Distributed control for the smart grid with communication constraints and cyber-security

Abstract

In the last few years networked control systems (NCS) have attracted the attention of the research community due to wide number applications that include communication networks into the feedback control loop to connect spatially distributed elements. Sensor measures and control actions are sent through private or shared communication links, offering many advantages such as reduced system wiring, low installation and maintenance costs, and increase on flexibility and adaption capability. One of the main examples of an NCS is the smart grid, which uses the flow of information of the different states of the power network to take smart decisions and improve the reliability and minimize power losses. In this work, we focus our attention on controlling physically isolated elements in the power grid, e.g., microgrids. We propose a control strategy based on the consensus algorithm that uses the frequency information transmitted through the communication network in order to assure frequency synchronization of several isolated microgrids. We use linear and nonlinear models in order to verify the capabilities of our controller, and we show that, under certain conditions, synchronization is achieved independently of the sampling period at which information is transmitted. Sending information through communications channels makes the system vulnerable to cyber-attacks, where an adversary may tamper sensors or actuators in order to access to private information and impact the power system, causing economical losses or even blackouts. A direct consequence of the sampled information property of our proposed controller is data minimization, which allows to improve privacy in the smart grid, and decrease or hide information from adversaries. Moreover, we study false data injection attacks, and we show how some detection mechanisms can minimize the impact of undetected attacks. We study the effects of injecting false information in a higher level control system, i.e., demand response and real time pricing, and we analyze the impact of injecting attacks at a certain frequency using sensitivity analysis. All our results are verified via simulation