General Stochastic Convergence Theorem and Stochastic Adaptive Output-Feedback Controller

This paper is devoted to the analysis methods/tools to stochastic convergence and stochastic adaptive output-feedback control. As the first contribution, a general stochastic convergence theorem is proposed for stochastic nonlinear systems. The theorem doesn't necessarily involve a positive-definite function of the system states with negative-semidefinite infinitesimal, essentially different from stochastic LaSalle's theorem (see e.g., [1]), and hence can provide more opportunities to achieve stochastic convergence. Moreover, as a direct extension of the convergence theorem, a general version of stochastic Barb ă lat's lemma is obtained, which requires the concerned stochastic process to be almost surely integrable, rather than absolutely integrable in the sense of expectation, unlike in [2]. As the second contribution, supported by the general stochastic convergence theorem, an adaptive output-feedback control strategy is established for the global stabilization of a class of stochastic nonlinear systems with severe parametric uncertainties coupled to unmeasurable states. Its feasibility analysis takes substantial effort, and is largely based on the general stochastic convergence theorem. Particularly, for the resulting closed-loop system, certain stochastic boundedness and integrability are shown by the celebrated nonnegative semimartingale convergence theorem, and furthermore, the desired stochastic convergence is achieved via the general stochastic convergence theorem.