Machine tool spindle thermal error modeling based on genetic algorithm optimization grey neural network
                                                               ZHENG Jinyong1, LIU Baoguo1,2, FENG Wei1,2
(1.School of Mechanical and Electrical Engineering, Henan University of Technology,Zhengzhou 450001, China; 2.Henan Key Laboratory for Superabrasive Grinding Equipment, Henan University of Technology, Zhengzhou 450001, China）

Abstract: Aiming at the problem that the machining accuracy was affected by the thermal performance of the machine tool spindle, the experimental research on the thermal error modeling direction of the machine tool spindle was carried out. Taking the CNC grinding machine spindle as the research object, with the temperature change data and thermal error data obtained by thermal characteristic test, the method of fuzzy clustering on the temperature data and the correlation coefficient method were used to select temperaturesensitive measuring points. A grey neural network thermal error prediction model based on genetic algorithm optimization was established, by optimizing the initial parameters of the gray neural network. In this model, the absolute error of the predicted output and the actual value of the gray neural network were used as the fitness function of the genetic algorithm, with the average relative error as the evaluation criteria for the prediction model, compared with gray neural network and BP neural network prediction results. The results indicate that the prediction model has higher prediction accuracy. Genetic algorithm can improve the prediction accuracy of gray neural network and be good for thermal error compensation systems by optimizing the initial parameters.

Key words: NC machine tool; thermal error; gray neural network; genetic algorithms; fuzzy cluster grouping