Different Sampling Rates on Neural NILM Energy Disaggregation

Energy disaggregation is a process to estimate an appliance power usage by providing only a household power reading. One of the techniques used in the energy disaggregation is to use neural network. However, when using neural network in the energy disaggregation, it can be time consuming for an online or real-time applications. In this paper, it is to investigate if lowering the sampling rate would reduce time for disaggregating an aggregated reading while still provide an acceptable result. The experiment in this paper using RNN with LSTM as a disaggregation algorithm on a public dataset of UKDALE. The results show that lowering the sampling rate can significantly reduce both training time duration and disaggregating time duration, but it reduces the performance of the algorithm by a little.