 79−84 [34] Beckel C, Kleiminger W, Cicchetti R, Staake T, Santini S. The eco data set and the performance of non-intrusive load monitoring algorithms. In: Proceedings of the 1st ACM Conference on Embedded Systems for Energy-Efficient Buildings. Memphis, USA: ACM, 2014. 80−89 [35] Powers J T, Margossian B, Smith B A. Using a rule-based algorithm to disaggregate end-use load profiles from premise-level data. IEEE Computer Applications in Power, 1991, 4(2): 42-47 doi: 10.1109/67.75875 [36] Bergés M, Goldman E, Matthews H S, Soibelman L. Training load monitoring algorithms on highly sub-metered home electricity consumption data. Tsinghua Science and Technology, 2008, 13(S1): 406-411 [37] Anderson K D, Bergés M, Ocneanu A, Benitez D, Moura J M F. Event detection for non intrusive load monitoring. In: Proceedings of the 38th Annual Conference on IEEE Industrial Electronics Society. Montreal, Canada: IEEE, 2012. 3312−3317 [38] Rudin L I, Osher S, Fatemi E. Nonlinear total variation based noise removal algorithms. Physica D: Nonlinear Phenomena, 1992, 60(1-4): 259-268 doi: 10.1016/0167-2789(92)90242-F [39] Laughman C, Lee K, Cox R, Shaw S, Leeb S, Norford L, et al. Power Signature Analysis. IEEE Power and Energy Magazine, 2003, 1(2): 56-63 doi: 10.1109/MPAE.2003.1192027 [40] Shaw S R, Leeb S B, Norford L K, Cox R W. Nonintrusive load monitoring and diagnostics in power systems. IEEE Transactions on Instrumentation and Measurement, 2008, 57(7): 1445-1454 doi: 10.1109/TIM.2008.917179 [41] Peach N, Basseville M, Nikiforov I V. Detection of abrupt changes: Theory and applications. Journal of the Royal Statistical Society Series A-Statistics in Society, 1995, 158(1): 185-186 doi: 10.2307/2983416 [42] Luo D, Norford L, Shaw S R, Leeb S B. Monitoring HVAC equipment electrical loads from a centralized location-methods and field test results. ASHRAE Transactions, 2002, 108(1): 841-857 [43] Jazizadeh F. Building energy monitoring realization: Context-aware event detection algorithms for non-intrusive electricity disaggregation. In: Proceedings of the 2016 American Society of Civil Engineers Construction Research Congress. San Juan, Puerto Rico, USA: ASCE, 2016. 839−848 [44] Jin Y W, Tebekaemi E, Bergés M, Soibelman L. Robust adaptive event detection in non-intrusive load monitoring for energy aware smart facilities. In: Proceedings of the 2011 IEEE International Conference on Acoustics, Speech and Signal Processing. Prague, Czech Republic: IEEE, 2011. 4340−4343 [45] Yang C C, Soh C S, Yap V V. Comparative study of event detection methods for non-intrusive appliance load monitoring. Energy Procedia, 2014, 61: 1840-1843 doi: 10.1016/j.egypro.2014.12.225 [46] De Baets L, Ruyssinck J, Develder C, Dhaene T, Deschrijver D. On the Bayesian optimization and robustness of event detection methods in NILM. Energy and Buildings, 2017, 145: 57-66 doi: 10.1016/j.enbuild.2017.03.061 [47] Zhu Z C, Wei Z Q, Yin B, Zhang S, Wang X. A novel approach for event detection in non-intrusive load monitoring. In: Proceedings of the 2017 IEEE Conference on Energy Internet and Energy System Integration (EI2). Beijing, China: IEEE, 2017. 1−5 [48] Afzalan M, Jazizadeh F, Wang J. Self-configuring event detection in electricity monitoring for human-building interaction. Energy and Buildings, 2019, 187: 95-109 doi: 10.1016/j.enbuild.2019.01.036 [49] Bhattacharjee S, Kumar A, RoyChowdhury J. Appliance classification using energy disaggregation in smart homes. In: Proceedings of the 2014 International Conference on Computation of Power, Energy, Information and Communication (ICCPEIC). Chennai, India: IEEE, 2014. 1−6 [50] Chou P A, Chang R I. Unsupervised adaptive non-intrusive load monitoring system. In: Proceedings of the 2013 IEEE International Conference on Systems, Man, and Cybernetics. Manchester, UK: IEEE, 2013. 3180−3185 [51] Egarter D, Bhuvana V P, Elmenreich W. PALDi: Online load disaggregation via particle filtering. IEEE Transactions on Instrumentation and Measurement, 2015, 64(2): 467-477 doi: 10.1109/TIM.2014.2344373 [52] Figueiredo M, Ribeiro B, De Almeida A. Electrical signal source separation via nonnegative tensor factorization using on site measurements in a smart home. IEEE Transactions on Instrumentation and Measurement, 2014, 63(2): 364-373 doi: 10.1109/TIM.2013.2278596 [53] Gabaldón A, Ortiz-García M, Molina R, Valero-Verdú S. Disaggregation of the electric loads of small customers through the application of the Hilbert transform. Energy Efficiency, 2014, 7(4): 711-728 doi: 10.1007/s12053-013-9250-6 [54] Kramer O, Klingenberg T, Sonnenschein M, Wilken O. Non-intrusive appliance load monitoring with bagging classifiers. Logic Journal of the IGPL, 2015, 23(3): 359-368 doi: 10.1093/jigpal/jzv016 [55] Wang X J, Lei D M, Yong J, Zeng L Q, West S. An online load identification algorithm for non-intrusive load monitoring in homes. In: Proceedings of the 8th International Conference on Intelligent Sensors, Sensor Networks and Information. Melbourne, Australia: IEEE, 2013. 