Robust Optimization And Computaional Analysis Of Power System Operation With Integrated Wind Power Generation

In recent years,since the environment gradually deteriorates and haze frequently occurs,emission reduction is imperative.As the most mature form of clean renewable energy,wind generation has been vigorously developed.Along with the installed capacity increasing year by year,there are more and more prominent adverse effects of highly uncertain wind power on the power system operation,which makes traditional scheduling and operation state analysis methods less effective.Therefore,it is necessary to study the optimal decision-making and computational analysis method for the power system operation,aiming at the uncertain characteristics of wind power.The paper focuses on the following issues,and obtains corresponding results:1)Uncertainty modeling of wind power forecast.It is difficult to accurately predict wind power.The errors have various sources and it is difficult to model them separately.Probability model based on statistical methods is a popular choice.The real and forecast values of wind power exhibit complex conditional dependence and spatial-temporal correlation,as well as non-Gaussian marginal distributions.Due to these obstacles,the simplified pair copula with flexible dependence structures is adopted to model the joint distribution of real and forecast values of wind power.The joint distribution can be converted to the conditional model of real wind power with respect to wind power forecast,which well fits the statistical characteristics of wind power.The proposed model updates the probability model of actual wind power according to the renewed forecasts,and reduces the conservatism consequently.Generated samples can be applied to optimization and computational analysis of the power system operation.2)Power system robust unit commitment considering the situation where the wind power can be completely absorbed.Under the premise of safe power supply,the power system pursues economic operation.The general ellipsoidal uncertainty set well fits correlated uncertain variables and is adopted to describe the correlated wind power.Using the affine policy and general ellipsoidal uncertainty set,the power system robust unit commitment model is constructed and transformed into a mixed integer second order cone programming that can be readily solved by off-the-shelf solvers.Then,feasibility in the scenarios out of the uncertainty set is discussed and probabilistic guarantees on the feasibility is used to decide the parameter of the uncertainty set.Finally,the proposed method reduces the conservatism of the solution by selecting appropriate uncertainty set and adjusting its parameter.3)Robust unit commitment that improves the admissible region of wind power.When wind power cannot be totally absorbed,especially during the heating period,the primary concern is to optimize the capability of absorbing wind power,namely the admissible region of wind power.Besides ordinary constraints of power systems operation,the cogeneration system robust unit commitment model considers additional thermoelectric coupling characteristics of heat units and thermal equilibrium.The proposed model adopts affine policies and sets apart admissible region from available region of wind power.The model aims at improving admissible region of wind power.This is a novel robust optimization problem with unknown uncertain interval uncertainty set.Such robust model is converted to a mixed integer linear programming.The proposed method fully exploits the regulating capacity of power systems to improve the admissible region and guarantees the robustness against any variations within the admissible region.4)Stochastic power flow calculation using multivariate dimension reduction.Since the stochastic power flow problem has high dimensionality,deduced multivariate dimension reduction decomposes the original stochastic power flow problem into two lower dimensional stochastic power flow subproblems.Then,the proposed method gives the formulae relating the solutions of the original problem to that of subproblems.The computation time is significantly reduced and is only proportional to the number of wind farms while the high accuracy remains.The proposed method also utilizes the multi-machine power balancing strategy.As a result,the proposed method shrinks the variation ranges of generators' outputs and can resist large power disturbance.