Document Type

Closed Project

Publication Date

Spring 2017

Instructor

Sule Balkan & Tim Anderson

Course Title

Energy Demand Forecasting

Course Number

ETM 510

Abstract

Generating and managing the electrical power is one of the important aspects of the electrical grid. The technical process might be easier to manage if the power load or the power demand is known ahead of time. Some information might be available, but they may not be enough. For example, the power demand usually does not equal to zero, so the lowest possible generated power could be known and it could not equal to zero. However, the highest possible demand might not be always available, and the variation might not be known. Preparing for the future is also important. If population increases, there is a higher probability that power demand will also be increasing. Planning new generation plants is important. To manage existing generation, or to plan for new generation plants, the energy demand might be a major contributor. Forecasting power is one of the ways that are used to get the power demand in the future. There are multiple ways to get a rough idea of the energy demand. One way is by looking at historical data, and processing that data. The historical data of the New England region, which includes Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island, and Vermont, will be used to establish a forecasting model. The only historical data that will be used are the 2015 and the 2016 data. There are three different forecasting methods used to predict the energy demand. The first method is the linear interpolation, the second method is the Arima model, and the third method is the neural networks. The prediction from each method will be submitted to the GEFCom2017 competition.

Description

This project is only available to students, staff, and faculty of Portland State University

Persistent Identifier

http://archives.pdx.edu/ds/psu/21457

Share

COinS