Multi-Agent Intelligent Simulator to estimate U.S. wholesale price of electricity

Author

Sueyoshi, Toshiyuki ; Goto, Mika

Author_Institution

Manage. Dept., New Mexico Inst. of Min. & Technol., Socorro, NM, USA

fYear

2010

fDate

10-13 Oct. 2010

Firstpage

3278

Lastpage

3283

Abstract

This study examines the price estimation capability of MAIS (Multi-Agent Intelligent Simulator) when two types of agents with different learning capabilities coexist in a power trading market. This study identifies that the proposed MAIS, considering the coexistence of different types of agents, can improve its estimation accuracy of wholesale electricity price. This study also reexamines the estimation capability of the MAIS on a data set generated by the mean reverting method. Using a real data set regarding PJM and its simulated data sets, we confirm that the proposed MAIS performs as well as the other well-known computer science approaches (SVM: Support Vector Machines, NN: Neural Networks, and GA: Genetic Algorithm) in terms of price estimation.

Keywords

genetic algorithms; multi-agent systems; neural nets; power engineering computing; power markets; pricing; support vector machines; US wholesale electricity price estimation; genetic algorithm; learning capabilities; mean reverting method; multiagent intelligent simulator; neural networks; power trading market; support vector machines; Artificial neural networks; Complexity theory; Electricity; Estimation; Gallium; Machine learning; Support vector machines; Agent-based Approach; Mean Reverting Model; Numerical Analysis; Power Trading;

fLanguage

English

Publisher

ieee

Conference_Titel

Systems Man and Cybernetics (SMC), 2010 IEEE International Conference on

Conference_Location

Istanbul

ISSN

1062-922X

Print_ISBN

978-1-4244-6586-6

Type

conf

DOI

10.1109/ICSMC.2010.5642314

Filename

5642314