Data-Driven Robust Passivity-Based Controller Design for Grid-Connected Converters

Due to the integration of renewable-based Distributed Generation Units (DGUs), Energy Storage Systems (ESSs) and electric car charging stations, more and more power electronic converters are connected to the grid. In grids with a high penetration rate of these converters, high current oscillation may happen which can damage the grid equipment or cause a blackout. Passivity theory is used to design converter controllers in order to alleviate this problem in a decentralized fashion. However, the proposed methods are mostly model-based and are sensitive to parametric uncertainty. This paper presents a robust data-driven frequency-domain grid-connected converter controller design with the possibility of defining robust passivity conditions on a performance channel. In addition to robust pas- sivity, the controller design guarantees tracking and disturbance rejection performance. All performance and passivity conditions are written as constraints on closed-loop sensitivity functions, then changed to frequency domain inequalities and convexified around an initial stabilizing controller. The controller design problem is converted to a convex optimization problem that can be solved efficiently using available solvers. The proposed method is validated by a Power-Hardware-in-the-Loop (PHIL) experiment including a switching inverter, real-time simulator and LCL−filter.