Thermal error prediction of machine tool spindle of GWOA-BiLSTM based on thermal image

Thermal error is an important factor affecting the machining accuracy of high precision CNC machine tools. In order to improve the machining accuracy and performance of the machine tool and reduce the thermal error generated during the operation of the machine tool, a grey wolf optimization algorithm (GWOA) and bidirectional long short-term memory (BiLSTM) neural network based on thermal images are proposed in our study. Firstly, the thermal imager is used to obtain the temperature field information of the spindle area of the machine tool. Secondly, the temperature sensitive points were screened by DBSCAN algorithm and correlation coefficient method. Then, by simulating the predation behavior of gray wolves, the optimal parameters of BiLSTM were found by searching in the parameter space. Finally, the thermal error is predicted by using the temperature-sensitive points and thermal displacement data obtained, and verified on the test machine. The experimental results show that compared with the root mean square error (RMSE) and mean absolute error (MAE) of the BiLSTM neural network prediction model optimized by GWOA, MAE decreased by 0.518 0 and 0.382 3 μm, respectively, and the coefficient of determination R² increased by 0.057 8. Compared with BiLSTM neural network model, GWOA optimized model has better prediction performance.