Non-instrusive Load Identification Method based on FHMM in a New Energy System

Load disaggregation aims to obtain a detailed indoor electricity usage based on comprehensive data monitored by smart meters. Real-time load disaggregation can help users make reasonable electricity-using plans under the time-of-use policy in smart grid and power IoT. It can also be used as the basis for load forecasting, and it is conducive to energy conservation in energy use terminal. This paper presents an energy management system to manage electricity process from power generation to load terminal, among which the smart meters are the core devices. The five components of this system (smart meters, communication networks, servers, dispatching automation system and energy services system) and their relationships with power IoT are described. The proposed method in this paper considers load disaggregation to be a factorial hidden Markov model (FHMM) which can be trained by the method using in the generalized additive models (GAMs). Numerical simulations indicate the effectiveness of the proposed method for energy disaggregation. Through the application of advanced metering infrastructure and integrated information system, this method can make a promotion for the use of intelligent electricity.