Distributed machine learning based smart-grid energy management with occupant cognition

It is challenging to process real-time data analysis and prediction for a smart grid in a building with consideration of both occupant profile and energy profile. This paper proposed a distributed and networked machine learning platform on smart gateways based smart grid. It can analyze occupants motion, provide short-term energy forecasting and allocate renewable energy resource. Firstly, occupant profile is captured by real-time indoor positioning system with Wi-Fi data analysis; and the energy profile is extracted by real-time meter system with electricity load data analysis. Then, the 24-hour occupant profile and energy profile are fused with prediction using an online distributed machine learning with real-time data update. Based on the forecasted occupant motion profile and energy consumption profile, solar energy source is allocated on the additional electricity power-grid in order to reduce peak demand on the main electricity power-grid. The whole management flow can be operated on the distributed smart gateway network with limited computation resource but with a supported general machine-learning engine. Experiment results on real-life datasets have shown that the accuracy of the proposed energy prediction can be 14.83% improvement comparing to SVM method. Moreover, the peak load from main electricity power-grid is reduced by 15.20% with 51.94% energy cost saving.