Modeling and Parameter Identification of Micropositioning System Based on the Expedited Aitken Method

The hysteresis of the positioning platform exhibits rate dependence. To address this issue, the combined modeling approach for hysteresis, integrating the improved gradient descent (GD) method, is proposed. This method incorporates a hysteresis operator at the input of a hybrid radial basis function-autoregressive (RBF-ARX) model. Subsequently, parameter identification is conducted using the gradient descent algorithm based on decomposition identification technology (DGD). Furthermore, the expedited Aitken method is introduced into the DGD to mitigate the impact of the learning rate on the algorithm’s convergence. Finally, the performance of the proposed method is validated through simulation and application on a micropositioning platform, indicating significant improvements in both convergence speed and parameter estimation accuracy.