Thermal error analysis and modeling for high-speed motorized spindles based on LSTM-CNN

The high-speed motorized spindle is the core part of high-precision machine tools, and its performance governs the machining precision of machine tools. The thermal deformation error is one of the main factors affecting the machining accuracy of the high-speed motorized spindle. The thermal error compensation system is an effective way to reduce thermal error. However, the performance of the compensation system primarily depends on the accuracy of the thermal error prediction model. Hence, it is crucial to establish an accurate and robust thermal error model. The conventional thermal error prediction models have low accuracy. Therefore, in this model, an effective thermal error prediction model is proposed based on the combination of long short-term memory (LSTM) and convolutional neural network (CNN). Taking high-speed motorized spindle as the research object, the principle of thermal error generation is analyzed for examining the influence of thermal error on the machining accuracy of the spindle. According to the principle of spindle deformation, an experimental platform of the motorized spindle is built to investigate the temperature and thermal deformation. The thermal error experiments are conducted under three different working conditions, and the temperature of the measurement point and the displacement offset of the spindle front end are determined to analyze the variation in the temperature and displacement over time. The K-harmonic means (KHM) clustering algorithm and grey relational analysis method (GRA) are used to optimize the temperature measurement points. Then, the LSTM-CNN thermal error prediction model is established and compared with the conventional model in terms of prediction performance and robustness. The results reveal that the proposed thermal error model is significantly better than the conventional model. Hence, the presented model can provide theoretical basis and technical support for thermal error compensation and optimized design of high-speed motorized spindle, and it is expected to be effectively applied and promoted in actual machining.