Neural network based algorithms for comprehensive collective axis limit avoidance on rotorcraft

Military rotorcraft are constrained by a number of operational limits designed to protect the structural integrity and controllability of the aircraft. There is interest in developing advanced control systems that provide Carefree Maneuvering capability on future rotorcraft, so that pilots can fly the entire flight envelope without requiring excessive workload to monitor limits. Recent advances in numerical algorithms and computing technologies such as neural networks are valuable tools in the implementation of such a control system. The present work is focused on developing advanced prediction algorithms for a comprehensive collective axis limit avoidance system for rotorcraft. The system is designed to simultaneously prevent exceedances of the continuous and transient transmission torque limits, transient RPM limit and engine temperature torque limit. The system is also used to assist in One Engine Inoperative (OEI) recovery and autorotation. A dynamic trim estimation algorithm using offline trained neural networks and polynomial curve-fits for torque prediction is used to calculate constraints on the collective stick for continuous torque limits. Different configurations of torque prediction are evaluated for possible implementation in existing and future rotorcraft A peak response estimation algorithm based on linear model of aircraft dynamics is used for transient limit prediction. The system is tested using a high fidelity, non-linear simulation model of the UH-60A Black Hawk helicopter. Results were generated for a variety of maneuvers in non-real-time simulations using a simple feedback controller to simulate the pilot. A stick limiting method that combines the precision of Automatic Flight Control System (AFCS) limiting with the limit over-ride capability of tactile cueing is proposed.