1−6 [56] Wang Z Y, Zheng G L. Residential appliances identification and monitoring by a nonintrusive method. IEEE Transactions on Smart Grid, 2012, 3(1): 80-92 doi: 10.1109/TSG.2011.2163950 [57] Racines D, Candelo J E, Montaña J. Non-intrusive electrical load monitoring system applying neural networks with combined steady-state electrical Variables. Tehnicki Vjesnik, 2018, 25(5): 1321-1329 [58] Lu L L, Park S W, Wang B H. Electric load signature analysis for home energy monitoring system. International Journal of Fuzzy Logic and Intelligent Systems, 2012, 12(3): 193-197 doi: 10.5391/IJFIS.2012.12.3.193 [59] Basu K, Debusschere V, Bacha S. Load identification from power recordings at meter panel in residential households. In: Proceedings of the 2012 International Conference on Electrical Machines. Marseille, France: IEEE, 2012. 2098−2104 [60] Tina G M, Amenta V A, Tomarchio O, Di Modica G. Web interactive non intrusive load disaggregation system for active demand in smart grids. EAI Endorsed Transactions on Energy, 2014, 1(3): 1-9 doi: 10.4108/ew.1.3.e1 [61] Hassan T, Javed F, Arshad N. An empirical investigation of V-I trajectory based load signatures for non-intrusive load monitoring. IEEE Transactions on Smart Grid, 2014, 5(2): 870-878 doi: 10.1109/TSG.2013.2271282 [62] Du L, He D W, Harley R G, Habetler T G. Electric load classification by binary voltage-current trajectory mapping. IEEE Transactions on Smart Grid, 2016, 7(1): 358-365 doi: 10.1109/TSG.2015.2442225 [63] Wang A L, Chen B X, Wang C G, Hua D D. Non-intrusive load monitoring algorithm based on features of V–I trajectory. Electric Power Systems Research, 2018, 157: 134-144 doi: 10.1016/j.jpgr.2017.12.012 [64] Baptista D, Mostafa S S, Pereira L, Sousa L, Morgado-Dias F. Implementation strategy of convolution neural networks on field programmable gate arrays for appliance classification using the voltage and current (V-I) trajectory. Energies, 2018, 11(9): Article No. 2460 doi: 10.3390/en11092460 [65] Yang C C, Soh C S, Yap V V. A systematic approach in load disaggregation utilizing a multi-stage classification algorithm for consumer electrical appliances classification. Frontiers in Energy, 2019, 13(2): 386-398 doi: 10.1007/s11708-017-0497-z [66] Liu Y C, Wang X, You W. Non-intrusive load monitoring by voltage-current trajectory enabled transfer learning. IEEE Transactions on Smart Grid, 2019, 10(5): 5609-5619 doi: 10.1109/TSG.2018.2888581 [67] Wild B, Barsim K S, Yang B. A new unsupervised event detector for non-intrusive load monitoring. In: Proceedings of the 2015 Global Conference on Signal and Information Processing (GlobalSIP). Orlando, USA: IEEE, 2015. 73−77 [68] He D W, Lin W X, Liu N, Harley R G, Habetler T G. Incorporating non-intrusive load monitoring into building level demand response. IEEE Transactions on Smart Grid, 2013, 4(4): 1870-1877 doi: 10.1109/TSG.2013.2258180 [69] Du L, Restrepo J A, Yang Y, Harley R G, Habetler T G. Nonintrusive, self-organizing, and probabilistic classification and identification of plugged-in electric loads. IEEE Transactions on Smart Grid, 2013, 4(3): 1371-1380 doi: 10.1109/TSG.2013.2263231 [70] Chang H H, Lin L S, Chen N M, Lee W J. Particle-swarm-optimization-based nonintrusive demand monitoring and load identification in smart meters. IEEE Transactions on Industry Applications, 2013, 49(5): 2229-2236 doi: 10.1109/TIA.2013.2258875 [71] Bouhouras A S, Gkaidatzis P A, Chatzisavvas K C, Panagiotou E, Poulakis N, Christoforidis G C. Load signature formulation for non-intrusive load monitoring based on current measurements. Energies, 2017, 10(4): Article No. 538 doi: 10.3390/en10040538 [72] Jimenez Y, Duarte C, Petit J, Meyer J, Schegner P, Carrrillo G. Steady state signatures in the time domain for nonintrusive appliance identification. Ingeniería e Investigación, 2015, 35(Sl): 58-64 [73] Chang H H, Lian K L, Su Y C, Lee W J. Power-spectrum-based wavelet transform for nonintrusive demand monitoring and load identification. IEEE Transactions on Industry Applications, 2014, 50(3): 2081-2089 doi: 10.1109/TIA.2013.2283318 [74] Chen K L, Chang H H, Chen N M. A new transient feature extraction method of power signatures for Nonintrusive Load Monitoring Systems. In: Proceedings of the 2013 IEEE International Workshop on Applied Measurements for Power Systems (AMPS). Aachen, Germany: IEEE, 2013. 79−84 [75] Gillis J M, Morsi W G. Non-intrusive load monitoring using semi-supervised machine learning and wavelet design. IEEE Transactions on Smart Grid, 2017, 8(6): 2648-2655 doi: 10.1109/TSG.2016.2532885 [76] Su Y C, Lian K L, Chang H H. Feature selection of non-intrusive load monitoring system using STFT and wavelet transform. In: Proceedings of the 8th International Conference on e-Business Engineering. Beijing, China: IEEE, 2011. 293−298 [77] Jazizadeh F, Becerik-Gerber B, Bergés M, Soibelman L. An unsupervised hierarchical clustering based heuristic algorithm for facilitated training of electricity consumption disaggregation systems. Advanced Engineering Informatics, 2014, 28(4): 311-326 doi: 10.1016/j.aei.2014.09.004 [78] Jiang L, Luo S H, Li J M. Intelligent electrical appliance event recognition using multi-load decomposition. Advanced Materials Research, 2013, 805-806: 1039-1045 doi: 10.4028/www.scientific.net/AMR.805-806.1039 [79] Borin V P, Barriquello C H, Campos A. Approach for home appliance recognition using vector projection length and Stockwell transform. Electronics Letters, 2015, 51(24): 2035-2037 doi: 10.1049/el.2015.2385 [80] Jimenez Y, Duarte C, Petit J, Carrillo G. Feature extraction for nonintrusive load monitoring based on S-transform. In: Proceedings of the 2014 Clemson University Power Systems Conference. Clemson, USA: IEEE, 2014. 