A Possibility Theory Based Approach For Building Optimal Bidding Strategies In Electricity Markets

In the electricity market environment, bidding strategies employed by generation companies may have significant impacts on their own benefits, and on the operating behaviors of an electricity market as well. Hence, how to develop optimal bidding strategies for generation companies or how to analyze strategic behaviors of them and hence to figure out the potential market power abuse is now a very active research area.In recent years, a considerable amount of literature has been produced on developing optimal bidding strategies for generation companies. Most of research so far is based on utilizing estimations of bidding behaviors of the rival participants. However, this kind of methods hasn't been used in realized bidding support systems because the technology of how to analyze strategic behaviors of competitors is immature. A novel possibility theory based approach to develop optimal bidding strategies is proposed and systematically studied in this dissertation.First, a possibility theory based approach is proposed in this work for estimating bidding behaviors. Based on the concept of possibility distribution, linguistic expert knowledge that is intrinsically fuzzy is translated into an elastic constraint on the values that may be assigned to a competitor's bid. Hence, this approach has the ability to make full use of the information of a decision maker, and deal with the information fusion problem conveniently.Secondly, a fuzzy regression model is developed to indicate the possibility distribution of a competitor's bidding coefficient estimation. The result reflects the incompleteness and preserves the uncertainty of the original data.Thirdly, the possibility theory based model for building optimal bidding strategies is analyzed to find appropriate solutions under different conditions. Supposed a competitor's bidding parameters are noninteractive, the optimization problem is solved by fuzzy interval analysis. When this is not the case, a technique of modified fuzzy stimulation is made used of by integrating fuzzy stimulation and fuzzy interval analysis.Finally, fuzzy chance-constrained programming provides a powerful means to deal with the intractable constraints functions such as those of generation companies' output.The new framework to build bidding strategies proposed in this paper is especially suitable for those electricity markets recently launched since sufficienthistorical bidding data is not available and hence probability methods cannot be employed. The model is practicable and computationally efficient. Numerical examples that illustrate the proposed model and the solution technique developed are analyzed in detail.