Study On A Novel Multi-objective Robust Unit Commitment Considering Temporal Correlation Of Wind And Peak Load Regulation Ability

With the increasing penetration of wind power,the volatility and uncertainty of the renewable resources brings significant challenges to unit commitment(UC)for power systems.Under this condition,power system operators have to increase the reserve margins to balance the demand and supply of the net load without violating the operating constraints of both generating units and transmission networks.In order to deal with the uncertainty of wind,it is essential to choose an appropriate model for uncertain factors to select the units and their outputs in the system,such as stochastic optimization(SO),fuzzy programming,robust optimization(RO)and so on.RO is independent on prior knowledge and does not need to assume probability distribution.So,it has become an effective tool for achieving reliable solutions under uncertain circumstances.However,wind power output often has the opposite of peak load regulation,and power system operators can not change output plan of wind with the human needs.Therefore,the robust unit commitment,which only considers the worst scene of wind power,is often due to the lack of peak regulation capacity,resulting in the phenomenon of abandonment of wind and load cutting.In addition,with the integration of large-scale wind power especially in coastal area of China,peak-valley difference of load continues to increase,which exacerbates the difficulty of peak load regulation.So,the peak load regulation is another key issue in UC that hasn't been paid enough attention,and it affects the security and economy of the system.Accordingly,the further study of robust optimization has important significance with consideration of the wind power plants' participation in peak load regulation when unit commitment is considered.Based on this,the following work has been done in this paper:(1)In order to ensure the robustness of the system,the objective function of peak regulation capability is established in this paper.By the method of non-parametric kernel density estimation,the upper and lower limits of wind power prediction are obtained,and the upper and lower limits of equivalent load of system are calculated.According to the fluctuation of equivalent load power,the demand of peak load regulation in the system is obtained under the worst scene,and then the objective function of the worst-case reduction peak capacity is established.(2)In order to consider the uncertainty and characteristics of peak load regulation of wind,the capacity of peak load regulation is taken as an objective function of robust unit commitment.At the same time,a multi-objective robust unit combination model considering the peak regulation of the system is established.(3)In order to improve the economy of robust optimization,we make full use of historical data of wind power to restrict the uncertainty of model in this paper.Firstly,the existence of time correlation of wind power output is proved by using Pearson correlation coefficient.Then,the scatter plot of the prediction error of wind power output is drawn,and it can be seen from the scatter plot that the closer the prediction time is,the stronger the wind power correlation.Therefore,only the correlation of wind p prediction error in two adjacent periods is considered in this paper.Finally,the probabilistic theory is adopted to constraint the above-mentioned model with temporal correlations of wind power in this paper,and the model of multi-objective robust optimization for unit commitment considering temporal correlation(MRUCT)is proposed to improve economy in a long run.(4)Then,a novel algorithm to solve MRUCT is put forward.It will decompose the problem into two stages.The first stage is to decide the start and stop state of the unit as a multi-objective optimization problem.The Pareto frontier of this multi-objective optimization solution can be obtained by normalized normal constraint(NNC).Then,the fuzzy mechanism is used to obtain the compromise solution.The second stage is to solve a "min-max" problem that determines the outputs of the specific units.This sub-problem is transformed into a mixed integer optimization problem by the dual linearization.(5)Finally,the simulation based on IEEE 118 node system verifies the effectiveness of the proposed model and algorithm.