Load Forecasting based on Neural Network

Jie Bao
AI Lab, Dept of Computer Science
Iowa State University

Problem Statement

See Term project proposal for Machine learning

Theory description

* Time series forecasting problem
* Neural network as a balck-box forecaster
* Local Regression for white-box model 

Experiment

* On Java
* On Matlab

Project Progress and Documents

* 2001-09-10: Get primary data and begin this project
* 2001-09-28: First Report (7 pages)
* 2001-10-25: Finish Model based on BP and regression. Second Report ( 8 pages). waiting for further data.
* 2002-03-02: Load and Weather Data Coming.
* 2002-03-15: Create Document List
* 2002-04-05: Presentation about Short Time Load Forecasting (STLP) based on Neural Net
* 2002-04-15: Term project proposal for Machine learning
* 2002-04-19: Neural network model : error 5.6% (24-hour)
* 2002-04-28: Moving average model: error 4.3%(24-hour)
* 2002-05-05: Using forcasted temperature: error 3.5%(24-hour)
* 2002-05-08: Report: Short-term Load Forecasting based on Neural network and Moving Average
* 2002-05: Using Local Regression ; Finish
* 2002-10-15: PowerPoint Slide for Short-term Load Forecasting based on Neural network and Local Regression
* 2003-10-27: Poster on JVA symposium