MINLP Modeling for Detection of SCADA Cyberattacks in Water Distribution Systems

Abstract

In recent years, automated systems of supervisory control and data acquisition (SCADA) have been widely used in water distribution networks, which have made the distribution system vulnerable to cyber-Attacks. One of the main limitations of the previously proposed cyberattack detection models is that tampering the SCADA's pressure readings would not be reflected on the estimated nodal demands. As a result, cyber-Attacks caused by such manipulations would not be detected. To address this challenge, this paper incorporates pressure readings in the detection algorithm as a constraint and develops a mixed integer nonlinear programming (MINLP) formulation to estimate the nodal demands of a network, as well as its flow rates and nodal pressure readings. These values are estimated using the SCADA readings for link flows, pump's status, tank's water level, and total demand of a district metering area. Nodal pressure heads are calculated using Bernoulli's and Hazen-Williams equations and pump's head equation.