1−5 [81] Cole A I, Albicki A. Data extraction for effective non-intrusive identification of residential power loads. In: Proceedings of the 1998 IEEE Instrumentation and Measurement Technology Conference. St. Paul, USA: IEEE, 1998. 812−815 [82] Patel S N, Robertson T, Kientz J A, Reynolds M S, Abowd G D. At the flick of a switch: Detecting and classifying unique electrical events on the residential power line. In: Proceedings of the 9th International Conference on Ubiquitous Computing. Innsbruck, Austria: Springer, 2007. 271−288 [83] Lin Y H, Tsai M S. Development of an improved time–frequency analysis-based nonintrusive load monitor for load demand identification. IEEE Transactions on Instrumentation and Measurement, 2014, 63(6): 1470-1483 doi: 10.1109/TIM.2013.2289700 [84] Norford L K, Leeb S B. Non-intrusive electrical load monitoring in commercial buildings based on steady-state and transient load-detection algorithms. Energy and Buildings, 1996, 24(1): 51-64 doi: 10.1016/0378-7788(95)00958-2 [85] Chang H H. Non-intrusive demand monitoring and load identification for energy management systems based on transient feature analyses. Energies, 2012, 5(11): 4569-4589 doi: 10.3390/en5114569 [86] Meehan P, McArdle C, Daniels S. An efficient, scalable time-frequency method for tracking energy usage of domestic appliances using a two-step classification algorithm. Energies, 2014, 7: 7041-7066 doi: 10.3390/en7117041 [87] Liang J, Ng S K K, Kendall G, Cheng J W M. Load signature study―part I: Basic concept, structure, and methodology. IEEE Transactions on Power Delivery, 2010, 25(2): 551-560 doi: 10.1109/TPWRD.2009.2033799 [88] Liang J, Ng S K K, Kendall G, Cheng J W M. Load signature study―part II: Disaggregation framework, simulation, and applications. IEEE Transactions on Power Delivery, 2010, 25(2): 561-569 doi: 10.1109/TPWRD.2009.2033800 [89] Gulati M, Ram S S, Singh A. An in depth study into using EMI signatures for appliance identification. In: Proceedings of the 1st ACM Conference on Embedded Systems for Energy-Efficient Buildings. Memphis, USA: ACM, 2014. 70−79 [90] Popescu F, Enache F, Vizitiu I C, Ciotirnae P. Recurrence plot analysis for characterization of appliance load signature. In: Proceedings of the 10th International Conference on Communications (COMM). Bucharest, Romania: IEEE, 2014. 1−4 [91] Patri O P, Panangadan A V, Chelmis C, Prasanna V K. Extracting discriminative features for event-based electricity disaggregation. In: Proceedings of the 2014 IEEE Conference on Technologies for Sustainability. Portland, USA: IEEE, 2014. 232−238 [92] Du L, Yang Y, He D W, Harley R G, Habetler T G. Feature extraction for load identification using long-term operating waveforms. IEEE Transactions on Smart Grid, 2015, 6(2): 819-826 doi: 10.1109/TSG.2014.2373314 [93] Chang H H, Chen K L, Tsai Y P, Lee W J. A new measurement method for power signatures of nonintrusive demand monitoring and load identification. IEEE Transactions on Industry Applications, 2012, 48(2): 764-771 doi: 10.1109/TIA.2011.2180497 [94] Chang H H. Genetic algorithms and non-intrusive energy management system based economic dispatch for cogeneration units. Energy, 2011, 36(1): 181-190 doi: 10.1016/j.energy.2010.10.054 [95] Du Y, Du L, Lu B, Harley R, Habetler T. A review of identification and monitoring methods for electric loads in commercial and residential buildings. In: Proceedings of the 2010 IEEE Energy Conversion Congress and Exposition. Atlanta, USA: IEEE, 2010. 4527−4533 [96] 武昕, 祁兵, 韩璐, 王震, 董超. 基于模板滤波的居民负荷非侵入式快速辨识算法. 电力系统自动化, 2017, 41(2): 135-141 doi: 10.7500/AEPS20160411001Wu Xin, Qi Bing, Han Lu, Wang Zhen, Dong Chao. Fast non-intrusive load identification algorithm for resident load based on template filtering. Automation of Electric Power Systems, 2017, 41(2): 135-141 doi: 10.7500/AEPS20160411001 [97] 程祥, 李林芝, 吴浩, 丁一, 宋永华, 孙维真. 非侵入式负荷监测与分解研究综述. 电网技术, 2016, 40(10): 3108-3117Cheng Xiang, Li Lin-Zhi, Wu Hao, Ding Yi, Song Yong-Hua, Sun Wei-Zhen. A survey of the research on non-intrusive load monitoring and disaggregation. Power System Technology, 2016, 40(10): 3108-3117 [98] Dong M, Meira P C M, Xu W, Freitas W. An event window based load monitoring technique for smart meters. IEEE Transactions on Smart Grid, 2012, 3(2): 787-796 doi: 10.1109/TSG.2012.2185522 [99] Kolter J Z, Jaakkola T. Approximate inference in additive factorial HMMs with application to energy disaggregation. In: Proceedings of the 5th International Conference on Artificial Intelligence and Statistics. Canary Islands, Spain: JMLR, 2012. 