Maneuver recognition and prediction of empennage flight loads of general aviation aircraft

Abstract

An improved method based on maneuver recognition using neural networks is presented for analyzing and predicting empennage flight loads of a general aviation aircraft. Linear accelerometer, angular accelerometer, rate gyro, and strain gage signals were recorded during flights for dutch-roll, roll, sideslip, level-turn, and pitch-up maneuvers. Sensor signals were filtered and used to train and validate the network. Strains in the horizontal tail spar were predicted successfully to within 50 us of their strain gage values resulting in significant improvements over previous results. This method can be used with low-cost flight data recorders to: 1) analyze flight maneuvers; 2) analyze flight loads of general aviation airplanes; and 3) establish empennage load spectra of airplanes already in-service where installation of strain gages is impractical.