Strategies for dealing with false data injection attacks in power grid measurement based on unsupervised learning

Under the attack of false data injection, existing power data recovery methods often suffer from problems such as poor data quality and weak self-healing ability to restore a normal power grid. This article innovatively proposes a strategy for responding to false data injection attacks in power grid measurement based on unsupervised learning. Firstly, based on the basic principles of generative adversarial network (GAN) algorithm and particle swarm optimization (PSO) algorithm, an improved generative adversarial network (IGAN) algorithm model and an improved particle swarm optimization (IPSO)algorithm model are constructed, respectively. Then, considering the correlation between power grid measurement data and their own time series characteristics and using the physical topology information of the power grid as specific labels, an improved method for recovering power grid measurement data using generative adversarial networks is proposed. Finally, simulation analysis verifies that the strategy proposed in this article can effectively solve the problem of false data caused by injection attacks on the power grid.