Constrained Near-Time-Optimal Sliding-Mode Control of Boost Converters Based on Switched Affine Model Analysis

Time-optimal (TO) control of dc-dc converters, using switching surface controllers, has been investigated extensively in the literature. Studies show that such controllers are of a minimum-switching and bang-bang nature and have the fastest possible response time. However, this method has three drawbacks: First, the inductors' maximum current goes beyond practical limits. Second, due to a marginal status in existence condition of the switching surface, stability of the controller is sensitive to parameter changes. Third, its nonlinear switching surface needs high computational power and is hard to implement. To overcome these problems, the switched affine model of boost converters is used in this paper to study the time response and equilibrium states. Then, the current-constrained TO state trajectory is studied, and its optimality is discussed. A constrained near-optimal (CNO) controller is proposed with a piecewise linear switching surface to guarantee the robustness of stability and the simplicity of implementation. A general condition for the existence of sliding mode and a novel method to check finite time reaching are proposed for arbitrary switching surfaces. The Lyapunov stability for the CNO controller is also discussed. The proposed method is validated on an experimental setup with a boost converter prototype and a TMS320F2812 processor board.