Temporal Dynamics of Vegetation Ignition With Hidden Markov Model Volatility Decoding

This is the authors' abstract. We don't add key points for this paper.

This research investigates the temporal dynamics of vegetation ignition during branch-touching-wire earth faults. The study applied image processing to study the relationship between electrical fault current and fire ignition by analyzing the temporal dynamics of ignition across species using data from controlled High-Impedance Fault (HIF) tests. The findings suggest large variability in ignition behaviour across species, influenced by the moisture content. A. Mearnsii and P. Radiata were the species with the quickest and slowest flame breakouts, consecutively. 10% of the sustained flame breakouts occurred within just 11.19 s of the branch touching the line. This research also provides insights into the required current sensitivity levels for early fault detection. A key research finding is that 89% of vegetation earth faults can be timely detected, before sustained flame ignition, if a 0.4 Armsearth-fault sensitivity can be achieved. This study also presented a novel methodology for decoding HIF volatility by integrating a volatility index using the gradient of the RMS current and a Hidden Markov Model (HMM). Analyzing the temporal volatility of RMS current, this method enables potential early detection of incipient arcing events that precede flame ignition.