1472−1482 [100] Figueiredo M, De Almeida A, Ribeiro B. Home electrical signal disaggregation for non-intrusive load monitoring (NILM) systems. Neurocomputing, 2012, 96: 66-73 doi: 10.1016/j.neucom.2011.10.037 [101] Yang C C, Soh C S, Yap V V. A systematic approach in appliance disaggregation using k-nearest neighbours and naive Bayes classifiers for energy efficiency. Energy Efficiency, 2018, 11(1): 239-259 doi: 10.1007/s12053-017-9561-0 [102] Hidiyanto F, Halim A. Knn methods with varied k, distance and training data to disaggregate nilm with similar load characteristic. In: Proceedings of the 3rd Asia Pacific Conference on Research in Industrial and Systems Engineering (APCORISE 2020). Depok, Indonesia: ACM, 2020. 93−99 [103] Nguyen M, Alshareef S, Gilani A, Morsi W G. A novel feature extraction and classification algorithm based on power components using single-point monitoring for NILM. In: Proceedings of the 28th Canadian Conference on Electrical and Computer Engineering (CCECE). Halifax, Canada: IEEE, 2015. 37−40 [104] Buddhahai B, Wongseree W, Rakkwamsuk P. A non-intrusive load monitoring system using multi-label classification approach. Sustainable Cities and Society, 2018, 39(10): 621-630 [105] Hoyo-Montano J A, León-Ortega N, Valencia-Palomo G, Galaz-Bustamante R A, Espejel-Blanco D F, Vázquez-Palma M G. Non-intrusive electric load identification using wavelet transform. Ingeniería e Investigación, 2018, 38(2): 42-51 [106] Su S, Yan Y T, Lu H, Li K P, Sun Y P, Wang F, et al. Non-intrusive load monitoring of air conditioning using low-resolution smart meter data. In: Proceedings of the 2016 IEEE International Conference on Power System Technology (POWERCON). Wollongong, Australia: IEEE, 2016. 1−5 [107] Singh M, Kumar S, Semwal S, Prasad R S. Residential load signature analysis for their segregation using wavelet—SVM. In: Proceedings of the 2015 Power Electronics and Renewable Energy Systems. New Delhi, India: Springer, 2015. 863−871 [108] Chui K T, Lytras M, Visvizi A. Energy sustainability in smart cities: Artificial intelligence, smart monitoring, and optimization of energy consumption. Energies, 2018, 11(11): Article No. 2869 doi: 10.3390/en11112869 [109] Su D L, Shi Q, Xu H, Wang W. Nonintrusive load monitoring based on complementary features of spurious emissions. Electronics, 2019, 8(9): Article No. 1002 doi: 10.3390/electronics8091002 [110] Chang H H, Lee M C, Lee W J, Chien C L, Chen N M. Feature extraction-based Hellinger distance algorithm for nonintrusive aging load identification in residential buildings. IEEE Transactions on Industry Applications, 2016, 52(3): 2031-2039 doi: 10.1109/TIA.2016.2533487 [111] Lin Y H, Hu Y C. Electrical energy management based on a hybrid artificial neural network-particle swarm optimization-integrated two-stage non-intrusive load monitoring process in smart homes. Processes, 2018, 6(12): Article No. 236 doi: 10.3390/pr6120236 [112] Yang C C, Soh C S, Yap V V. A systematic approach to ON-OFF event detection and clustering analysis of non-intrusive appliance load monitoring. Frontiers in Energy, 2015, 9(2): 231-237 doi: 10.1007/s11708-015-0358-6 [113] Cipriano X, Vellido A, Cipriano J, Martí-Herrero J, Danov S. Influencing factors in energy use of housing blocks: A new methodology, based on clustering and energy simulations, for decision making in energy refurbishment projects. Energy Efficiency, 2017, 10(2): 359-382 doi: 10.1007/s12053-016-9460-9 [114] De Baets L, Develder C, Dhaene T, Deschrijver D. Detection of unidentified appliances in non-intrusive load monitoring using Siamese neural networks. International Journal of Electrical Power and Energy Systems, 2019, 104: 645-653 [115] 黄雅婷, 石晶, 许家铭, 徐波. 鸡尾酒会问题与相关听觉模型的研究现状与展望. 自动化学报, 2019, 45(2): 234-251Huang Ya-Ting, Shi Jing, Xu Jia-Ming, Xu Bo. Research advances and perspectives on the cocktail party problem and related auditory models. Acta Automatica Sinica, 2019, 45(2): 234-251 [116] Zhong M, Goddard N, Sutton C. Signal aggregate constraints in additive factorial HMMS, with application to energy disaggregation. In: Proceedings of the 27th International Conference on Neural Information Processing Systems. Montreal, Canada: ACM, 2014. 3590−3598 [117] Alshareef S, Morsi W G. Application of wavelet-based ensemble tree classifier for non-intrusive load monitoring. In: Proceedings of the 2015 IEEE Electrical Power and Energy Conference (EPEC). London, Canada: IEEE, 2015. 397−401 [118] Liu H, Wu H P, Yu C M. A hybrid model for appliance classification based on time series features. Energy and Buildings, 2019, 196: 112-123 doi: 10.1016/j.enbuild.2019.05.028 [119] Tabatabaei S M, Dick S, Xu W. Toward non-intrusive load monitoring via multi-label classification. IEEE Transactions on Smart Grid, 2017, 8(1): 26-40 doi: 10.1109/TSG.2016.2584581 [120] Wu X, Gao Y C, Jiao D. Multi-label classification based on random forest algorithm for non-intrusive load monitoring system. Processes, 2019, 7(6): Article No. 337 doi: 10.3390/pr7060337 [121] Li D, Dick S. Residential household non-intrusive load monitoring via graph-based multi-label semi-supervised learning. IEEE Transactions on Smart Grid, 2019 10(4): 4615-4627 doi: 10.1109/TSG.2018.2865702 [122] Bishop C M. Pattern Recognition and Machine Learning. New York: Springer, 2006. 