Nonintrusive disaggregation of residential air-conditioning loads from sub-hourly smart meter data

An analysis on the energy consumption of circulating pumps of residential swimming pools for peak load management

“…First, Although most of steady-state signatures based NIALM methods do not emphasize the importance of the sampling frequency, an implicit assumption is that the sampling rate needs to be high enough to capture most of state changes of major appliances [19,20,24,30], or the prior information of use patterns of appliances must be available [18]. NIALM needs to captures state changes (events) of major appliances, and they would fail if an appliance's state has changed multiply times during two successive samples [19,30].…”

“…For NIALM methods, features selection and load disaggregation are two key issues. At present two types of features of appliances are usually used: steady-state signatures [18,19,[23][24][25] and transient signatures [17,21,29]. Steady-state signatures of appliances are signatures when appliances are in stable states, such as the active power, the reactive power, and the power factor angle.…”

“…Compared with intrusive appliance load monitoring (IALM) which measures each appliance energy consumption with a specific meter, NIALM has many advantages, such as lower cost of installation and maintenance, smaller space requirements and greater reliability. NIALM is a very hot topic in recent years, and currently there are plenty of NIALM methods [17][18][19][20][21][22][23][24][25][26][27][28][29]. For NIALM methods, features selection and load disaggregation are two key issues.…”

“…However, all of these NIALM methods cannot be used for the CPRSP energy consumption, due to the very low sampling frequency (hourly data) and no prior information on CPRSP in this work. Although most of steadystate signatures based NIALM methods do not emphasize the importance of the sampling frequency, an implicit assumption is that the sampling frequency needs to be high enough to capture most of state changes of major appliances [19,20,24,30], or the prior information of use patterns of appliances must be available [18]. From another aspect, the estimation of the CRPSP energy consumption can be converted to estimations of base load and the temperature-dependent energy consumption, and some temperature-based building energy disaggregation methods such as degree days [31,32] and change-point models [33,34] may be used.…”

“…Compared with IALM which measures each appliance's energy consumption with a specific meter, NIALM has many advantages, such as lower cost of installation and maintenance, smaller space requirements and greater reliability [17]. Currently there are plenty of NIALM methods [17][18][19][20][21][22][23][24][25][26][27][28][29]. Among these methods, most of them are event-based, meaning that appliances are monitored by detecting the change of some steady-state [19,[23][24][25] or transient-state features [17,21,29] related to current or voltage, and then corresponding energy consumptions are estimated by supervised [17,20] or unsupervised [19,23] methods, depending on whether prior information is available.…”