Date of Graduation


Document Type


Degree Type



Statler College of Engineering and Mineral Resources


Mechanical and Aerospace Engineering

Committee Chair

Mario Perhinschi.


Research efforts have increased in recent years toward the development of intelligent fault tolerant control laws, which are capable of helping the pilot to safely maintain aircraft control at post failure conditions. Researchers at West Virginia University (WVU) have been actively involved in the development of fault tolerant adaptive control laws in all three major categories: direct, indirect, and hybrid. The first implemented design to provide adaptation was a direct adaptive controller, which used artificial neural networks to generate augmentation commands in order to reduce the modeling error. Indirect adaptive laws were implemented in another controller, which utilized online PID to estimate and update the controller parameter. Finally, a new controller design was introduced, which integrated both direct and indirect control laws. This controller is known as hybrid adaptive controller.;This last control design outperformed the two earlier designs in terms of less NNs effort and better tracking quality. The performance of online PID has an important role in the quality of the hybrid controller; therefore, the quality of the estimation will be of a great importance. Unfortunately, PID is not perfect and the online estimation process has some inherited issues; the online PID estimates are primarily affected by delays and biases. In order to ensure updating reliable estimates to the controller, the estimator consumes some time to converge. Moreover, the estimator will often converge to a biased value. This thesis conducts a sensitivity analysis for the estimation issues, delay and bias, and their effect on the tracking quality. In addition, the performance of the hybrid controller as compared to direct adaptive controller is explored.;In order to serve this purpose, a simulation environment in MATLAB/SIMULINK has been created. The simulation environment is customized to provide the user with the flexibility to add different combinations of biases and delays to the explored derivatives. Biases were considered in the range -500% to 500% and delays in the range 0.5 to 40 seconds. The stability and control derivatives considered in this research effort are a combination of decoupled derivatives in the three channels, longitudinal, lateral, and directional. Numerous simulation scenarios and flight conditions are considered to provide more credibility to the obtained results. In addition, a statistical analysis has been conducted to assess the results. The performance of the control laws has been evaluated in terms of the integral of the error in tracking the three desired angular rates, pitch, roll, and yaw. In addition, the effort of the neural networks exerted to compensate for tracking errors is considered in the analysis as well.;The results show that in order to obtain reliable estimates for the investigated derivatives, the estimator needs to generate values with less than five seconds delay. In addition, derivatives estimates are within 50% or -15% off the exact values. Moreover, the importance of updating derivatives depends on the maneuver scenario and the flight condition. The estimation process at quasi-steady state conditions provides reliable estimates as opposed to estimation during fast dynamic changes; also, the estimation process has better performance at large rate of change of derivatives values.