Advances in optimization and prediction techniques: Real-world applications

Article type: Other

Authors: Troncoso Lora, Alicia

Affiliations: Area of Computer Science, University Pablo de Olavide, Seville, Spain E-mail: [email protected]

Abstract: This paper describes a time-series prediction method based on the k-Weighted Nearest Neighbours (k-WNN) algorithm and a simple technique to deal with nonconvex, nonlinear optimization problems by solving a sequence of Interior Point (IP) subproblems. The proposed prediction methodology is applied to obtain the 24-hour forecasts of two real time series: the demand and the energy prices in the competitive Spanish Electricity Market. The proposed optimization method is applied to the optimal scheduling of the electric energy production in the short-term.

Keywords: Machine Learning, forecasting, nonconvex nonlinear optimization

Journal: AI Communications, vol. 19, no. 3, pp. 295-297, 2006