Title :
Effect of Agent´s Action Domain Representation Method in Agent-Based Electricity Market Simulation
Author :
Jing, Zhaoxia ; Chen, Haoyong ; Ngan, H.W. ; Wang, Jianhui
Author_Institution :
Sch. of Electr. Eng., South China Univ. of Technol., Guangzhou
Abstract :
Bid format is a crucial feature of electricity market mechanism. Correspondingly, in agent based electricity market simulation, how to model agent´s bid format and represent action domain is also an important aspect to construct a valid learning method. In this paper, two methods to generator agents´ action domains are presented and their effects to the simulation result are analyzed on a 4-generator system. The result shows that in agent-based simulation, agent action domain representation can strongly affect the simulation result. With all other parameters the same, different action domain generating methods lead to different average clearing prices which means different capability of executing market power.
Keywords :
learning (artificial intelligence); multi-agent systems; power engineering computing; power markets; agent action domain representation method; agent-based electricity market simulation; learning method; reinforcement learning; Costs; Electricity supply industry; Laboratories; Learning systems; Microeconomics; Power generation economics; Power system dynamics; Power system economics; Power system interconnection; Power system modeling;
Conference_Titel :
Power and Energy Engineering Conference, 2009. APPEEC 2009. Asia-Pacific
Conference_Location :
Wuhan
Print_ISBN :
978-1-4244-2486-3
Electronic_ISBN :
978-1-4244-2487-0
DOI :
10.1109/APPEEC.2009.4918312
