Convex Optimization-Based Fault Location Combining Perspective Relaxation and Iterative Reweighting in Multiterminal DC Systems

Abstract

Accurate fault location in multiterminal DC (MTDC) systems is hindered by topological ambiguity and measurement synchronization uncertainties. This article presents a convex optimization-based fault location framework characterized by low computational burden and global optimality. By introducing a continuous perspective relaxation strategy, the nonconvex combinatorial search problem is reformulated into a strictly convex rotated second-order cone programming (RSOCP) model. This guarantees global optimality, overcoming the vulnerability of traditional analytical methods to synchronization errors, while bypassing the local minima traps of heuristic algorithms and the extrapolation vulnerability of data-driven models. To enhance robustness, a Cauchy M-estimator-based iteratively reweighted least squares mechanism is integrated, allowing the framework to autonomously suppress synchronization noise and extreme measurement deviations. The RSOCP model is solved via the primal-dual interior-point method. Dynamic evaluations on four-terminal and 13-terminal meshed MTDC systems verify the framework’s accuracy and reliability even under low sampling frequencies, high fault resistances, and adverse noisy conditions.