Machine Learning-Based Cyber-Attack Detection in Photovoltaic Farms

Abstract

In this paper, a machine learning technique is proposed for the detection of cyber-attacks in Photovoltaic (PV) farms using point of common coupling (PCC) sensors alone. A comprehensive cyberattack model of a PV farm is first developed to consider operating conditions variability. The attack model specifically includes two types of cyber-attacks that are historically more difficult to detect. A Convolutional Neural Network (CNN) using µPMU plus figures of merit is proposed and compared with other machine learning techniques using raw electric waveform and micro-phase measurement units (µPMU), respectively. Finally, a cyber-physical security testbed of an IEEE 37-bus distributed grid with PV farms is developed. A real-time simulation, detection, and visualization framework is designed to demonstrate the feasibility of the proposed method in a real-world application. Results show that the proposed machine learning methods can achieve adequate detection accuracy and robustness under various attack scenarios.INDEX TERMS Grid-tied power electronics converters, photovoltaic farms, cyber-attack detection, waveform, micro-phasor measurement units. Nomenclature

Smart Grid Resilience for Grid-Connected PV and Protection Systems under Cyber Threats

“…In addition, this study also examines the impact of PV irradiance within normal operation on the voltage and current of a PV system, which exhibit variations throughout the day. Table 1 presents a detailed summary of recent studies that enhance smart grid resilience for cyber threats with DER systems [16][17][18][19][20][21][22][23][24]. In general, the studies [19][20][21] focused on studying the impact of cyber attacks on power protection systems, while [5,15] investigated the impact of cyber threats on microgrid control systems.…”

“…The IEDs in smart grids exchange messages using the GOOSE protocol to provide high-speed communication for exchanging critical information. Critical events in power system protection and inverter control applications, such as inter-trip and blocking, need to be delivered in less than 3 ms of time [18]. Therefore, GOOSE messages are directly attached to the data link layer using the CSMA/CD (Carrier Sense Multiple Access with Collision Detection) protocol.…”

Cyber-physical testbeds for smart grid and electric vehicle-charging infrastructure

“…This may possibly lead to performance degradation or instability during data availability attacks. TSA detection scheme [12] and FDIA detection [13], focus solely on attack detection and necessitate a training phase to comprehend the system's normal behavior. Relying heavily on extensive training data introduces complexities such as increased memory usage, substantial data prerequisites, and privacy concerns.…”