Research On Network Attack Detection And Control Strategy Of Integrated Energy System

With the increase of intelligent measurement equipment in the integrated energy system(IES),the problem of information security in the process of data transmission has become increasingly severe.The falsification of measurement data and the randomness of predicted values greatly increase the difficulty of optimal scheduling of IES.Although the integration of hydrogen equipment can improve the utilization efficiency of new energy in IES,the inflammability and explosiveness of the hydrogen infiltration puts forward higher requirements for the operation of IES.Therefore,the dual effects of cyber attacks and hydrogen infiltration have brought a series of difficulties and challenges to the safe and stable operation of IES.In this context,the potential impact of false data injection attacks(FDIA)on the IES is first analyzed,and corresponding detection methods are proposed.For hydrogen penetrated integrated energy system(HPIES),a mathematical method for optimal scheduling and a method based on machine learning are proposed respectively.Finally,the intra-day optimization method of HPIES is studied,and a short-term optimization response method for network attack scenarios is proposed to ensure the economy and stability of the system in the attack scenario.The main work of this paper is as follows:(1)The impact of FDIA on the integrated energy system is analyzed,and the transmission characteristic of network attacks between different energy networks is studied.A FDIA detection method for IES based on machine learning is proposed.Aiming at the problem of high missed detection rate of this method when the measured value modification range is small,a FDIA detection method based on cold and hot network simulation is proposed.A dual detection mechanism for FDIA in IES is formed,which is based on machine learning and supplemented by cold and hot network simulation.A detailed HPIES equipment and network model is established.Three models of mixed-integer linear programming,conditional value-at-risk and robust optimization are used for day-ahead economic dispatch.The results of calculation examples show that both conditional value-at-risk and robust optimization can withstand part of the impact of uncertainty.Using the characteristics of long and short-term memory artificial neural networks that are suitable for processing time-related problems,a date optimization method for HPIES based on machine learning is proposed,and a control scheme is generated by learning historical scheduling plans.This method is superior to the mathematical programming method in terms of economics and intraday adjustments under high uncertainty scenarios,and is suitable for day-ahead economic dispatch of HPIES.(3)On the basis of the previous optimization method,the intraday control method and defense strategy of HPIES under the attack scenario are proposed.Through global constraints,the day-ahead scheduling plan and the intra-day economic scheduling are bound to form a day-ahead-rolling-short-term three-tier control plan.The global constraints are passed from the day-ahead scheduling plan to the short-term optimization through rolling optimization,which can guarantee the optimal global economy of the system within a day under normal circumstances.When an attack is detected,the operating curve of the key equipment of the system can be continuously revised by the global constraint to achieve the purpose of resisting FDIA.Finally,combined with the FDIA detection framework,a short-term control strategy of HPIES considering the FDIA is proposed.