610−631 [123] Zia T, Bruckner D, Zaidi A. A hidden Markov model based procedure for identifying household electric loads. In: Proceedings of the 37th Annual Conference of the IEEE Industrial Electronics Society. Melbourne, Australia: IEEE, 2011. 3218−3223 [124] Ghahramani Z, Jordan M I. Factorial hidden Markov models. Machine Learning, 1997, 29(2-3): 245-273 [125] Shaloudegi K, Gyürgy A, Szepesvári C, Xu W. SDP relaxation with randomized rounding for energy disaggregation. In: Proceedings of the 29th Conference on Neural Information Processing Systems (NIPS 2016). Barcelona, Spain: Curran Associates, 2016. 4985−4993 [126] Bonfigli R, Principi E, Fagiani M, Severini M, Squartini S, Piazza F. Non-intrusive load monitoring by using active and reactive power in additive factorial hidden Markov models. Applied Energy, 2017, 208: 1590-1607 doi: 10.1016/j.apenergy.2017.08.203 [127] Mengistu M A, Girmay A A, Camarda C, Acquaviva A A, Patti E. A cloud-based on-line disaggregation algorithm for home appliance loads. IEEE Transactions on Smart Grid, 2019, 10(3): 3430-3439 doi: 10.1109/TSG.2018.2826844 [128] Qin J H, Wan Y N, Yu X H, Kang Y. A newton method-based distributed algorithm for multi-area economic dispatch. IEEE Transactions on Power Systems, 2020, 35(2): 986-996 doi: 10.1109/TPWRS.2019.2943344 [129] Sandryhaila A, Moura J M F. Big data analysis with signal processing on graphs: Representation and processing of massive data sets with irregular structure. IEEE Signal Processing Magazine, 2014, 31(5): 80-90 doi: 10.1109/MSP.2014.2329213 [130] Ortega A, Frossard P, Kovačević J, Moura J M F, Vandergheynst P. Graph signal processing: Overview, challenges, and applications. Proceedings of the IEEE, 2018, 106(5): 808-828 doi: 10.1109/JPROC.2018.2820126 [131] Shuman D I, Narang S K, Frossard P, Ortega A, Vandergheynst P. The emerging field of signal processing on graphs: Extending high-dimensional data analysis to networks and other irregular domains. IEEE Signal Processing Magazine, 2013, 30(3): 83-98 doi: 10.1109/MSP.2012.2235192 [132] Zhao B C, Stankovic L, Stankovic V. On a training-less solution for non-intrusive appliance load monitoring using graph signal processing. IEEE Access, 2016, 4: 1784-1799 doi: 10.1109/ACCESS.2016.2557460 [133] He K, Stankovic L, Liao J, Stankovic V. Non-intrusive load disaggregation using graph signal processing. IEEE Transactions on Smart Grid, 2018, 9(3): 1739-1747 doi: 10.1109/TSG.2016.2598872 [134] Qi B, Liu L Y, Wu X. Low-rate nonintrusive load disaggregation for resident load based on graph signal processing. IEEJ Transactions on Electrical and Electronic Engineering, 2018, 13(12): 1833-1834 doi: 10.1002/tee.22746 [135] Mauch L, Yang B. A new approach for supervised power disaggregation by using a deep recurrent LSTM network. In: Proceedings of the 2015 IEEE Global Conference on Signal and Information Processing (GlobalSIP). Orlando, USA: IEEE, 2015. 63−67 [136] Mauch L, Yang Bin. A novel DNN-HMM-based approach for extracting single loads from aggregate power signals. In: Proceedings of the 2016 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). Shanghai, China: IEEE, 2016. 2384−2388 [137] Bonfigli R, Felicetti A, Principi E, Fagiani M, Squartini S, Piazza F. Denoising autoencoders for non-intrusive load monitoring: Improvements and comparative evaluation. Energy and Buildings, 2018, 158: 1461-1474 doi: 10.1016/j.enbuild.2017.11.054 [138] Xia M, Liu W A, Xu Y Q, Wang K, Zhang X. Dilated residual attention network for load disaggregation. Neural Computing and Applications, 2019, 31(12): 8931-8953 doi: 10.1007/s00521-019-04414-3 [139] Barsim K S, Yang B. On the feasibility of generic deep disaggregation for single-load extraction. arXiv: 1802.02139, 2018. [140] Tan C Q, Sun F C, Kong T, Zhang W C, Yang C, Liu C F. A survey on deep transfer learning. In: Proceedings of the 27th International Conference on Artificial Neural Networks and Machine Learning. Rhodes, Greece: Springer, 2018. 270−279 [141] De Baets L, Ruyssinck J, Develder C, Dhaene T, Deschrijver D. Appliance classification using vi trajectories and convolutional neural networks. Energy and Buildings, 2018, 158: 32-36 doi: 10.1016/j.enbuild.2017.09.087 [142] Singh S, Majumdar A. Deep sparse coding for non-intrusive load monitoring. IEEE Transactions on Smart Grid, 2018, 9(5): 4669-4678 doi: 10.1109/TSG.2017.2666220 [143] Singhal V, Maggu J, Majumdar A. Simultaneous detection of multiple appliances from smart-meter measurements via multi-label consistent deep dictionary learning and deep transform learning. IEEE Transactions on Smart Grid, 2019, 10(3): 2969-2978. doi: 10.1109/TSG.2018.2815763 [144] Nalmpantis C, Vrakas D. Machine learning approaches for non-intrusive load monitoring: From qualitative to quantitative comparation. Artificial Intelligence Review, 2018, 52(1): 217-243 [145] Kim H, Marwah M, Arlitt M, Lyon G, Han J W. Unsupervised disaggregation of low frequency power measurements. In: Proceedings of the 7th SIAM International Conference on Data Mining. Mesa, USA: SIANM, 2011. 