Adaptive structure detector for linear implicit systems

The synthesis of the descriptor variable reconstructor requires the full knowledge of which internal structure is active, that is to say, if the system is either of first order or second order with or without the dominant zero; i.e. we have to synthesize an internal structure detector. We present an adaptive structure detector based on a normalized gradient algorithm projected along a given hyper-sphere, whose convergence in finite time is proved. We propose the reconstructor and give simulation results and concluding remarks