Dynamic State Estimation for a Back-to-Back VSC-HVDC Sending-End System Under Short-Circuit Faults: An Event-Triggered Model-Matching Approach

Short-circuit faults alter the algebraic network relations of a back-to-back voltage-source-converter high-voltage direct-current (VSC-HVDC) sending-end system, creating structural mismatch when a dynamic state estimator retains the normal-network model. This paper develops an event-triggered model-matching approach within a single unscented Kalman filter (UKF). A calibrated innovation criterion combines with given fault information to activate the matching shunt coefficient and update its covariance treatment. Measurement correction uses the updated statistics, and subsequent predictions incorporate the fault-network relation through an explicit network update. Two fault-reactance cases, Xf=0.1 and 0.2p.u., are evaluated using reference trajectories and synthetic measurements generated by the reduced-order sending-end model. The proposed method yields lower RMSEs than fixed-model and covariance-adaptive UKFs for the four reported synchronous-generator states during fault and recovery intervals. For Xf=0.2p.u., the rotor-angle RMSE over the [1.0,1.2)s fault interval is 0.0132rad, compared with 0.5411 and 0.3320rad for the two comparators, respectively. These results show how matching the network relation within the UKF improves fault-period estimation and recovery in the studied sending-end equivalent.