The prospective prediction of thermal error of NC machine tools based on neural network algorithm

The thermal deformation error of CNC machine tools accounts for 40% -70% , which is a key problem that needs to be solved urgently among many factors affecting the numerical control accuracy error. The present situation of thermal error compensation for NC machine tools is investigated, and the key problem of thermal error compensation is summarized and analyzed, this paper compares and analyzes the algorithm of RNN, points out the implementation scheme of using LSTM neural network algorithm to solve the hysteresis of thermal error compensation, and combs out the concrete steps of computing solution of LSTM neural network, the Prediction Model of LSTM neural network algorithm is constructed. The experiment shows that the prediction accuracy based on LSTM algorithm is higher than that based on traditional thermal error prediction, the residual mean square error between the predicted value and the actual value is 0.945, which verifies the validity of the research.