Identification of nonstationary models with application to myoelectric signals for controlling electrical stimulation of paraplegics

Abstract
It has been shown that the estimates of the nonstationary identifier proposed by A. Kitagawa and W. Gersch (1985) do indeed exist for all inputs because the formation is uniformly observable with probability one, and that the identifier is stable because the formulation is uniformly controllable. Some of the complicating factors concerning their nonstationary identification algorithm are clarified to establish its optimality and stability. Among these are the nonlinear and time varying nature of the formulation. It provides proofs that this nonstationary identifier's estimate exists, is stable, and is optimal for Gaussian noise inputs and is also optimal over a limited class of identifiers for non-Gaussian noise inputs and mean squared error loss function. Experimental results are included which demonstrates the superior performance of the nonstationary identifier over a piecewise stationary identifier operating on nonstationary electromyographic data.