Benchmark of machine learning algorithms on capturing future distribution network anomalies

The conventional distribution network is undergoing structural changes and becoming an active grid due to the advent of smart grid technologies encompassing distributed energy resources (DERs), aggregated demand response and electric vehicles (EVs). This establishes a need for state estimation-based tools and real-time monitoring of the distribution grid to correctly apply active controls. Although such new tools may be vulnerable to cyber-attacks, cyber-security of distribution grid has not received enough attention. As smart distribution grid intensively relies on communication infrastructures, the authors assume in this study that an attacker can compromise the communication and successfully conduct attacks against crucial functions of the distribution management system, making the distribution system prone to instability boundaries for collapses. They formulate the attack detection problem in the distribution grid as a statistical learning problem and demonstrate a comprehensive benchmark of statistical learning methods on various IEEE distribution test systems. The proposed learning algorithms are tested using various attack scenarios which include distinct features of modern distribution grid such as integration of DERs and EVs. Furthermore, the interaction between transmission and distribution systems and its effect on the attack detection problem are investigated. Simulation results show attack detection is more challenging in the distribution grid.