Modeling Of Wind Power Uncertainty And Its Application In Power System

To address increasing environmental pollution and energy storage problems,wind power generation is developed and utilized extensively worldwide.However,it challenges the operation,dispatch,and planning of power system a lot,due to the intermittent and volatile nature,especially when it is accessed by the bulk power system.The characterization of wind power random character is exactly one of basic and imperative works to deal with these problems.Thus,this dissertation focuses on the modeling of wind power random characteristics and its application in power system.The major contributions include:1)A convex set named pair-wise convex hull is proposed to characterize the uncertain region of wind power.Firstly,the concept of convex hull is introduced.Then,the construction method of pair-wise convex hull is described.It is proved that the pair-wise convex hull contains the convex hull and is contained in the box set.Numerical studies show that this method can effectively reduce the computational burden of convex hull and is less conservative than other convex sets.In addition,the pair-wise convex can reduce the decision conservativeness of robust dispatch problems.2)A method based on the quotient gradient system(QGS)is proposed to characterize the dispatchable region of wind power.Firstly,the constraints based on the AC power flow equation is established.Then,the feasible region of these constraints is transformed into the regular stable equilibrium manifold of the quotient gradient dynamic system.We have proved the equivalence between the regular stable equilibrium manifold and the feasible region,and the global convergence of the quotient gradient system is also proved.Finally,the quotient gradient system is applied to fully describe the wind power dispatchable region.To speed up the calculation process,region expansion method is proposed to quickly depict this region.3)A method based on R-vine Copula is proposed to model the correlations between adjacent wind farms.Firstly,the historical data of the same period with similar power generation outputs are selected as inputs.Then,the optimal Copula function decomposition is determined by maximizing the summation of Kendall rank coefficient.At the same time,the optimal pair-Copula function is selected using AIC criterion.Simulation results demonstrate that the proposed model has higher flexibility and fitting accuracy than multivariate Copula,C-vine,and D-vine Copula based models.A probabilistic power flow algorithm based on R-Vine Copula model is proposed.Firstly,scenarios are generated by R-vine Copula model.Then,these scenarios are clustered using K-means method with distance constraints.Finally,the accurate power flow model based calculation is implemented on all cluster center senarios,and the remainings,cluster member senarios(cluster center senarios are excluded),are approximated by the method which is based on the model that retains the nonlinear part of Taylor expansion.Results of case studies indicate that this proposed method can address both complex correlations between variables and the tradeoff between the accuracy and the computation speed effectively.4)To improve the security of the system with wind power integrated,a rolling window economic dispatch model is proposed.A method to find the minimum number of controls to remove the overloaded power on the line is proposed.Firstly,an optimization model with minimum number of controls as the objective function is established.Then,a three-stage method with screening,ranking and detailed analysis is adopted to solve this prolem.Experimental results show that the three-stage method can quickly solve the large-scale nonlinear mixed integer programming problem(MINLP)and obtain the controls that meet the requirements of online operation.