A first-hitting-time framework for risk-based adaptive tripping and prognostics in gas-insulated switchgear

Abstract

Gas-insulated switchgear (GIS) protects itself against dielectric breakdown by tripping when the SF6 pressure falls below a fixed level. This logic cannot distinguish a slow leak from a rapid pressure loss, and a single threshold forces a compromise: set high, it de-energises the equipment unnecessarily during chronic leaks; set low, it cannot guaran-tee that the pressure remains above the critical level by the time the breaker completes its operation during a fast leak. This article proposes a trip rule in which the breaker operates once breach of critical level within the protection chain's reaction-time budget becomes near-certain under a conservative upper bound on the estimated leak rate. The method cou-ples choked-orifice leak physics with an interacting multiple-model estimator of pressure and leak conductance, operating on temperature-compensated measurements; the probabil-ity of first hitting the critical level within the reaction budget is available in closed form through the inverse-Gaussian first-passage distribution. The proposed method is compared with four threshold-crossing baselines in a five-way ablation study across leak severities. The same running estimator additionally supplies a short-term breach-risk probability and a long-term remaining-useful-life estimate with an 80% prediction interval. Of the five methods, only the proposed one keeps the pressure above the critical level at breaker com-pletion in every tested leak class while avoiding unnecessary tripping, and its remaining-time interval attains 80–100% empirical coverage of its nominal level. A first-hitting-time criterion thus provides uniformly safe operation and accurate prognostics from a single es-timator, where threshold-crossing rules must trade one against the other.