Adaptive feedforward construction method for precision motion stage based on iterative learning control

Feedforward compensation strategy is becoming increasingly important on improving control performance of the modern ultra-precision motion stages since the motion stages behave more flexible dynamic characteristics increasingly. Among numerous feedforward compensation strategies, iterative learning control with basis functions (ILC-BF) attracts more interests since the ILC-BF strategy enables excellent improvement of the control performance for varying tasks. However, most existing ILC-BF methods require artificially selecting the basis functions to construct feedforward controller in advance which results in inconvenience and limited performance improvement due to a possible inaccurate feedforward controller structure. Moreover, the data-driven parameter tuning strategy of most ILC-BF methods usually suffers from the biased-estimation issue caused by the presence of measurement noise, which leads to suboptimal parameter estimation results. In view of the above issues, an adaptive feedforward construction strategy is established to enhance the learning ability of the ILC-BF algorithm. Considering the estimation accuracy, instrumental variables (IVs) are involved to achieve the unbiased estimation in adaptive process of feedforward construction. Finally, the effectiveness of feedforward construction and the trajectory applicable for varying tasks are illustrated through the simulation as well as the experiments on a precision motion stage.