Data-Driven Feedforward Parameter Tuning Optimization Method Under Actuator Constraints
Liangliang Yang, Hui Zhang
Abstract
Feedforward control includes iterative force injection-based feedforward control and model-based feedforward control. Model-based feedforward control enables high performance for nonrepetitive motions in industrial motion systems. In particular, model-based feedforward control with a parameterized input shaping filter and a feedforward controller can effectively implement trajectory tracking tasks without relying on mathematical models of systems. However, the pre-existing model-based feedforward control does not consider the saturation boundary of the system actuator, which may lead to saturation of the feedforward force for actual systems with actuator constraints. Therefore, this article aims to develop a feedforward control method that obtains high tracking performance for nonrepetitive trajectory under actuator constraints. In particular, a parameterized input shaping and feedforward control method based on basis functions is pursued, which is combined with a data-driven iterative optimization approach to cope with nonrepetitive trajectory tracking tasks and actuator constraints. Simulations and experiments on a motion system confirm the optimal trajectory tracking performance under actuator constraints and robustness to nonrepetitive trajectory.