Suboptimal synthesis of a robust decentralized control for large-scale mechanical systems (Q2266685)

A large-scale mechanical system is modeled as a collection of n actuators described as \(\dot x_ i=A_ ix_ i+b_ iu_ i+f_ iP_ i\), which are coupled through a set of nonlinear relations \(P_ i=q_ i(x_ 1,x_ 2,...,x_ n;d)\), where d is a vector of system parameters. A set of nominal trajectories is prescribed together with a region of allowable parameters such that each nominal trajectory is achievable by a nominal decentralized control in the absence of coupling (i.e., when \(P_ i=0)\). The aim of the paper is to design additional control to stabilize the overall system around each of the nominal trajectories in the presence of coupling and for all allowable parameter values. An iterative procedure is established to choose unique linear local controllers, which is based on a LQ optimization approach. If all nominal trajectories can not be accomodated by local controllers alone, a force feedback is implemented as a global control to compensate for the effect of coupling.