Wind Power Forecasting And Its Application In Optimal Joint Operation Of Wind-Energy-Storage Systems

China’s wind power installed capacity is the first in the world.The large-scale wind power integration increases the difficulty of dispatching power systems to improve the economic dispatching capacity of wind farms and reduce the conservativeness of power system planning decisions.The output of wind farms is characterized by randomness,volatility,and unschedulability,which makes wind power forecasting more difficult.In addition,it increases the difficulty of economic dispatching of the power grid and poses a serious threat to operate the power system safely and stably,which attracts great attention of experts and scholars worldwide.Nevertheless,the domestic and international research on wind power forecasting and its applications in the joint optimization operation of wind-thermal-energystorage systems are still under study.Based on the above research background,this thesis focuses on the wind power foreacasting and its optimization on the operation of wind-thermal-energy-storage systems.From the perspectives of the wind power forecasting,the statistical distribution,and the windthermal-energy-storage system application,this thesis reviews the state-of-the-art of wind power forecasting and its domestic and foreign research status of the problems of optimizing the operation of wind-thermal-energy-storage systems.The wind power forecasting method based on the k-division multi-modeling is studied.The application of Gaussian odel distribution in forecasting error data analysis is also studied.A stochastic optimization simulation platform is proposed and used to optimally place and fully adjust the energy storage device.The platform has established an optimal scheduling model with wind power generation and energy storage system considering the opportunity cost constraint to improve the utilization efficiency of wind power in the integrating connection process.Regarding the wind power forecasting method,the k-division multi-modeling method is combined with the particle swarm optimization algorithm for NARX system identification.Firstly,the observation data of wind power is divided into k sets.Secondly,the current optimal kernel function is obtained by the particle swarm optimization algorithm.The kernel function is suitable for the k data sets.Finally,the relevant criteria is used to terminate the algorithm to obtain the optimal model.The simulation results show that the proposed algorithm can generate robust models with better generalization performance.And simulation experiments has demonstrated the performance of the proposed algorithm.With regards to the statistical analysis of wind power prediction error,a wind power forecasting error distribution based on the Gaussian mixture model is studied.The expectation maximization algorithm is used to analyze the wind power load forecasting error data from the statistical point of view,which is theoretically proved.The rationality and advantage of this method is that regardless of its statistical distribution,the probability density function of all wind power forecasting errors can be approximated by using a Gaussian mixture model.Then,appropriate submodel reductions are made.The effectiveness of the Gaussian mixture model in the statistical analysis of wind power forecasting error is proved.With regards to the simulation of wind storage systems,a stochastic optimization model is introduced to optimally place and fully adjust the capacity of the energy storage device to minimize the cost of running and interrupting the load.The probability density function is used to characterize the uncertainties of the power system,including the wind power,the load,and the equipment availability.The proposed method can be used to evaluate the reliability and operability of wind power integration and can increase the available capacity of the energy storage device.The economics of the energy storage device and the cost of its electrical service reliability are incorporated into the overall cost-benefit analysis of the system and used to determine the most economical plan.The simulation results show that the wind power combined power output model can effectively improve the overall performance of the wind turbine in the process of grid connection and proposes an optimal solution that can increase the wind power penetration rate and the reliability of grid connection,that is,the energy storage needs to be placed.Finally,an economic scheduling model platform considering opportunity cost constraints is developed for the wind storage joint operation system.In order to maximize the overall economic benefits of power system operation and minimize the waste air volume of wind power generation,an optimal dispatching model including wind power generation and energy storage system is established,and then the sequential linear programming method is used to calculate and solve the proposed problem,so as to obtain the optimal dispatching operation scheme.Simulation results verify the correctness and effectiveness of the proposed method.When the wind power permeability is 25.98%,the operation cost saved by using Gaussian mixture distribution is about 25.05%.When the wind power permeability is about 21.65%,the operation cost saved by using Gaussian mixture distribution is about 14.51%.