1 ConvEDNet: A Convolutional Energy Disaggregation Network Using Continuous Point-On-Wave Measurements

Energy disaggregation means separating the distribution system-level net-load measurements into native load and photovoltaic (PV) generation. This paper proposes a causal, context-aware, and fully-convolutional deep-learning network for simultaneous PV-load energy disaggregation using continuous point-on-wave measurements. The proposed network, called Conv-EDNet, uses an encoder-decoder framework combined with a separator network for performing disaggregation in the time domain. The separator network harnesses the power of stacked dilated temporal convolutions for learning two weighting functions that perform the disaggregation using the non-negative encoder output. Finally, the decoder converts the weighted encoder output into time-domain native-load and PV generation measurements. Further, Conv-EDNet+, which combines the Conv-EDNet with a gated recurrent unit-based discriminator for adversarial learning, is proposed. Numerical evaluations of the proposed approaches outperform the present state-of-the-art methods widely used for disaggregation tasks.