Compensation adaptive robust control for a linear motor–driven stage system with state and input constraints based on gated recurrent unit architecture

Abstract

Motion control of mechatronic systems with uncertainties and physical constraints, while ensuring robustness and achieving better performance, such as high tracking accuracy and fast response, has always been a hot topic. However, the most current related works only focus on how to guarantee system stability under constraints, and few consider comprehensive performance. This paper investigates gated recurrent unit (GRU)-based compensation adaptive robust control (ARC) for uncertain linear motor–driven stage system with state and input constraints. To achieve rapid and precise motion control, a dual-loop control structure is employed, where GRU and ARC are the outer loop and the inner loop, respectively. First, the ARC control law is used to deal with the parameters uncertainty and external disturbances in the system, which further improves the tracking accuracy. A GRU neural network is then constructed and capable of implementing precise prediction ahead of the actual system output. Through choosing suitable loss function and training model, it can effectively minimize prediction error under state and input constraints. Comparative experiment results demonstrate the superiority and validity of the proposed scheme on the basis of GRU and ARC.