Unit Commitment Optimization Strategy Considering Cyber-attacks And Wind Power Uncertainty

Uncertain factors in the power system are becoming more diverse,more complicated and more difficult to quantify due to the increasing consumption of distributed renewable energy and the rocketing number of smart meters,whic h will certainly bring new challenges to the safety and economy of the power system operation.Among these uncertain factors,the uncertainty of wind power in the physical layer and the uncertainty of cyber attacks in the cyber layer are the major security threats to the power system.However,there is little focus on the detection of wind power scenario attacks and the methods of preventing power system day-ahead scheduling model from being attacked by abnormal wind power scenarios.For this foreseeable problem,this paper will first construct new methods to reduce the probability of wind power being tampered from wind power scenario detection method and wind power scenario reduction method.And then establish a two-stage wind power safety consumption inter val optimization model based on the tested wind power forecasted scenarios.The complexity and solving process of the traditional security-constrained unit commitment model are finally simplified on the base of the obtained wind power safety consumption interval.Firstly,wind power simulation samples are generated by a trained generative adversarial network to enhance the diversity of initial datasets,so as to improve the accuracy of the abnormal scenario detection algorithm based on decision tree and the n prevent attackers from influencing the quality of wind power simulation scenario and day-ahead unit commitment model by tampering with wind power forecasted scenarios.Secondly,an improved wind power scenario reduction method considering spatial-temporal correlation and cyber attacks is proposed aiming to select the scenarios that obey the physical characteristics while deleting the attacked abnormal scenarios and ensure the quality of the simulation scenarios.Moreover,principal component analysis and a clustering method are adopted in our method to ensure the calculating speed and the diversity of selected final scenarios.Thirdly,based on the wind power forecasted scenarios,a wind power safety consumption interval optimization model is constructed by integrating the proposed slack line power flow constraint and dual wind power constraints into a wind power consumption capacity evaluation model,which is to obtain the precise wind power consumption space coordinating with the actual conditions of the power system operation next day and compress the abnormal fluctuation range of wind power.Finally,the traditional security-constrained unit commitment model is simplified based on the obtained wind power safety consumption interval and the number of unce rtain constraints in the model are reduced.Meanwhile,two sub-models are used to modify the day-ahead scheduling results to reduce the redundant reserve capacity of the power system and the on-off operation frequency,improving the economy of the operation.In this paper,two wind power scenario processing methods are improved to reduce the probability of wind power scenario being attacked,and the effectiveness and feasibility of these improved two methods are verified by the historical data of two wind fields.Then,based on the obtained non-abnormal wind power scenarios,the simulation experiments are carried out on the 6-node system,IEEE 30-node system and IEEE 118 node system with several wind farms.The experimental results demonstrate the feasibility and economy of the wind power safety consumption interval optimization model,the simplified security-constrained unit commitment model and the two auxiliary sub-models.Moreover,the proposed optimization strategy can make the improved SCUC model superi or to the deterministic unit commitment model in both wind power consumption and operation cost.