Non-intrusive Load Monitoring based on Convolutional Neural Network with Differential Input

Non-intrusive load monitoring (NILM) is a process for analyzing load in a building and deducing what appliances are working as well as their individual energy consumption. Compared with intrusive load monitoring, NILM is low cost, easy to deploy, and flexible. NILM installed in smart grids can provide information for decision making for energy management and therefore support energy-related industrial services. In this paper, we propose a NILM-based energy management system for appliance-level load monitoring service and a convolutional neural network based model with differential input. Experiment shows that the proposed model with differential input outperforms the existing models with raw input.