Abstract

This paper considers the phenomenon of 'pole slipping' in magnetic gears and couplings as a result of torque overload. Specifically, previously reported work on optimized servo speed control and pole-slip detection in magnetic gears is extended through the development of a new control scheme to prevent pole slipping due to combined controller and load torque overload. By utilizing a model predictive control (MPC) strategy, the controller's principal objective is to prevent the magnetic gear from pole slipping by invoking hard constraints on the amount of controller torque that can be applied for a given steady-state load torque. A custom demonstrator drivetrain incorporating a magnetic coupling (1:1 magnetic gear) is used to experimentally verify pole-slip prevention using an implementation of explicit MPC via multiparametric quadratic programming (mp-QP). It is shown that while conventional MPC is restricted to systems with relatively low sample rates, due to the need to solve the constrained optimization problem in real time, an alternative explicit form of MPC can be readily utilized at sample rates more typically found in electrical drive applications. The underlying principles and benefits afforded by the proposed techniques are validated using simulation and experimental measurements from the drivetrain test facility, using only motor-side sensor measurements and load-side state estimates via a discrete-time observer. © 1996-2012 IEEE.