Locational marginal price forecasting by locally linear neuro-fuzzy model

Author

Sarafraz, Fahimeh ; Ghasemi, Hassan ; Monsef, Hassan

Author_Institution

Sch. of Electr. & Comput. Eng., Univ. of Tehran, Tehran, Iran

fYear

2011

Firstpage

1

Lastpage

4

Abstract

Price forecasting in competitive electricity markets plays a crucial role for any decision making. This is a difficult task since price time series are non-stationary, and with variable mean and variance, and also have periodic monthly and seasonal behavior. This paper introduces an approach to forecast several-hours-ahead electricity locational marginal price (LMP) using locally linear neuro-fuzzy (LLNF) model for the PJM market. The autocorrelation method is used to make the appropriate input vectors. The LLNF model leads to more accurate results compared to the Multi Layer Perceptron (MLP) neural network.

Keywords

decision making; forecasting theory; fuzzy set theory; neural nets; power markets; pricing; LLNF model; PJM market; autocorrelation method; decision making; electricity markets; locally linear neuro-fuzzy model; several hours-ahead electricity locational marginal price forecasting; Correlation; Electricity; Forecasting; Neurons; Predictive models; Testing; Training data; electricity market; locally linear neuro-fuzzy model; locational marginal price forecasting; multi layer perceptron neural network;

fLanguage

English

Publisher

ieee

Conference_Titel

Environment and Electrical Engineering (EEEIC), 2011 10th International Conference on

Conference_Location

Rome

Print_ISBN

978-1-4244-8779-0

Type

conf

DOI

10.1109/EEEIC.2011.5874828

Filename

5874828