Feedforward-adaptive neural network control for ultra-precision positioning in magnetic levitation motion stages

The objective of this study is to develop a feedforward-adaptive neural network controller (F-ANNC) for precise trajectory tracking and robust disturbance rejection in magnetic levitation motion stage (MLMS) systems. Initially, the dynamic behavior of the MLMS is modeled to capture the fundamental characteristics of the system. The F-ANNC is then constructed by integrating model-based high-order trajectory feedforward control with an adaptive neural network (ANN) compensator and a linear feedback controller. The neural network component is designed using a multi-layer radial basis function neural network (ML-RBF-NN), enabling adaptive estimation and compensation for system uncertainties and external disturbances. This approach allows the neural network to dynamically adjust based on changes in system behavior without requiring precise system parameter settings. The adaptive control laws are derived from Lyapunov stability theory, ensuring both system stability and robustness. Extensive simulations and experiments validate the proposed F-ANNC, demonstrating its superior performance in managing parameter uncertainties and external disturbances, thereby confirming its effectiveness in ultra-precision MLMS applications.