Sequence-to-Sequence Load Disaggregation Using Multiscale Residual Neural Network

With the increased demand on economy and efficiency of measurement technology, nonintrusive load monitoring (NILM) has received more attention as a cost-effective way to monitor electricity provide feedback users. Deep neural networks have been showing great potential in field disaggregation. In this article, first, new convolutional model based residual blocks is proposed avoid degradation problem whose traditional or less suffer from when network layers are order learn complex features. Second, we propose dilated convolution curtail excessive quantity parameters obtain bigger receptive multiscale structure mixed data features targeted way. Third, give details about generating training test set under certain rules. Finally, algorithm tested real-house public set, UK Domestic Application Level Electric (UK-DALE), with three existing networks. The results compared analyzed, shows improvements F1 score, MAE, well complexity across different appliances.