Data-driven methods are extensively used in machine tool thermal error modeling, yet the black-box and open-loop operation mode devoid of mechanism support makes it difficult to ensure the robustness and prediction performance, particularly under new operating conditions. This paper proposes an applicability evaluation method for machine tool spindle to establish a semi-closed-loop feedback between thermal error modeling and prediction processes. A temperature range index (TRI) is introduced through inductive analysis to quantitatively assess the applicability of the thermal error model. The temperature sensitive points (TSPs) are then re-screened based on the TRI, leading to the reconstruction of an improved thermal error model. The results demonstrate a strong correlation between the TRI index and the prediction performance of the thermal error model. The proposed applicability evaluation index and model reconstruction method address the limitations of existing data-driven thermal error models, further enhancing their predictive accuracy.
Xu et al. (Thu,) studied this question.