Search for: All records

In this paper, a data-driven control strategy is proposed to regulate the dc bus voltage for permanent magnet synchronous generators (PMSGs) with an active rectifier. The proposed technique utilizes input/output data from a black-box model of the system, ensuring accuracy in predicting system behavior under the persistence of excitation of the input. A data-driven predictive control strategy is then formulated to generate the current references to an inner loop in order to accurately regulate the dc bus voltage. Control hardware-in-the-loop results of the PMSG and active rectifier for dc microgrid applications validate the effectiveness of the method and assess the feasibility of implementing the controller in hardware. 

DC microgrids have widely adopted hierarchical control architecture through distributed generation units (DGUs) to enhance reliability and scalability. However, this makes the system vulnerable to false data injection attacks (FDIAs), which can disrupt system stability or shift the operating point. While observers are commonly used to detect FDIAs, some FDIAs can be stealthy, or observers lack sufficient sensitivity for reliable identification. To address this, we propose a quickest change detection (QCD) method based on an unknown input observer (UIO) estimation error model to detect the FDIAs that are stealthy to the UIOs. The Ergodic CuSum algorithm is designed and can be efficiently updated using estimation error observations. The approach is validated through Simulink and real-time simulations. 

Not AvailableDC microgrid systems commonly feature a hierarchical control architecture with multiple interconnected distributed generation units (DGUs), requiring the integration of communication layers. This integration introduces a potential vulnerability, as malicious attackers can exploit the system by injecting false data, which could result in a shift in the operating point of the system or make the entire system unstable. To overcome this issue, this article proposes a data-driven unknown input observer (UIO) to detect and identify false data injection attacks (FDIAs) in the system. The data-driven UIOs are designed using only historical input/output data, which can be collected through simulations or experimental results. The developed UIOs do not require knowledge of the microgrid parameters. The proposed data-driven UIOs are then validated through Simulink and hardware-in-the-loop real-time simulation case studies to detect FDIAs in the secondary control of dc microgrids. The results show that the proposed observers can effectively detect and localize FDIAs in the communication links of the system. 

A lifting approach is proposed for the modeling and controller design of nonlinear power electronics converters with Constant Power Loads (CPLs). The analysis of converters with active loads is challenging due to the nonlinear CPL disturbance, which is modeled as a rational function. An extra state is proposed to transform the system into polynomial form more suitable for stability and controller design using linear matrix inequalities and sum of squares methods. The proposed approach ensures the lifted system represents the original model without approximation errors. Methods are then developed for the estimation of the region of attraction and controller design using sum of squares programming. Real time simulation results are presented to demonstrate the performance and feasibility of implementation of the proposed controller.