Finite horizon noninteraction and fault detection through almost controllability subspaces

The structural conditions for noninteracting control are extended so as to provide an effective tool to single out and handle the cases where noninteraction can be guaranteed for a finite time rather than for an infinite time. On the assumption that the extended conditions hold, finite horizon noninteraction is achieved through feedforward dynamic units also including finite impulse response systems. The design procedure is strictly geometric and exploits the basic properties of controllability and almost controllability subspaces. The dual counterpart in the context of fault detection and isolation introduces a structural means to identify and treat the cases where the residuals which can be generated are significant only in a limited period