数控机床热误差预测模型的评估方法

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

数控机床热误差产生机理复杂,通常使用温度场监测数据数学建模对热误差进行预测,加工工况和所处环境的复杂使得数据驱动建模方法的优劣难以全面评价,因此建立了一套热误差预测模型的评价指标体系,并提出了相应的评估方法,从热误差预测模型的鲁棒性、准确性、有效性、稳定性与相关性方面对预测模型的性能进行了综合评价。以ZK5540A型重型数控机床在空转工况下所监测的温度场数据为例,采用所提评价指标和评估方法分别对3种不同的数据驱动建模方法的性能指标进行了计算与分析,验证了评估指标和方法的有效性。

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.