PSAT-FMP: An asset-aware fire monitoring platform for critical transmission infrastructure

Wildfires pose a recurrent risk to transmission infrastructure, particularly in areas with dense vegetation. Existing satellite based fire information services support broad area monitoring, but their outputs are not always suitable for infrastructure oriented assessment, especially in the presence of persistent industrial thermal sources or spatially localized events. This study presents the PSAT-FMP, an asset aware fire monitoring system designed for transmission infrastructure protection through sensor specific satellite workflows and context aware alert generation. The platform combines NOAA-20 VIIRS and Sentinel-2 A MSI imagery with georeferenced infrastructure data, exclusion zones, and proximity based severity classification within an AWS based operational workflow, while the fire detection component is implemented with a U-Net architecture and an EfficientNet-B7 backbone. In Minas Gerais, Brazil, the deployment generated 226 fire alerts associated with 56 monitored transmission lines. On the evaluated datasets, the segmentation models achieved test IoU values of 0.90 for NOAA-20 and 0.94 for Sentinel-2 A. Targeted comparative assessments relative to FIRMS further showed that the proposed workflow can reduce industrial thermal responses propagated as alerts and identify localized fire related events through the Sentinel-2 A branch in situations not represented in coarser broad scale fire products. These results indicate that asset aware fire monitoring can provide operationally relevant support for transmission infrastructure protection. The study contributes an applied framework that combines sensor specific fire detection, spatial contextualization, and severity classified alert generation within a unified monitoring workflow.