Abstract
Three-level T-type converters (3LT${}^{\mathbf{2}}$Cs) provide inherent redundancies that can be used to improve system-level reliability, where fault-tolerant control (FTC) techniques are the enabling algorithms. Existing FTC methods primarily focus on compensating redundant vectors and adjusting dwell times, while neglecting the impact on modeling and control of grid-side currents and neutral-point (NP) voltage under fault conditions. This work formulates a unified fault-induced state space model that facilitates the construction of cost functions in finite-control-set model predictive control (FCS-MPC). The modeling is based on the investigation of the possible cases of single-switch and double-switch open-circuit faults, and the analysis of additional terms introduced by the faults. The proposed FTC method adopts a dual-loop control structure, i.e., the outer loop uses a proportional integral (PI) controller to automatically increase the reference value of the dc bus voltage during vertical faults and converts it into the grid current reference, while the inner loop combines fault-induced grid-side current and NP voltage cost functions in FCS-MPC, adjusting weighting factors based on fault conditions to prioritize grid-side current quality. Experimental results demonstrate the proposed FTC under various open-circuit fault scenarios.

This publication has 31 references indexed in Scilit: