Energy Consumption and Production Forecasting Using Artificial Neural Network for Optimal Energy Cost Savings

With the goal of reducing their carbon footprint, Instituto Superior de Engenharia de Lisboa (ISEL) has put in place several measures in the past few years. One has been to install a solar photovoltaic (PV) system for self-consumption. Load forecasting and PV production models were developed, based on measured data, resorting to Artificial Neural Networks (ANN), with the optimization of ISEL’s energetic needs in mind. To test the forecasting models, an optimization system using linear programming was also developed, focusing on minimizing the cost of energy consumption from the grid, resorting to the management of an Energy Storage System (ESS). The aim is to understand the impact on the optimization system using as entries the measured and the forecasted data by the ANN. The results obtained validate the ANN forecast model and demonstrate its robustness and predictive capacity.

I. Introduction

An ANN is a computational model inspired by the human brain’s inner workings, which is capable of learning from data and using that information to make decisions in specific tasks. The neural network can be trained, using large amounts of data, and used to identify patterns and make predictions [1], [2].