On Statistical Modeling and Forecasting of Energy Usage in Smart Grid
Wei Yu, Dou An, David W. Griffith, Qingyu Yang, Guobin Xu
Developing effective energy resource management in the smart grid is challenging because the entities in both the demand and supply sides experience various uncertainties. This paper addresses the issue of quantifying uncertainties on the energy demand side and the developed techniques in the paper can be applied to the energy generation side as well. The statistical modeling analysis approaches are developed to derive a statistical distribution of energy usage and the machine learning based approaches such as the Support Vector Machines (SVM) and neural network are developed to conduct the accurate forecasting on energy usage. The extensive experiments of proposed approaches are conducted using a real-world meter reading data set. The experimental data shows that the statistical distribution of meter reading data can be largely approximated with a Gaussian distribution and the two SVM-based machine learning approaches achieve a high accuracy of energy usage forecasting. Extensions to other smart grid applications (e.g., forecasting energy generation and determining optimal demand response) are discussed as well.