Researches On Optimal Scheduling Of Power Systems With Large-scale Wind Power Integration

With the increasing trend of global climate change and the greenhouse effect,countries around the world are paying unprecedented attention to the development of renewable energy.Under the new situation,China has also put forward the goal of achieving the carbon emission peak and carbon neutrality successively around the middle of this century.The achievement of the double-carbon goal puts forward higher requirements for the proportion of wind power,photovoltaic and other renewable energy in the power systems.On the one hand,as renewable energy sources are volatile,intermittent and stochastic in nature,and as their installed capacity continues to rise,the output of renewable energy sources will fluctuate randomly over a wider range.It is particularly important to better model the uncertain output of renewable energy sources.On the other hand,the penetration rate of renewable energy in the power systems is increasing year by year,and the traditional operation and scheduling methods are facing great challenges in the consumption of renewable energy,and how to fully take into account the uncertainty of renewable energy output in the optimal scheduling model also needs to be addressed.In recent years,with the rapid development of Io T technology,system operators have been able to grasp huge amounts of wind power and photovoltaic output data.If the valid information in the historical data is fully exploited and applied to the operation and scheduling of the power systems,it will be of great value for both the day-ahead unit commitment(UC)problem and the intra-day economic dispatch(ED)problem of the power systems.Based on this,this paper focuses on the optimal scheduling of power systems under large-scale wind power integration,with the aim of providing practical suggestions for the scheduling of the systems,which mainly consists of the following four parts:1)Firstly,to address the problem that the existing robust uncertainty sets are not effectively constructed to exploit the value of historical data,an enhanced uncertainty set for the data-adaptive robust unit commitment(UC)model is developed.The cumulative distribution function(CDF)is estimated using the imprecise dirichlet model(IDM)driven by the wind power historical data to obtain the output interval of wind power at a certain confidence level,which is used as the upper and lower bounds for the wind power output in the robust uncertainty set.A linear constraint of 0-1 auxiliary variables for positive/negative fluctuations of different wind farms is introduced into the uncertainty set,so that the fluctuations of wind farms in the same area are limited to a certain perturbation range,thus ensuring the synchronous fluctuation characteristics of neighbouring wind farms.Under the min-max-min three-level robust framework,the model is decomposed into a two-stage optimization problem with a master problem and subproblems using the C&CG algorithm and iterating over each other to find the optimal solution.Numerical analysis verifies that the proposed model can make full use of the valid information of wind power historical data and improve the practicality of the scheduling scheme.2)Next,a data-driven distributionally robust unit commitment model is proposed to address the problem that traditional robust scheduling decisions are too conservative due to their focus on the worst-case scenarios of wind power.The Dirichlet Process Mixture Model(DPMM)is used to classify the multi-modal wind power prediction error dataset into multiple subsets with different probability distribution characteristics and the corresponding weights,based on which an ambiguity set of probability distribution considering the first-order moment information of random variables and the Wasserstein distance is constructed.The physical meaning of the worst-case probability distribution of random variables is explained intuitively through mathematical derivation,and a two-stage distributionaly robust optimisation algorithm based on the cutting plane method is proposed to find the optimal unit commitment solution that can withstand the worst-case probability distribution of wind power.The case study verifies that the proposed method can greatly reduce the conservatism of traditional robust optimization and improve the economics of the scheduling scheme.3)Then,a stochastic economic dispatch model for power systems that considers the correlation between different wind farms is developed to address the problem that the wind power stochastic scenarios selected in traditional stochastic economic dispatch models tend to ignore the correlation between different wind farms.The Pair-Copula method is applied to construct a probability density function(PDF)of multi-dimensional variables to capture the tail correlation between different marginal distribution functions(MDFs),and to generate a large number of wind power scenarios considering the correlation of wind power to characterise the uncertainty of wind power output.The mean and variance of the generation costs under different wind power scenarios are used to measure the economic cost and risk cost of the scheduling scheme,so as to construct a multi-objective economic dispatch model that takes into account the expected costs and economic risks,and the optimal Pareto frontier is obtained through the group search optimizer with multiple producers(GSOMP).The proposed model is validated to be effective in considering the correlation between different wind farms and provides a more reliable scheduling scheme for the power systems.4)Finally,to address the problem that the classical mean-variance model ignores the value of wind power prediction scenario in the application of the stochastic economic dispatch problem,a stochastic economic dispatch method considering wind power uncertainty based on the mean-tracking error model is proposed.The arrangement of the generator units under the wind power prediction scenario,namely,the pre-schedule,is obtained through the deterministic scheduling model,which is then considered in the multi-objective optimisation model in the form of tracking errors.By minimising the errors between the generation cost under different wind power scenarios and those of the pre-schedule,the optimal scheduling solution can be found.Numerical analysis verifies that the proposed method can absorb the uncertain wind power output and make as few adjustments to the pre-schedule as possible,effectively exploiting the value of wind power prediction scenario and improve the practicability of the scheduling scheme.This paper provides a research basis for the optimal scheduling of wind power output under different degrees of uncertain information,and extends the application of data-driven methods based on historical data in uncertain optimal scheduling of power systems,which is of great value for the improvement of the actual operation of power systems.