747−758 [146] Makonin S, Popowich F. Nonintrusive load monitoring (NILM) performance evaluation. Energy Efficiency, 2015, 8(4): 809-814 doi: 10.1007/s12053-014-9306-2 [147] Lange H, Bergs M. Efficient inference in dual-emission FHMM for energy disaggregation. In: Proceedings of the 30th AAAI Conference on Artificial Intelligence Artificial Intelligence for Smart Grids and Smart Buildings. Phoenix, USA: AAAI, 2016. 248−254 [148] Pereira L, Nunes N. An experimental comparison of performance metrics for event detection algorithms in NILM. In: Proceedings of the 4th International Workshop on Non-Intrusive Load Monitoring. Austin, USA: IEEE, 2018. 1−4 [149] Maasoumy M, Sanandaji B M, Poolla K, Vincentelli A S. Berds-berkeley energy disaggregation data set. In: Proceedings of the 26th International Conference on Neural Inform−ation Processing Systems. Lake Tahoe, USA: Curran Associates, Inc., 2013. 1−6 [150] Anderson K, Ocneanu A, Benitez D, Carlson D, Rowe A, Bergés M. Blued: A fully labeled public dataset for event-based non-intrusive load monitoring research. In: Proceedings of the 2nd KDD Workshop on Data Mining Applications in Sustainability. Beijing, China: ACM, 2012. 12−16 [151] Barker S, Mishra A, Irwin D, Cecchet E, Shenoy P, Albrecht J. Smart*: An open data set and tools for enabling research in sustainable homes. In: Proceedings of the 2012 Workshop on Data Mining Applications in Sustainability (SustKDD 2012). Beijing, China: ACM, 2012. 1−6 [152] Nambi A S N U, Lua A R, Prasad V R. LocED: Location-aware energy disaggregation framework. In: Proceedings of the 2nd ACM International Conference on Embedded Systems for Energy-Efficient Built Environments. Seoul, Korea (South): ACM, 2015. 45−54 [153] Reinhardt A, Burkhardt D, Zaheer M, Steinmetz R. Electric appliance classification based on distributed high resolution current sensing. In: Proceedings of the 37th Annual IEEE Conference on Local Computer Networks. Clearwater, USA: IEEE, 2012. 999−1005 [154] Makonin S, Popowich F, Bartram L, Gill B, Bajić I V. AMPds: A public dataset for load disaggregation and eco-feedback research. In: Proceedings of the 2013 IEEE Electrical Power and Energy Conference. Halifax, Canada: IEEE, 2013. 1−6 [155] Kelly J, Knottenbelt W. The UK-DALE dataset, domestic appliance-level electricity demand and whole-house demand from five UK homes. Scientific Data, 2015, 2: Article No. 150007 doi: 10.1038/sdata.2015.7 [156] Batra N, Gulati M, Singh A, Srivastava M B. It’s different: Insights into home energy consumption in India. In: Proceedings of the 5th ACM Workshop on Embedded Systems for Energy-Efficient Buildings. Roma, Italy: ACM, 2013. 1−8 [157] Murray D, Stankovic L, Stankovic V. An electrical load measurements dataset of united kingdom households from a two-year longitudinal study. Scientific Data, 2017, 4: Article No. 160122 doi: 10.1038/sdata.2016.122 [158] Shin C, Rho S, Lee H, Rhee W. Data requirements for applying machine learning to energy disaggregation. Energies, 2019, 12(9): Article No. 1696 doi: 10.3390/en12091696 [159] Monacchi A, Egarter D, Elmenreich W, D＇Alessandro S, Tonello A M. GREEND: An energy consumption dataset of households in Italy and Austria. In: Proceedings of the 2014 IEEE International Conference on Smart Grid Communications. Venice, Italy: IEEE, 2014. 511−516 [160] Gao J K, Giri S, Kara E C, Bergés M. PLAID: A public dataset of high-resoultion electrical appliance measurements for load identification research: Demo abstract. In: Proceedings of the 1st ACM Conference on Embedded Systems for Energy-Efficient Buildings. Memphis, USA: ACM, 2014. 198−199 [161] Jazizadeh F, Afzalan M, Becerik-Gerber B, Soibelman L. EMBED: A dataset for energy monitoring through building electricity disaggregation. In: Proceedings of the 9th International Conference on Future Energy Systems. Karlsruhe, Germany: ACM, 2018. 230−235 [162] Parson O, Fisher G, Hersey A, Batra N, Kelly J, Singh A, et al. Dataport and NILMTK: A building data set designed for non-intrusive load monitoring. In: Proceedings of the 2015 IEEE Global Conference on Signal and Information Processing (GlobalSIP). Orlando, USA: IEEE, 2015. 210−214 [163] Batra N, Kukunuri R, Pandey A, Malakar R, Kumar R, et al. Towards reproducible state-of-the-art energy disaggregation. In: Proceedings of the 6th ACM International Conference on Systems for Energy-Efficient Buildings, Cities, and Transportation. New York, USA: ACM, 2019. 193−202 [164] 白昱阳, 黄彦浩, 陈思远, 张俊, 李柏青, 王飞跃. 云边智能: 电力系统运行控制的边缘计算方法及其应用现状与展望. 自动化学报, 2020, 46(3): 397-410Bai Yu-Yang, Huang Yan-Hao, Chen Si-Yuan, Zhang Jun, Li Bai-Qing, Wang Fei-Yue. Cloud-edge intelligence: Status quo and future prospective of edge computing approaches and applications in power system operation and control. Acta Automatica Sinica, 2020, 46(3): 397-410 [165] Kairouz P, McMahan H B, Avent B, Bellet A, Bennis M, Bhagoji A N, Bonawitz K, et al. Advances and open problems in federated learning. Machine Learning, 2021, 14(1-2): 1-210 [166] Wu Z H, Pan S R, Chen F W, Long G D, Zhang C Q, Yu P S. A comprehensive survey on graph neural networks. IEEE Transactions on Neural Networks and Learning Systems, 2021, 32(1): 4-24 doi: 10.1109/TNNLS.2020.2978386 期刊类型引用(55)
1. 刘俊雨，白牧可，张立梅. 基于耦合特征和改进随机森林的非侵入式农业负荷辨识. 供用电. 2025(01): 106-115 . 
