基于数据驱动的机床热误差补偿技术研究

Research on Thermal Error Compensation Technology of Machine Tools Based on Data Driving

随着先进制造技术的发展,对数控系统的高速、高精、高效、智能化方面的要求越来越高,为此,实现热误差补偿在内的补偿技术具有重要的现实意义.热误差的非线性问题是实现误差补偿的难点之一,传统的补偿方法采用离线线性拟合的方法,造成较大的拟合误差且实时性很难保证,为此传统的基于精确模型的热误差补偿技术遇到了发展瓶颈,提出了采用数据驱动的方法实现机床热误差补偿,采用模糊神经网络作为学习模型,结合加工过程实时采集到的误差数据,提供最优的补偿策略,减小了拟合误差,提高热误差补偿的补偿精度,并通过试验进行测试,测试结果表明所提出的补偿方案相对于传统的方案,显著改善了拟合精度,从而验证了该方案的可行性.

With the development of advanced manufacturing technology,the requirement of high speed,high precision,high efficiency and intellectualization of NC system is getting higher and higher.Therefore,the compensation technology including thermal error compensation has important practical significance.The non-linearity of thermal error is one of the difficulties in error compensation.The traditional method uses off-line linear fitting method,which results in large fitting error and hard to guarantee real-time performance.Traditional thermal error compensation technology based on precise model meets the bottleneck of development.A data-driven method is proposed to realize thermal error compensation of machine tools.Fuzzy neural network is used as learning model.Combining with the error data collected in real-time processing,the optimal compensation strategy is provided,and a data-driven method is developed.The fitting error is reduced and the compensation accuracy of thermal error compensation is improved.The test results show that the proposed compensation scheme significantly improves the fitting accuracy compared with the traditional scheme,thus verifying the feasibility of the scheme.