Subtask Gated Networks for Non-Intrusive Load Monitoring

Abstract

Non-intrusive load monitoring (NILM), also known as energy disaggregation, is a blind source separation problem where a households aggregate electricity consumption is broken down into electricity usages of individual appliances. In this way, the cost and trouble of installing many measurement devices over numerous household appliances can be avoided, and only one device needs to be installed. The problem has been well-known since Hart's seminal paper in 1992, and recently significant performance improvements have been achieved by adopting deep networks. In this work, we focus on the idea that appliances have on/off states, and develop a deep network for further performance improvements. Specifically, we propose a subtask gated network that combines the main regression network with an on/off classification subtask network. Unlike typical multitask learning algorithms where multiple tasks simply share the network parameters to take advantage of the relevance among tasks, the subtask gated network multiply the main network's regression output with the subtask's classification probability. When standby-power is additionally learned, the proposed solution surpasses the state-of-theart performance for most of the benchmark cases. The subtask gated network can be very effective for any problem that inherently has on/off states.

Scale- and Context-Aware Convolutional Non-Intrusive Load Monitoring

“…Following previous studies [2,15], we use data of houses 2-6 to create the training set and leave the data for house 1 as the test set for the REDD dataset. Disaggregation is implemented for fridge, dishwasher, and microwave.…”

“…Estimating the power consumption sequence of an appliance with length s using the aggregate consumption signal corresponding to the same time window is difficult as contextual information outside the window is not considered. Thus, it is suggested to add windows of length w to both sides for the input aggregate sequence [2,16]. Specifically, with input sequence x t := (x t−w , · · · , x t+s+w−1 ), y i t := (y i t , · · · , y i t+s−1 ) is the output sequence for the ith appliance.…”

“…For simplicity, we omit the superscript i and subscript t hereafter. The structure of the two sub-networks proposed in [2,15] is illustrated in Fig. 2.…”

“…As the gating operation is separately performed for each dilation rate, a rich combination of features at different time scales can be achieved. We then concatenate p and obtain p (2) and s (2) (note that s (2) contains features with the rectified linear unit (ReLU) activation function instead of the sigmoid function). Both p (2) and s (2) are processed by a convolutional layer with a kernel size of 1, yielding p (3) and s (3) , the inputs to the self-attention modules.…”

“…The integration of domain knowledge further enriches the design of CNN architectures. An on/off state classification subnetwork can be added in parallel to the regression sub-network so that the model can learn from on/off state information directly [2,18]. The work in this paper adopts the structure of subtask gated network (SGN) [2] as a starting point.…”