Data-Driven Iterative Feedforward Tuning With Nonlinearity Compensation Based on Bang–Bang Funnel Control for Linear Motor Systems

Iterative feedforward tuning (IFFT) effectively improved tracking accuracy in linear motor systems. Nonlinear disturbances, such as friction, not only introduces repetitive errors but also destabilizes the estimation of feedforward parameters after convergence in existing IFFT methods. Although improving IFFT robustness against nonlinear disturbances enhances system performance, it inevitably reduces efficiency. To address the tradeoff between efficiency and robustness, a novel Data-driven iterative feedforward tuning method combined with nonlinearity compensation based on bang–bang funnel control (IFFT-BBF) is proposed. The IFFT-BBF utilizes data sampled over iterations to compute the projection state of tracking errors using the projection theorem. The approach enables the physical quantification of the compensation degree for feedforward parameters. When tracking error projections deviates significantly from the predefined state funnel, the bang–bang iteration loop is activated to quickly drive the error projections back toward the funnel. Subsequently, the robust iteration loop is triggered to ensure convergence of tracking error projections to the zero-state, corresponding to optimal feedforward parameters. The proposed IFFT-BBF is validated in a linear motor system with 1 μm resolution, reducing the nonzero root mean square tracking error to 1.85 μm. Comparative experiments demonstrate that the standard deviation of IFFT-BBF is no greater than 29.6% of similar methods, with minimal compromise to efficiency. This flexible approach significantly enhances both the stability and efficiency of IFFT.