Design and Simulation-Based Verification of a Training Instrumentation Framework for Long-Endurance Hybrid-Electric VTOL UAS Operations

Abstract

Long-endurance hybrid-electric vertical-takeoff-and-landing unmanned aircraft systems combine automated flight modes, energy constraints, payload control, and extended mission monitoring. Conventional training records commonly show completion outcomes but do not reconstruct how an operator recognized an event, selected a procedure, and maintained the aircraft within a safe envelope. This study develops a product-neutral training instrumentation framework that converts time-aligned telemetry and operator-event records into reproducible performance evidence. The framework monitors flight mode, navigation integrity, command-and-control status, energy state, environmental limits, payload state, acknowledgments, checklist actions, and recovery outcomes. A four-component score represents procedural accuracy, response timeliness, flight-envelope compliance, and safety outcome, with non-compensatory gates for critically weak procedural, envelope, or safety evidence. Functional verification used 1,600 reproducible seeded synthetic traces across eight abnormal-event scenarios. The framework classified 95.4% of traces consistently with their injected labels, with a macro-averaged F1 score of 0.946. Additional tests examined missing measurements, bounded noise, weight and threshold changes, and deliberately conflicting component patterns. A competency audit covered 87% of the independently defined weighted requirements and identified assembly and configuration, post-flight data integrity, and field maintenance as requiring separate assessment. Because the injected labels determined the generating distributions, the agreement result verifies internal implementation consistency rather than predictive validity. The results do not demonstrate training effectiveness or human performance.