百度学术2. 丁健，陈鉴祥，刘旺朋，丁一帆，王谱宇. 基于改进型颜色编码与密集连接网络的非侵入式负荷监测. 浙江电力. 2025(02): 3-12 . 百度学术3. 王谱宇，丁一帆，陈鉴祥，刘兴江，方凯杰，程含渺，张小平. 基于动态谐波导纳参数的非侵入式负荷监测数据模拟生成方法. 中国电机工程学报. 2025(03): 923-936 . 百度学术4. 黄颖祺，颜钟宗，郝芃斐，温和. 工业非介入式负荷监测研究综述. 中国测试. 2025(01): 11-23 . 百度学术5. 杨舒惠，黎静华，韦善阳. 基于图像分类网络的非侵入式负荷辨识算法的运算成本优化. 电力自动化设备. 2024(01): 141-146 . 百度学术6. 罗平，朱振宇，樊星驰，孙博宇，张帆，吕强. 基于时序生成对抗网络的居民用户非侵入式负荷分解. 电力系统自动化. 2024(02): 71-81 . 百度学术7. 李子凯，岳宝强，杨波，周忠堂，王春宝. 适于低功率状态的多特征融合负荷分解方法. 山东电力技术. 2024(01): 68-76 . 百度学术8. 刘凯，符玲，杨金刚，熊思宇，蒿保龙，刘丽娜. 基于简化HMM和时间分段的非侵入式负荷分解算法. 电力自动化设备. 2024(02): 198-203+210 . 百度学术9. 季坚莞，胡文军，王闯. 基于物联网的非侵入式负荷状态监控系统设计. 湖州师范学院学报. 2024(02): 53-64 . 百度学术10. 韩林池，高放，赵子巍，郭苏杭，李想，张冬冬，武新章. 基于多尺度卷积与Informer混合模型的非侵入式负荷监测方法. 电力自动化设备. 2024(03): 134-141 . 百度学术11. 林顺富，李毅，沈运帷，林屹峰，李东东. 基于全卷积去噪自编码器与卷积块注意力模块的非侵入式居民负荷分解模型. 电力自动化设备. 2024(03): 127-133 . 百度学术12. 侯林超，朱武，汤德清. 基于深度学习网络的非侵入式负荷分解方法. 计算机仿真. 2024(02): 137-140+445 . 百度学术13. 徐瑞琪，刘丹丹. 基于多尺度特征融合和多头自注意力机制的非侵入式负荷监测. 科学技术与工程. 2024(06): 2385-2395 . 百度学术14. 杨苗，游文霞，刘玥，汪芯茜. 基于颜色编码和残差神经网络的非侵入式负荷识别. 电工材料. 2024(02): 94-99 . 百度学术15. 刘岩，王玉君，杨晓坤，李文文，郭磊. 基于KPCA和XGBoost算法的非侵入式负荷辨识方法. 电测与仪表. 2024(05): 71-77 . 百度学术16. 张瀚文，李鹏，郎恂，沈鑫，梁俊宇，苗爱敏. 基于UMAP流形特征提取和KELM的非侵入式负荷监测方法研究. 电子器件. 2024(02): 448-457 . 百度学术17. 汪敏，张孟健，禹洪波，熊炜，袁旭峰，邹晓松. 基于EEMD和特征降维的非侵入式负荷分解方法研究. 电测与仪表. 2024(06): 80-86 . 百度学术18. 王丹宇，刘君，周亚同，何静飞. 面向用电负荷分解的特征融合与Transformer模型. 电力系统及其自动化学报. 2024(06): 129-136 . 百度学术19. 李利娟，刘海，刘红良，张青松，陈永东. 融合外部注意力机制的序列到点非侵入式负荷分解. 上海交通大学学报. 2024(06): 846-854 . 百度学术20. 张媛，王飞，张照锋，崔秀华，翟琳. 基于小波变换与神经网络的非侵入式家电负荷监测研究. 电子器件. 2024(03): 749-756 . 百度学术21. 卞海红，孙鑫. 基于改进SVIT算法的非侵入式居民负荷监测方法. 电子设计工程. 2024(16): 87-91+96 . 百度学术22. 韩小地，郭家兴，钱康，刘文瑞. 基于双分支特征集成LSTM的非侵入式负荷分解研究. 山东电力技术. 2024(10): 31-43 . 百度学术23. 王保义，邓晓平，张桂青，阎俏，袁帅. 基于高低频数据融合的非侵入式负荷监测系统设计与应用. 建筑科学. 2024(10): 62-69+142 . 百度学术24. 张媛，王飞，崔秀华，翟琳，张照锋. 基于改进GMM算法的非侵入式负荷监测方法研究. 电子器件. 2024(05): 1382-1388 . 百度学术25. 黄宇，蒲曾鑫，丁超. 一种基于VAE的非侵入式家庭用电能耗分解方法. 电力大数据. 2024(08): 40-47 . 百度学术26. 刘希臣. 基于NB-IoT的变电站电气设备实时负载监测技术. 现代电子技术. 2023(02): 149-153 . 百度学术27. 张石清，王伟，钱亚冠，赵小明，杜磊，章为昆. 面向深度学习的非侵入式负荷监测研究进展. 计算机系统应用. 2023(03): 25-47 . 百度学术28. 姚刚，王赟，王远亮，宋子浩. 基于多模式集成深度融合的非侵入式负荷分解模型. 控制与信息技术. 2023(01): 1-10 . 百度学术29. 刘航，刘春阳，赵浩然，田立军，刘俊伟. 