Magnetic servo levitation by sliding-mode control of nonaffine systems with algebraic input invertibility

Magnetic Servo Levitation (MSL) is an important actuation principle with potential applications ranging from ultra-high-precision positioning to high-speed rail systems. This paper describes a nonlinear controller design technique for MSL that has inherent robustness to both parametric uncertainties and unmodeled dynamics. Most of the currently available literature on sliding mode considers nonlinear systems that are linear (affine) in the input action. The proposed technique allows designing sliding-mode controllers for the family of nonaffine problems that have an input nonlinearity algebraically invertible with respect to the available control action. This differs from the standard approach of input feedback linearization, and is based on a modified sliding condition that can be used to synthesize a switching control law. An equivalent control term can also be included, substantially enhancing the performance of the controller. Experimental results show that the proposed technique can achieve excellent tracking at high speeds in a fast-tool servo system actuated by MSL.