Evaluation method for thermal error prediction models of computer numerical control machine tools

Due to the complicated mechanism of Computer Numerical Control (CNC) machine tool's thermal error, they are usually predicted by mathematical modeling based on temperature field monitoring data. The complexity of working conditions and environment makes it difficult to evaluate the advantages and disadvantages of data-driven modeling method comprehensively. Therefore, a set of evaluation index system of thermal error prediction model was established and the corresponding evaluation method was proposed. The performance of the prediction model was evaluated comprehensively from the aspects of robustness, accuracy, effectiveness, stability and correlation. By taking the temperature field data of ZK5540 heavy CNC machine tool with air-cutting conditions as an example, the performance of three different data-driven models were calculated and analyzed by using the evaluation index and evaluation method proposed, which verified the effectiveness of the evaluation index and method. © 2019, Editorial Department of CIMS. All right reserved.