Adaptive Kalman Filter-based Information Attack Detection In Smart Grids

As a new generation of power grid,smart grid uses modern information technology to realize the integration of power and information.Smart grid can deliver power in a more effective way.At the same time,information and communication technology improves the efficiency of the system,because of the complexity of the information physical system,it also makes the system more vulnerable to malicious internal personnel or hostile competitors launch network attacks,resulting in serious consequences.This paper studies the detectability of smart grid attack under structural vulnerability and the detection of smart grid under false data injection attack.The main work of this paper is divided into two parts as follows:In the first part,in view of the situation that the smart grid system with structural vulnerability suffers from information attack,this paper analyzes the attack detectability of the attacked system with the help of graph theory from the perspective of structural control theory.On this basis,the necessary and sufficient conditions for simplifying the topology of complex power system are given.Based on the simplified graph,the necessary and sufficient graph theory conditions are proposed to solve the problem of attack detectability.In the second part,the attack detection algorithm of adaptive Kalman filter is proposed to detect the attack when smart grid system suffers from false data injection attack.For the case of nonlinear system and unknown noise,an adaptive square root unscented Kalman filter with adaptive ability to noise and stable filtering is designed to realize state estimation.Based on the obtained state estimation value,a random variable is constructed by using the generalized central limit theorem,and a bilateral hypothesis test method is proposed according to the statistical characteristics of the random variable.The detection algorithm is relaxed.For the limitation of the nature of noise in the system,it is not necessary to make the noise be linear Gauss assumption,and analyze and compare the algorithm in this paper and the existing algorithm from the angle of false detection rate.In this paper,the attack detection problem is studied from the perspective of topologyand system state of smart grid,and the attack detection problem is solved with a more complete framework.