A Decentralized Adaptive Architecture for Resilient Current Sharing in DC Microgrids Under Current Sensor Integrity Attacks

Effective current sharing among distributed generation (DG) units is crucial for microgrid stability, efficiency, and reliability. However, current sensors, essential for monitoring and control, might be subject to cyberattacks. This paper develops a novel resilience-enhancing framework based on a decentralized adaptive architecture for current sharing in DC microgrids under integrity cyberattacks in current sensor measurements. The proposed framework employs two key components for each DG unit; a resilient unknown input observer (UIO) and a dynamic virtual current sensor. The proposed resilient UIO and virtual sensors estimate the correct current from inaccurate current sensor measurements and baseline control inputs in the presence of uncertain and unknown load conditions and power lines. Our proposed approach consists of a mathematically rigorous input-output stability analysis. It is shown that the interconnection of the proposed framework with microgrids’ baseline controllers enables DC microgrids to adapt to and compensate for sensor uncertainties without compromising overall performance. Specifically, the augmented framework with the baseline microgrid controller proposed guarantees asymptotic stability of the closed-loop dynamical system under time-invariant current sensor uncertainties and uniform ultimate boundedness under time-varying current sensor uncertainties. Several case studies on a DC microgrid illustrate the efficiency and performance of the proposed resilient current sharing scheme. Note to Practitioners—This work addresses a critical cybersecurity challenge in DC microgrids: attacks on current sensors that can disrupt proper load sharing among distributed generation units and affect system stability. The proposed solution introduces a resilience-enhancing framework that works alongside existing microgrid controllers, requiring minimal modification to the installed setup. For each distributed generator unit, the framework adds two key components: a resilient observer that detects sensor anomalies and a virtual sensor that provides reliable current measurements even when physical current sensors are compromised. The main features of the proposed framework are: (i) it functions without requiring communication among distributed generators unit (decentralized design), it adapts automatically to changing load conditions (load-independent design), (iii) it does not require detailed knowledge of line parameters (graph-independent design), (iv) it maintains stable operation even under current sensor attacks (resilience), and (v) it works with existing microgrid secondary control systems (retrofit control). The implementation of the proposed framework requires only local measurements and control inputs, making it cost-effective and straightforward to deploy in both new and existing DC microgrids. The framework’s ability to handle both constant and time-varying sensor attacks provides robust protection against various cybersecurity threats while ensuring resilient power sharing.