Abstract
The present thesis addresses the methods for solving the control allocation problem and their applications. The topic finds ready applications in various engineering fields like aerospace, robotics, ground vehicles and many other safety critical systems. These systems use re- dundant actuators to improve system performance, reliability and reconfigurability. But coordination of those actuators and their control is a challenging task. Few methods for distributing the control commands among the actuators in the presence of saturation and failure of actuators have been proposed in this thesis.
First, an unconstrained control allocation method is taken up that involves finding a solution to an underdetermined system. The problem is solved using a system theoretic approach where a linear closed-loop system is designed. The convergence rate of the proposed method that can be tuned is comparable to existing numerical methods.
Next, two methods are proposed to solve an constrained control allocation problem that handles the saturation limits more conveniently. The allocation problem is framed as a fixed point problem and solved sequentially. Further, a Newton based method is proposed to solve the same formulation. Through several examples, the convergence properties of the proposed methods are compared with existing methods.
The Newton based method is followed by a lookup table based direct allocation method, which usually requires negligible computation. But in the event of actuator fault, normally new lookup tables are required for allocation, and thus the advantage is lost. A new allocation method is proposed that uses pre-fault tables and fault information to allocate command among healthy actuators. The method is used for fault tolerant control.
Lastly, two controller design methods are proposed for a class of overactuated systems in order to meet a level of disturbance attenuation. For overactuated systems, two types of designs are possible. Either a controller is designed to generate actuator command directly, or a controller-allocator pair is designed where the allocator distributes control command among actuators. It is shown that under normal operation both these design methodologies can meet the same disturbance attenuation level. But the allocator based method performs
