A Functional Time Series Framework for Probabilistic Health Monitoring on a Hexacopter: Experimental Evaluation via a Series of Flight Tests

This study presents a statistical framework for detecting and quantifying damage in multicopter propeller blades through functionally pooled (FP) time series models. The methodology integrates model-based analysis with statistical time series techniques, employing FP models characterized by parameters that are explicit functions of damage magnitude. The framework is experimentally validated using a hexacopter executing figure-eight flight patterns at constant velocity and altitude under turbulent conditions. Damage sizes in the form of blade trimming ranging between 2 mm and 10 mm are investigated using data from the on-board inertial measurement unit (IMU). The proposed approach achieves both effevtive damage detection and precise magnitude estimation using short data segments (4 seconds) of single-channel measurement signals obtained during flight. Results demonstrate accurate damage size predictions with 99% confidence bounds across multiple signal sources, including linear accelerations and angular rates. The framework’s effectiveness is comprehensively evaluated through statistical analysis of 400 independent test data sets and damage estimations per damage state using statistical analysis.