Modeling and Optimization of a Large-Load Magnetic Levitation Gravity Compensator

This article presents the modeling and optimization design of a large load magnetic levitation gravity compensator. The proposed magnetic levitation gravity compensator comprises three layers of 2-D permanent magnet (PM) arrays. The middle mover layer can generate a large passive levitation force to compensate for the gravity of large loads in some magnetic levitation systems, such as the measurement framework in lithography machines or large space optical equipment to be tested on the ground. To accurately predict the passive levitation force, an improved magnetic charge model is proposed, in which the actual working points of each PM are considered. The accuracy of the improved model is verified via 3-D finite-element simulation. Genetic algorithm is then adopted as the parameter optimization method to reduce the levitation force stiffness within the effective vertical displacement to the greatest extent possible. Compared with the levitation force performance without parameter optimization, the sensitivity of the levitation force with vertical displacement is significantly reduced. In addition, the mechanical structure is designed, and the mechanical strength is checked. Finally, a magnetic levitation gravity compensator prototype with a passive levitation force of 6200 N is manufactured and tested. The tested values of the levitation force match well with the analytical and simulation results. The tested levitation force stiffness within 4 mm is less than 50 N/mm and basically remains unchanged, which is a superior performance for magnetic levitation systems.

Automated summary

This article presents the modeling and optimization design of a large load magnetic levitation gravity compensator. The proposed compensator consists of three layers of 2-D permanent magnet arrays. It generates a large passive levitation force to compensate for the gravity of large loads in magnetic levitation systems. An improved magnetic charge model is proposed to accurately predict the passive levitation force and a genetic algorithm is adopted for parameter optimization. A prototype is manufactured and tested, showing excellent performance.