The Research On Power Supplier Price Forecasting And Bidding Strategy

With profound reformation undergoing in electricity market, the research of bidding strategies for generators is significant. This thesis studies with emphasis how a dynamoelectric enterprise in market background of the power plant and grid is separated should scientifically constitute bidding strategy, participate in market completion and obtain maximum income.In the first and second part, this thesis introduces not only the history of development of power market and its mode, but also the formation and characteristics of current structure of China's power market. And then analyzes the basic bidding ways of dynamoelectric side.In the third part, Fuzzy Theory is used for price forecasting. In traditional method of exponential smoothing, only before datas were used for calculating, and brings the lagging to data sequence. To eliminate this issue, an exponential smoothing method based on Fuzzy Theory is advanced. The forth part is about the arithmetic of ANN based on BP model, namely the advanced of traditional BP arithmetic. One alterable step and scale BP arithmetic based on comparability of model and probability of accepting BP arithmetic is used, to enhance the convergence rate of learning process of BP network, also avoid the stagnation problem to some extent. It indicates that the ANN'S efficiency and precision by this way can be ameliorated by the simulation of real data.In the fifth part of this paper least squares support vector machines is applied to forecast the price. The local minima problem is solved in support vector machine. So support vector machine is more robust and accurate for. forecasting and it is considered to be the substitution of artificial neural network. Least squares support vector machine is an expansion of standard support vector machine. It is faster and easier to use. The forecasting result of least squares support vector machine model is better than the artificial neural network under the same training sample and input vector.In the sixth part, the thesis puts up about the bidding strategy of dynamoelectric companies based on Game Theory. Standard forms of Cournot model and Bertrand model were analyzed. Standard form of Coumot model was extended. The competitive strategy models based on the lowest cost of power generation and the most of power generation were specified at the same time. Then a Cournot model with generators' capability constraints under complete information was established. On the basis of this, a Cournot model with generators' capability constraints under incomplete information was established. The model under incomplete information was specified in two situations: Generation companies have some estimate functions of other generation units' cost; Generation companies have a distributing function of other generation units' cost-estimate. Then the methods to obtain Cournot-Nash equilibrium states of all kinds of Cournot models above were introduced. When getting the optimal solution of the output of generation units, the units bidding curves can be obtained. From the units bidding curves, the units bidding strategies can be obtained, stimulation and analysis was illustrated.Finally, a practical bidding strategy of generator and its issues are discussed. Then the paper summarizes the use of the methods advanced before and uses game strategy to make the stepwise bidding strategy more scientific.