Research on Temperature Field Prediction Model of Electric Spindle Based on Improved BP Neural Network

Abstract

Abstract The motorized spindle is the core component of CNC machine tools. In order to ensure its processing performance and processing safety, the temperature field of motorized spindle is studied. The three-dimensional model of the motorized spindle is established, and the convective heat transfer coefficient of the internal heat load and the simulation boundary condition are calculated by combining the heat transfer theory. The simulation is carried out by the finite element analysis software, and the internal temperature distribution of the motorized spindle under thermal steady state is calculated. Based on the numerical simulation analysis method and the thermal balance test method, the data basis for the prediction model of the motorized spindle temperature field is provided. The traditional BP neural network algorithm and PSO-BP neural network algorithm are used to predict the temperature of the motorized spindle measuring point under specific working conditions, and the temperature field prediction results are compared and analyzed. The results show that the PSO-BP neural network prediction model has good compatibility for variable data input, and the prediction results show little difference, which has high prediction accuracy and robustness.