Beyond the Short-Circuit Ratio: Control-Aware Online Assessment of Dynamic System Strength via Gray-Box Identification of Converter Dynamics

Conventional network-only short-circuit ratios (SCRs) do not quantify converter-control dependence or the reliability of a measurement-derived dynamic model. This paper combines an EMS network prior with structured converter identification, a return-difference margin, conditional uncertainty propagation, and event-dependent reporting. The unsigned margin is interpreted only with an independent stability check and a specified port normalization. The original IEEE 39-bus study is extended to include 1800 independent matched-model records, 1400 deliberate-mismatch trials, 1200 threshold-test records, repeated control changes, and coupled models at all three converter buses. Matched-model ellipsoid projections show 99–100% scalar-margin coverage, whereas direct scalar intervals show 91.5–97%, including one undercovered condition. A 0.1 ms relative measurement-channel delay gives zero ellipsoid coverage despite a median point error of 0.502%, establishing a measurement-calibration limitation. Three-converter coupling reduces the baseline margin from the independent-model value of 0.676 to 0.580. Under the same disturbance, two nominally stable PLL tunings produce voltage norms of 0.00818 and 0.01184, distinguishing their compliance with a specified 0.01 response limit. An event-triggered shorter-memory update combined with a complete post-event observation-window restart recovers all 24 tested changes within 12 s. The results support reliability-aware small-signal monitoring across the stated topologies and controller variants, while retaining explicit limitations on model mismatch, finite-band interpretation, and transfer to field or nonlinear converter behavior.