Streamflow forecasting by a feedforward neural network with Levenberg-Marquardt training using the MATLAB toolboxes

• Grant number: 08/01229-3
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: May 01, 2008
• End date: July 31, 2010
• Field of knowledge: Engineering - Electrical Engineering - Power Systems
• Principal Investigator: Anna Diva Plasencia Lotufo
• Grantee: Klayton Antonio Moreira Araújo
• Host Institution: Faculdade de Engenharia (FEIS). Universidade Estadual Paulista (UNESP). Campus de Ilha Solteira. Ilha Solteira , SP, Brazil

Abstract

Load Forecasting is very important for planning and operation of electrical power systems. Since the advance of neural networks and the application in forecasting problems, they are very interesting and the statistical traditional methods like ARIMA of Box and Jenkins become unused. Feedforward neural network with training with backpropagation is one of the most used methods found on the literature, but the convergence is not so fast. The Levenberg-Marquardt training algorithm is a very useful alternative due to the faster convergence. This work intends to use a feedforward neural network trained with Levenberg-Marquardt algorithm by the toolboxes of MATLAB applied to electrical load forecasting with data from a Brazilian electrical power company.