Flight control system design of F-16 aircraft using robust eigenstructure assignment

In this thesis, the design of a modern flight control system using robust eigenstructure assignment for F-16 aircraft is studied. In the first chapter, the literature reviews of eigenstructure assignment and stability robustness are presented. The developments of the stability analysis beginning from SISO to MIMO approach are presented in terms of Nyquist theorem and singular value theory. The effect of singular value theory in the design and analysis of the eigenstructure assignment is emphasized. In the second chapter, F-16 aircraft mathematical model is introduced. The mathematical model includes six main components which are aerodynamic model, equations of motion, athmosphere model, engine model, actuator model and sensor models. Aerodynamic model is based on the wind tunnel test results of NASA Langley Research Center. In the third chapter, trim and linearization algorithms are derived. Trim algorithms are constructed for three flight conditions which are straight and level flight, climb and descent and coordinated turn. The search algorithm for trim operation is based on the Nelder-Mead optimization. Linearization algorithms are developed by decoupling the aircraft motion into two as longitudinal and lateral-directional. Longitudinal motion is further decoupled into two as short-period and phugoid aproximation. Similiarly, lateral-directional motion is decoupled into two as dutch-roll and roll-spiral approximation. These approximations are achieved by using the wind axis and stability axis state variables. In the fourth chapter, design of longitudinal CSAS is presented. The design technique is based on pole assignment. Pole assignment is performed by controller form decomposition. Five longitudinal CSAS are designed by pole assignment technique. The first three uses angle of atttack, pitch rate and normal acceleration feedbacks as command variables consecutively. The last two are the modified versions of pitch rate and normal acceleration CSAS which don't use angle of attack feedback. Control systems are designed in a fashion that the resultant closed loop characteristic polynomials will be the same. Hence, the eigenvalue sensitivity and stability robustness comparison are performed with the same design requirements. The effectiveness of the designed controllers are illustrated by nonlinear simulations. In the fifth chapter, design of lateral-directional CSAS is presented. Two design techniques are investigated which are non-robust and robust eignstructure assignment. Non-robust eigenstructure assignment is evaluated with two different eigenvector structures. The control law synthesis is performed by controller form decomposition which computes non-orthonormal null space basis vectors. The optimization cost function of the non-robust design is constructed in a way that it only includes mode decoupling properties of the system. Robust eigenstructure assignment is evaluated with a single eigenvector structure but with two different design methodologies. Both design methodologies are based on the orthonormal null-space basis vectors. The first one is design with fixed eigenvalues. This technique uses a fixed eigenvalue set and searchs for the optimal robust solution with four different cost functions. The first two cost functions are selected for eigenvalue snsitivity minimization and the last two are for stability optimization. Second design methodology is design with flexible eigenvalues. This technique mixes the eigenvalue locations into the optimization problem. The same cost functions are used as in the previous method and the feedback matrices are computed with a single step optimization procedure. Hence, the eigenvalue sensitivity and stability robustness comparison are performed for all the computed feedback matrices. Furthermore, the effectiveness of the designed controllers are illustrated by nonlinear simulations. In the last chapter, high angle of attack control problem is evaluated. A well-known design methodology is presented as in nonlinear dynamic inversion. Nonlinear dynamic inversion is applied at two steps which are slow and fast inversions. Nonlinear dynamic inversion and robust eigentructure assignment is compared with two high angle of attack maneuvers. These maneuvers are carried on with three different angle of attack values and the results are illustrated graphically.