Hide/Show Apps

Simulator based evaluation of adaptive envelope protection algorithms for active sidestick controllers

Ünal, Zeynep
In this thesis, a simulator environment with an active control system is developed for testing different force feedback maps for flight envelope limit avoidance. Previously developed flight envelope protection algorithm; named direct adaptive limit margin estimation method is improved with Single Hidden Layer Neural Network. Neural network based adaptive models are developed online using concurrent learning algorithm for weight update laws. Concurrent learning method uses both current data and recorded past data for adaptation. In this study, a Linear Parameter Neural Network and a Single Hidden Layer Neural Network are utilized and compared. The performance of single hidden layer neural network estimates are found to be more accurate for model error compensation. Three different force feedback maps are designed for pilot cueing with active side stick. Proposed force maps are tested on simulator environment. Performance of different force maps found to be dependent on type of limit parameters. Simulator based tests are conducted for rotorcraft model load factor limit avoidance and fixed wing aircraft load factor and angle of attack limit avoidance.