Combined Unfixed-Structure and Fixed-Structure Data-Driven Feedforward Control Approach for Ball Screw Feed-Drive System
Tao Huang, Yueting Kang, Yangjun Pi, Min Li
Abstract
Ball screw feed-drive system (BSFDS) is a widely used high-precision positioning mechanism, and its positioning accuracy is strongly limited by friction, vibration, etc. Those factors are difficult to model or to model completely. Since the data-driven control approach is not model dependent, it is suitable to overcome the previous problems. Iterative learning control (ILC) is one of the general unfixed-structure data-driven approaches, but it is restricted by its residual error. Therefore, this article proposes a combined unfixed-structure and fixed-structure data-driven feedforward control method. Notably, the unfixed-structure part adopts a standard norm-optimal ILC, and the fixed-structure part is a finite impulse response (FIR) filter structure. Then, the corresponding optimization algorithms are proposed to design an unfixed-structure ILC feedforward controller and to optimize the parameters of the fixed-structure FIR feedforward controller. Furthermore, the convergence and trajectory tracking performances of the proposed method are analyzed. The proposed data-driven feedforward control strategy is experimentally validated and compared with a preexisting FIR approach and standard norm-optimal ILC. The results show that the proposed approach has better convergence performance and trajectory tracking performance, and represents an effective strategy to improve the positioning accuracy in BSFDS.