Fault tolerant control allocation for a thruster-controlled floating platform using parametric programming

The task in control allocation is to allocate a specified generalized force to a redundant set of control effectors where the associated actuator control inputs are constrained, and other physical and operational constraints and objectives should be met. In this paper we consider a convex quadratic approximation to a control allocation problem for a thruster-controlled floating platform with eight rotatable azimuth thrusters where the high level controller is assumed to specify three generalized forces; surge, sway and yaw. The optimization problem is solved explicitly by viewing the generalized forces as a vector of parameters and utilizing parametric programming techniques, leading to a continuous piecewise affine (PWA) function implementing the optimal control allocation. Experimental results from a test basin with a 1:100 scale model of a platform are presented. It is shown how thruster and machinery failure scenarios can be handled by automatic reconfiguration of the control allocation, exploiting symmetry of the thruster configuration.