Strategic bidding of ancillary services for a hydro power producer

摘要

This paper presents an agent-based simulator for examination of a secondary control market dominated by hydro power producer as decision support for one of the market participants. Proposed is a Q-learning algorithm for determining possible strategic behavior. Adaptive learning is made possible by application of certain characteristics to agents quantity price pairs bids. Considered are for each agent its portfolio of different hydro power plants with their water values estimated by a stochastic dynamic programming scheme. The simulator is applied to the Swiss system where strategic behavior will be shown. Additionally it is analyzed how single agents could make use of strategic behavior in case of special occurrences in the market.