| An important step in control design for elastic systems is the determination of the number and location of control system components, namely sensors. The number and placement of sensors can be critical to the robust functioning of active control systems, especially when the system of interest is a large high-speed aeroelastic vehicle. The position of the sensors affects not only system stability, but also the performance of the closed-loop system. In this dissertation, a new approach for sensor placement in the integrated rigid and vibrational control of flexible aircraft structures is developed. Traditional rigid-body augmentation objectives are addressed indirectly through input-output pair and compensation selection. Aeroelastic control suppression objectives are addressed directly through sensor placement. A nonlinear programming problem is posed to minimize a cost function with specified constraints, where the cost function terms are multiplied by appropriate weighting factors. Cost function criteria are based on complex frequency domain geometric pole-zero structures in order to gain stabilize or phase stabilize the aeroelastic modes. Specifically, these criteria are based on dipole magnitude and complementary departure angle. In turn, the control design approach utilizes one of the classical methods known as Evans root migration to exploit the pole-zero structures resulting from sensor placement. Desirable complementary departure angles can lead to significant aeroelastic damping improvement as loop gain is increased, while favorable dipole magnitudes can virtually eliminate the effects of aeroelastics in a feedback loop. Appropriate constraints include minimum phase aeroelastic zeros to avoid common problems associated with right-half plane zeros. To achieve desirable flight control system characteristics via optimal sensor locations, different kinds of blending filters for multiple sensors are investigated. Static filters, as well as dynamic filters with fixed or variable parameters and fixed or variable compensator parameters, are considered. For every cost function, there are several local minima indicating many distributions of the sensors are available. By evaluating the cost for each minimum, the global optimum can then be found. |
