Research On Detection And Defense In Smart Grid GPS Spoofing Attack

Phasor Measuring Unit(PMU)are considered as an important part of smart grid,where the PMUs are feasible to collect accurate electric power data.However,GPS Spoofing Attack(GSA)will compromise some PMUs and cause the PMUs collecting false data,which could incur some generators trips and some other bad consequences.Simultaneously,it is difficult for many previously proposed schemes to detect the slow continuous GSA.So as to solve the above problems,this paper focus on detection and defense in smart grid GSA.Firstly,this paper proposes an adaptive generalized accumulation GSA detection scheme based on the adaptive sliding window and generalized accumulation algorithm.Secondly,considering the low data repairing accuracy in many previously proposed GSA defense schemes,this paper proposes a GSA protection scheme based on BiLSTM and self-attention generative adversarial network,where bidirectional long-and short-term memory network and self-attention are employed so as to improve the accuracy on repairing data.The details are described as follows:(1)To guarantee the timeliness on detecting the slow continuous GSA and improve detection accuracy,this paper proposes an adaptive generalized accumulation GSA detection algorithm,where adaptive sliding window is employed to collect electric power data,and then an improved Kalman filter is employed to estimate the state of our proposed system model.Finally,generalized accumulation algorithm is employed to estimate the probability of being attacked so as to detect and distinguish different types of GSAs.Further,we test and analyze our algorithm on detecting mutation,slow continuous and hybrid GSAs,which are simulated on a standard IEEE-39 node network.The experimental results show that our proposed scheme is feasible to reduce the computation costs,the mean-squared errors and improve the detection rate while detecting the above three GSAs,compared with other previously proposed schemes.(2)To improve the data repairing accuracy and reduce computation costs,this paper proposes a GSA protection scheme based on BiLSTM and self-attention generative adversarial network,where a GSA detection model is employed to estimate PMUs collected electric power data and eliminate false data,and then an improved WGAN-GP data repairing model is employed to deal with the remaining data so as to complete(repair)the collected electric power data.Finally,PMUs transfer the repaired electric power data to control center.Further,we test and analyze our scheme on detecting GSAs,which are simulated on a standard IEEE-39 node network.The experimental results show that our proposed scheme is feasible to improve the detection rate on detecting GSAs,improve the AUC scores and improve the data repairing accuracy,compared with other previously proposed schemes.