| With the acceleration of advancing process of power marketablization and the implementation of operation mechanism of " Separation of power plant from power network, competitive price for listing power network" power generating enterprises have become the main body of market competition. Under the power market environment, how hydropower plant participates in market competition, optimizes production process of power generation and adopts the tactic price offers to achieve the maximum profits of power generation has become the foremost problem concerning with its survival and benefits.Owing to the complexity of hydropower plant operation, little research work has been done on hydropower plant, but much has been done on the optimal operation of the competitive price for listing power network of the coal-fired power plant in the country. Based on the analysis of the existing conditions, directions and difficult points of power market theory research both home and abroad, this thesis studies the following aspects with the prediction of marginal electricity prices in power system and the competitive price tactics as the main contents:(1)The artificial neural network method is adopted to predict the marginal electricity price in the system.The electricity price prediction model is established on the basis of similar searching and weighted regression. The effectiveness of the method is proved via the actual prediction, and the effect of network learning specimen upon the predicted results is analyzed.The second-day marginal electricity price prediction model is established on the basis of the similar searching and neural network. The high non-linear fitting capacity of neuralnetwork is used to achieve the rapid response to data variations; and the neural network is trained through the improved BP algorithms and variable learning speed reversion spreading algorithms, whereby the prediction accurateness of work-day marginal electricity price has been greatly improved.Also, the prediction model of week-end marginal electricity price has been improved so as to shorten the number of concentrated training days, to reduce data of work days in input items, to increase the historical data of nearing weekends, to lower the prediction interference, whereby the prediction accurateness of weekend electricity price has been improved. (2) The chaotic theory is adopted to predict the marginal electricity price in the system.With an aim at the changeable characteristics of electricity price in power market in our country, the chaotic performances of time sequences of electricity prices and loads can be used to re-construct the space of accurate electricity consequence phase. Accordingly, the electricity price prediction model based on fast BP network is established via tracing the evolution tendencies of neighboring points of phase space so as to carry out the prediction of electricity price of Chuan-Yu power network, with the satisfactory results obtained.The phase space reconstruction theory of chaotic dynamic system, in combination of the non-linear reflecting and pan-capacity of neural network, can be used to establish the prediction model; and a kind of new prediction method is suggested to realize the complete tracing of phase point evolution process and to predict "the price nails", whereby improving the prediction accurateness and effectively solving the problem of negative prediction, with the satisfactory results obtained.(3 )The advantages and disadvantages of various error analysis methods in electricity price prediction are analyzed; and the medium value relative error is recommended to evaluate the accurateness of electricity price prediction method. The effectiveness of error analysis method has been proved via the analysis of computing examples.(4) With an aim at the hydropower stations with different types and different regulating capacities, various kinds of constraints should be taken into account based on the second-day marginal electricity price prediction. With the great benefits of power generation as the objective, the optimal operation model should be established so as to carry out the optimal computation of reservoir operation. It is proved via the analysis and contrast of actual calculating examples that the benefits of power generation in hydropower stations are obviously improved.(5) This thesis analyzes the power market structure and operation mechanism in our country, particularly with the price mechanism of power market in power generation, trading patterns and competitive price patterns, and also describes the formation mechanism of electricity price and the components of power generation costs and further analyzes the cost performances of power generating sets, on the basis of which, this thesis analyzes the competitive price tactics of the clearing price of power generating plant based on the marketprediction, the competitive price taclics based on the price offering behaviors of other power generation firms, the competitive price tactics based on the game theory and other optimal methods. Accordingly, some hydropower plant with the yearly maximum profits as the objective is analyzed and calculated so as to obtain the optimal structure tactics of electricity quantities in the case of the competitive price for listing power network.(6) This thesis studies the structures of decision-making supporting system and the corresponding functions of the competitive price for listing power network of hydropower plant, and carries out the designs of structure, network connection and functional model blocks of decision-making supporting system of the competitive price for listing power network of hydropower plant.
|
PDF Full Text Request