HELICOPTER FLIGHT DATA FEATURE-EXTRACTION OR COMPONENT LOAD MONITORING

Helicopter night data are analyzed using univariate and multivariate t echniques to extract features of relevance for rotor system component load prediction. The vibratory component of four rotor system loads is examined; main rotor pushrod, rotor blade normal bending, lag damper, and main rotor shaft bending load. Univariate relationships between t hese loads and fixed system parameters are examined and basic trends a re highlighted, Multivariate approaches including multiple linear regr ession and artificial neural network analyses are utilized to create l oad prediction models. Fixed system parameters form the basis of the m odels and include pilot control positions and aircraft state parameter s. Generally, the loads can he predicted quite well during steady leve l night, as well as moderate and high-g flight where fatigue damage is most likely to occur. Low speed and hovering night and flight conditi ons with low engine torque are the most difficult night conditions for accurate load prediction. Significant parameters in the regression an d neural network models are identified and several night regimes are d efined that can be used to improve load prediction accuracy.