This paper discusses the H∞ control problem for a class of nonlinear stochastic systems with both state-and disturbance-dependent noise. By means of Hamilton-Jacobi equations, both infinite and finite horizon nonlinear stochastic H∞ control designs are developed.Some results on nonlinear H∞ control of deterministic systems are generalized to a stochastic setting. We introduce some useful concepts such as "zero-state observability" and "zero-state detectability" which, together with the stochastic LaSalle invariance principle, yield some valuable consequences in infinite horizon nonlinear stochastic H∞ control.
Introduction.In practice, when the exogenous disturbance enters the system, an H ∞ control design is often first considered, when a control law is sought to efficiently eliminate the effect of the disturbance; see [12], [13] and the references therein. Theoretically, study of H ∞ control first starts from the deterministic linear systems, and the derivation of the state-space formulation of the standard H ∞ control leads to a breakthrough; details can be found in the prize-winning paper [17]. From the viewpoint of the state space, the linear H ∞ control problem can be converted into the study of a game-theoretic Riccati equation, and the "completion of square methodology," similar to the linear quadratic (LQ) and linear quadratic Gaussian (LQG) theories, can be applied; see [15], [16], [20], [21], [25], and [30]. Soon after the appearance of [17], the nonlinear H ∞ control problem (on deterministic systems) was investigated by many authors; see [5], [14], [18], and [19]. From the time-domain perspective, an H ∞ norm of the transfer function is nothing else but the L 2 -induced norm of the input-output operator with initial state zero. This important feature makes it possible to develop nonlinear or stochastic H ∞ theory. Van der Schaft [5] made a contribution to the state feedback H ∞ control for nonlinear deterministic systems with infinite time horizon, where a relatively deep tool, i.e., the strict relation between Hamilton-Jacobi equations (HJEs) and invariant manifolds of Hamiltonian vector fields, was applied. Using this tool, he showed that a local solution to the primal nonlinear H ∞ control exists if its linearized H ∞ control problem is solvable. The authors of [18] and [19] dealt with the output feedback H ∞ control of nonlinear deterministic systems with incomplete state information, and a separation principle was obtained. In [5], [18], and [19], the differential geometric approaches