Research On Modeling And ADP Algorithm For Unit Commitment Problem In Power System With Wind Turbines

The unit commitment stochastic problem of generating units with wind turbines is to determine the "on" and "off" status and generate dispatch of each unit to fulfill the demands under the practical scheduling constraints and the uncertainty condition of wind power.The research can increase the proportion of grid-connected wind power,and reduce fossil energy consumption scientifically and effectively.It also reduces the cost,then improves the energy utilization.Therefore,the research on this problem has become one of the frontiers and hot topics.In the power scheduling system containing wind turbines,the number of wind turbines is large.If we model and solve the single wind turbine,the dimension of the model is so high,that the computation is very difficult and the computation time is rather long.Aiming at solving this problem,according to the batch division principle based on the location and generation parameters,we group the units into different batches to schedule,and aggregate the parameters.Then we use batch modeling strategy to reduce the dimension of the model.In view of the characteristics of the problem model,the approximate dynamic programming(ADP)algorithm is designed to solve the problem.Finally,a wind-thermal unit stochastic scheduling decision support system is developed.The main research contents are as follows:1)Considering the impact of the fluctuation and randomness of wind power on the power system,scenario probability is adopted to quantitatively describe the uncertainty caused by wind turbines generation and depict the possibility of future occurrences.At the same time,considering that the problem is difficult to solve due to the fact that the number of wind turbines is large and the dimension of the model is high,batch modeling strategy and corresponding parameter aggregation methods are proposed to reduce the complexity of the problem based on the batch division principle which makes use of the similarity of generation parameters.2)For the power system unit commitment scheduling optimization problem with wind turbines,the research mainly focuses on determining the unit commitment and power generation at each time,to achieve the minimum cost of the system’s operation,under the constraints such as the system load,spinning reserve,power generation capacity climbing ability,minimum start-up and shut-down time.Based on the scenario tree method,a batch-based mixed integer programming model is established,and the validity of the proposed model is verified by numerical experiments.3)In view of the characteristics of the problem model,two ADP algorithms are designed.The first is to adopt the myopia strategy,and the second employs the approximate value function policy to approximate the generation cost in the future.Through numerical experiments on random data,compared with CPLEX,the first strategy can get approximate solutions in a short time,while the second strategy can get better approximate optimal solutions compared with myopic strategies.The above results show the effectiveness of the proposed ADP algorithm.4)Based on above model and algorithm,the support system on wind-thermal units stochastic scheduling is developed.This thesis analyzes the needs of power system scheduling,designs the database structure,and develops the Web version of the front interface.Finally we realize the function of random scheduling of thermal units and wind turbines,to achieve a friendly interface display,and also simplify the operation of dispatchers.