基于深度学习的公共楼宇电-气负荷非侵入式分解. 电网技术. 2023(03): 1188-1197 . 百度学术30. 王德文，貟青青. 基于时域卷积网络和自注意力的非侵入式负荷监测方法. 电力科学与工程. 2023(03): 42-51 . 百度学术31. 王家驹，王竣平，白泰，张然，丁熠辉，杨林，张姝. 基于设备特征多层优选和CNN-NLSTM模型的非侵入式负荷分解. 电力科学与技术学报. 2023(01): 146-153 . 百度学术32. 王传君，缪巍巍，曾锃，李世豪，蒋姝. 基于复合特征的非侵入式电力负荷识别方法研究. 电子器件. 2023(01): 218-222 . 百度学术33. 刘仲民，赵丹阳. 基于VMD-CNN的非侵入式负荷监测方法研究. 电气传动. 2023(06): 53-61 . 百度学术34. 王赟，葛泉波，姚刚，王梦梦，姜淏予. 基于多注意力机制集成的非侵入式负荷分解算法. 南京信息工程大学学报(自然科学版). 2023(03): 304-314 . 百度学术35. 钱玉军，包永强，姜丹琪，张旭旭. 基于LSTM和序列到点模型的非侵入式负荷分解技术. 电子器件. 2023(03): 841-848 . 百度学术36. 鲍海波，杨舒惠，陈子民，郭小璇，黎静华. 事件检测类非侵入式负荷监测算法综述. 电力系统自动化. 2023(13): 94-109 . 百度学术37. 钱玉军，包永强，姜丹琪，张旭旭，雷家浩. 基于通道注意力机制和BIGRU的非侵入式负荷分解方法. 电力信息与通信技术. 2023(07): 1-10 . 百度学术38. 李岢淳，李兵. 基于IPSO-FHMM的非侵入式负荷分解. 计算机系统应用. 2023(08): 214-220 . 百度学术39. 洪德华，刘翠玲，赵林燕，雷沁怡，王海鑫. 基于多域特征分析与选择的电力数据识别方法. 水电能源科学. 2023(09): 211-215 . 百度学术40. 齐屹，陈书畅，林才华，张逸. 基于阻抗分离的工业谐波源非侵入式识别方法. 电力系统及其自动化学报. 2023(10): 148-158 . 百度学术41. 李梓彤，杨超，陈飞. 基于PCA-SSA-ELM的非侵入式负荷识别方法. 电子器件. 2023(05): 1357-1363 . 百度学术42. 符明，段斌. 基于格拉姆角差场图像编码的非侵入式负荷识别方法. 激光与光电子学进展. 2023(24): 87-95 . 百度学术43. 吴玉刚，迟长春. 基于VMD的家用电器负荷特征提取与识别. 上海电机学院学报. 2023(06): 317-323 . 百度学术44. 欧飞，刘敏. 用于需求响应能力评估的概率负荷识别方法. 电力科学与工程. 2022(01): 8-15 . 百度学术45. 王毓琦，高嵩，万校宏，李元元，杨子江. 电网负荷分类评价反馈算法研究. 山东电力技术. 2022(03): 20-24+58 . 百度学术46. 崔昊杨，蔡杰，陈磊，江超，江友华，张驯. 基于颜色编码的非侵入式负荷细粒度识别方法. 电网技术. 2022(04): 1557-1567 . 百度学术47. 龙世渚，涂来. 基于智能家居的家庭用电监测系统设计. 物联网技术. 2022(04): 27-31 . 百度学术48. 林顺富，詹银枫，李毅，李东东. 基于CNN-BiLSTM与DTW的非侵入式住宅负荷监测方法. 电网技术. 2022(05): 1973-1981 . 百度学术49. 王毅，王萧阳，李松浓，陈涛，侯兴哲，付秀元. 基于单分类结合模糊宽度学习的负荷辨识方法. 电子技术应用. 2022(05): 51-55+60 . 百度学术50. 安琪，梁宇飞，王耀强，王占彬，李争，李峥，安国庆. 基于K-means聚类与PSO特征优选KNN的分级负荷识别方法. 河北科技大学学报. 2022(03): 249-258 . 百度学术51. 杨丹旭，宋亚奇，岳建任，李莉. 基于投票方差的非侵入式负荷波动检测. 电力自动化设备. 2022(07): 46-50+110 . 百度学术52. 蔡雨露，聂玉虎，崔文朋，郑哲，刘瑞，池颖英. 基于谐波分量与有效值的神经网络负荷分解. 电子技术应用. 2022(08): 123-126 . 百度学术53. 任智仁，汤博，周弼，薛雷，易灵芝，刘西蒙. 基于CNN-BiGRU-Attention的非侵入式负荷分解. 电机与控制应用. 2022(08): 41-46+52 . 百度学术54. 陈俊，彭勇刚，凌家源，蔡田田，邓清唐. 基于概率稀疏自注意力模型的非侵入式负荷分解. 电网技术. 2022(10): 3932-3941 . 百度学术55. 葛翔，山宪武，宋秩行，陈术，高弘毅，孙永辉. 基于非侵入式负荷监测数据的居民相似邻里查找模型. 南京理工大学学报. 2022(05): 600-605 . 百度学术其他类型引用(87)
-
下载:
计量
- 文章访问数: 4507
- HTML全文浏览量: 3495
- PDF下载量: 735
- 被引次数: 142
