Energy Management Research Journal (EMRJ)

Research Article

Energy Consumption and Saving Calculations Using Nearest Neighbor and Artificial Neural Network Models

Kevin Eaton¹, Nabil Nassif ², Pyrian Rai¹ and Alexander Rodrigues²

¹Department of Computer Science, University of Cincinnati, Cincinnati, Ohio, United States 45221.

²Department of Civil and Architectural Engineering and Construction Management, University of Cincinnati, 2850 Campus Way Drive, 790 Rhodes Hall, Cincinnati, Oh 45221.

Submitted: December 13, 2018; Accepted: March 23, 2019.

Abstract

Demand for building units with advanced saving techniques has grown and continues to grow. Any company or organization has plenty of need for saving energy, whether the motive behind energy saving is saving money or aiding in saving the environment. Buildings now have the technology to track and produce highly accurate electoral outputs. These can be used for improving the functionality of energy models. This paper discusses typical data-based building energy models and proposes new improvements by utilizing an artificial neural network. Using sub-hourly and hourly electric energy consumptions, five different data-based models are utilized and compared to one another. The first two models are linear one regressor models. The first is a linear fit model and the second is a linear change point fit. The third model is a two regressor model using a linear fit. The fourth model is a proposed Classification Learning model using three regressors. The fifth model is a proposed Artificial Neural Network model using three regressors. The two types of data collected are simulation data and actual data. There are four buildings in total; two with simulation data and two with actual data. The results show that the proposed Artificial Neural Network model and K Nearest Neighbor model can provide accurate predictions for the data as compared to traditional linear modeling techniques. These models are then utilized to calculate saving percentage, which is then compared to the actual percentage.

Keywords

Energy saving; Energy prediction; Artificial Neural Network; Classification learner; Linear estimation.

Downlaod Full Text >>