Abstract—The implementation of the smart grid will greatly improve the efficiency of energy supply by detecting, predicting, and reacting to real-time local changes of energy uses. To this end, energy usage prediction of household buildings is critically important to facilitate the implementation of smart grid. This study used a single house as a prototype, employed different observed features, advanced data analysis approach, and artificial neural network model to predict real-time dynamics of house energy usage. Data analysis revealed that among the 26 observed features, only the top ten most important features were helpful and could maximize the neural network model performance. The resultant model has the great predictive capability on energy usage, thus provided a promising framework to improve the smart grid implementation. 

Index Terms—building energy use, machine learning, principal component analysis, recurrent neural network

Cite: Mengxuan Sun, Jinglin Zhao, and Heidan Shang, "Building Energy Consumption Prediction with Principal Component Analysis and Artificial Neural Network," International Journal of Electronics and Electrical Engineering, Vol. 8, No. 2, pp. 36-39, June 2020. doi: 10.18178/ijeee.8.2.36-39