Thermal error modeling and compensation of spindle based on gate recurrent unit network

In view of the serious hysteresis and nonlinear relationship between the thermal error of CNC machine tool spindle and the temperature rise of spindle measuring points, a spindle thermal error prediction model combining sparrow search algorithm (SSA) and gate recurrent unit (GRU) is proposed. Taking the spindle of a precision machine tool as the research object, the thermal error and the temperature field of the spindle in idling state are measured. Select the temperature of the measuring point of the spindle as the input and the thermal error in Z-direction as the output, the thermal error prediction model is established by using GRU network. SSA is used to optimize the training parameters of GRU network, and finally a prediction model of SSA-GRU spindle Z-direction thermal error considering the influence of natural environment is established. The performance of the established model is verified by taking the test data of variable speed working condition as the robustness test set. The results show that SSA-GRU can be used for thermal error modeling and compensation, and the Z-direction thermal error of the machine tool spindle can be controlled within 8 μm, which has better prediction accuracy than the traditional network model.