Methodology for the Prediction of the Empennage In-Flight Loads of a General Aviation Aircraft Using Backpropagation Neural Networks

摘要

Backpropagation neural networks have been used to predict strains resulting from maneuver loads in the empennage structure of a Cessna 172P. The purpose of this research was to develop a methodology for the prediction of strains in the tail section of a general aviation aircraft that does not require installation of strain gages and to determine the minimum set of sensors necessary for these prediction that is suitable for small transport aircraft. ACE-100 provides accepted methods for estimating the fatigue life of empennage structures. It is based on airplane normal accelerations at the center of gravity (CG Nz). While there is a strong correlation between CG Nz and wing loads, little correlation exists with empennage loads. It was found that significant loads were induced in the horizontal and vertical tails during roll and dutch-roll maneuvers indicating that the fatigue load spectra derived using ACE-100 may significantly underestimate the fatigue cycles experienced by the empennage. This report provides a methodology for monitoring and measuring in-flight tail loads using Neural Networks. It does not require the installation of strain gages on each airplane. This is an inexpensive and effective technique for collecting empennage load spectra for small transport airplanes already in-service where installations of strain gages are impractical. Linear accelerometer, angular accelerometer, rate gyro, and strain gage signals were collected in flight using DAQBook portable data acquisition system for dutch-roll, roll, sideslip left, sideslip right, stabilized g tum left, stabilized g tum right, and push-pull maneuvers at airspeeds of 65 KIAS, 80 KIAS, and 95 KIAS. The sensor signals were filtered and used to train the neural networks. Modular Neural Networks were used to predict the strains.