Gonenc Gursoy

Concurrent Learning Enabled Adaptive Limit Detection for Active Pilot Cueing

Neural-network-based adaptive dynamic models are commonly used to estimate allowable control travel and the proximity to a limiting flight condition in the design of advanced envelope protection algorithms for fly-by-wire aircraft. In this paper, linear models are compensated with adaptive neural networks, which use instantaneous sensor data as well as past flight history information for concurrent learning. A law for collecting appropriate training data into the history stack is established. It is observed that using the proposed time history data for online neural network training provides more accurate dynamic trim and control limit predictions compared to using instantaneous sensor data only. Simulation results for a fixed-wing aircraft during maneuvers show comparisons between the different adaptation schemes.

Non-Iterative Adaptive Limit and Control Margin Estimation with Concurrent Learning

In this paper, the adaptive neural network based online limit and control margin estimation algorithms for envelope protection are improved using a non-iterative limit margin estimation methodology. The fixed point solution assumption is removed. Functional relations between the fast aircraft states and the control inputs are generated online using concurrent learning neural networks with guaranteed signal bounds. Estimates of the optimal adaptive weights are obtained through concurrent adaptation using minimum singular value maximization for recording necessary data. The allowable control travel not to exceed an imposed flight envelope limit is estimated using control sensitivity estimations. This information can be used to cue pilots or limit controller commands to ensure a safe flight. A nonlinear aircraft model is used to show the effectiveness in simulation.

Engine Limit Detection and Avoidance for Helicopters with Multiple Limits

In this paper, adaptive neural network based approximate models with concurrent learning schemes are used to estimate approaching limits of multiple parameters of a helicopter engine, namely the Turbine Gas Temperature and the Gas Generator Speed. The collective control channel is used to estimate the corresponding allowable control travel in order not to exceed an imposed limit. This information can be used to cue pilots or to design control systems that ensure a safe flight. Simulation results are obtained through a high fidelity simulation environment with coupled helicopter and engine dynamics.

Helicopter Slung Load Simulations Using Heli-Dyn+

This paper demonstrates a fast and convinent way to model the dynamics of a helicopter with a slung load. A new software tool called Heli-Dyn+ is used to generate high fidelity helicopter math models. One of the benefits of Heli-Dyn+ is that helicopter math models can be exported as dynamic libraries into different software environments. In this paper we develop a generic 3-DOF slung load dynamic model in MATLAB/Simlulink. Then the slung load model is integrated with a nonlinear 6-DOF helicopter model exported from Heli-Dyn+. Simulation results showing longitudinal and lateral responses of the combined model are presented for various flight scenarios.

Direct Adaptive Limit and Control Margin Estimation with Concurrent Learning

In this paper, two vital signals to enable flight envelope protection, namely the onset to the flight envelope (limit margin) and the available control travel to reach the limit boundary (control margin), are estimated using improved noniterative adaptive neural-network-based approximate models. The adaptive elements use current and past information (concurrent learning) and have guaranteed signal bounds. Minimum singular value maximization is used to record data for concurrent learning. Results showed better convergence properties of the network weights compared with results in the literature in which only the current data is used for network weight updates. Two methods are introduced to calculate control margins from approximate models. None of the introduced methods require iteration and therefore remove previously introduced assumptions related to iteration convergence. A nonlinear fixed-wing aircraft model is used to show the effectiveness in simulation for estimating limit and control margins and av...

Kalman Filter Based Modification on Helicopter Adaptive Control

In this paper, a Kalman Filter based e-modification term is used in the update law of an online learning adaptive neural network controller for a helicopter application. The controller is applied to a high fidelity helicopter simulation model. The applications include both an example on a single-input single-output controller structure using a linear in the parameter neural network, but also a multi-input multi-output neural network structure to compensate uncertainties in the attitude control of a helicopter. It is observed that using a Kalman Filter based e-modification provides improved command following and requires less control effort compared to traditional e-modification.

Long Term Learning Adaptive Neural Network Estimator Based Limit Detection

Abstract Dynamic adaptive models are commonly used to estimate allowable control travel and the proximity to a limiting flight condition in the design of advanced envelope protection algorithms for fly by wire aircraft. In this paper linear models are compensated with adaptive neural networks to build adaptive models of relevant aircraft dynamics. A stack of data collected during flight is used to update the network weights online. The data stack is made up of instantaneously measured data and recorded data during simulations. It is observed that by using recorded data in a stack can cancel out new modeling errors in a short time and results with better predictions of approaching limits compared to using instantaneous data only.