高档数控机床早期故障累积机理与动特性退化趋势预测研究

The domestic CNC machine tools reliability is one of fundamental factors for machine tools industry. With the rapid development of high speed and accuracy, as well as extensive usage of new structures and materials for CNC machine tools ,the difficulties of ensuring machines safety and reliability have increased dramaticly.Dynamic performance degradation mechanism and the performance degradation process evaluating of CNC machine tools are necessary for the safe and reliable operation process. The prediction and recognition of CNC machine tools working condition exactly have become short plate in manufacturing process. Based on the practical problem encountered during the manufacturing process with high speed and variable load conditions , focusing on studying relative theories and methods about working reliability, the research mainly engages studies concerning regularity on propagation of early fault, and early fault features mining optimization extraction，dynamatic state of performance state identification assessment based on multi-information fusion and traceability in order to solve key problems mentioned above. Several innovative theories, including regularity on propagation of early fault of CNC machine tools , searching propagation path of early fault based on multi-manifold variation space, analying relative dynamic performance evaluation prediction as well as degration tracing technology, are proposed. The outcomes of the research will be applied to develop CNC machine tools working state evaluation and predication system partially used in analizing CNC machine tools working reliability by combining the fault mechanism and the running state information. It will provide directive theories, methods and tools to insure the CNC machine tools working safety in real-time.

国产数控机床的可靠性是行业发展急需解决的问题。尤其是随着新技术新材料产品的要求，数控机床功能越来越复杂。研究数控机床的动态性能退化机理，评估数控机床的性能衰退过程对数控机床的安全可靠运行非常必要。本项目针对状态识别存在的难点问题，对高速变负载工况下的数控机床运行状态的动态性能开展研究。在方法学上，拟探索出一条数控机床动态性能退化机理、早期故障特征挖掘优化提取、性能状态辨识评估与溯源的集成方法，构建基于多源征兆信息的运行失稳劣化状态识别模型，改变目前不能定性判定数控机床状态，合理维护的现状；在基础理论上，在数控机床早期故障累积规律、多流形空间域、故障趋势敏感特征优化获取策略，性能评估和预测、退化溯源的相关理论和技术进行系统研究，突破融故障机理、运行状态信息和系统固有特性的数控机床状态评价等理论难点，为实现数控机床运行状态实时性能评估，数控机床的运行可靠提供理论和方法

国产数控机床的可靠性是行业发展急需解决的问题。尤其是随着新技术新材料产品的要求，数控机床功能越来越复杂。研究数控机床的动态性能退化机理，评估数控机床的性能衰退过程对数控机床的安全可靠运行非常必要。本项目提出了五轴球头铣削加工动力学模型构建与稳定性预测方法，为铣削稳定性预测奠定了可靠基础。建立了考虑刀具-工件交互效应的五轴球头铣削动力学模型，揭示了多种交互效应下刀轴姿态对五轴球头铣削稳定性的影响规律，可更加准确地预测五轴球头铣削的稳定性。提出了基于状态信息的多域空间流形学习的早期故障敏感特征优化获取方法，揭示故障演变累积机理，建立敏感特征与典型故障的映射关系。构建基于多源信息融合数控机床运行状态评估模型。研究轴心轨迹的提纯方法和自动识别技术，建立基于轴心轨迹的数控机床动特性劣化识别方法，基于S试件加工精度与数控机床运行状态的映射模型，真实反映数控机床服役阶段的性能状态与运行可靠性演变规律，发现数控机床运行过程中的薄弱环节，进行溯源定位和控制调整，改变目前不能定性判定数控机床状态和合理维护的现状。构建了数控机床多源信息融合可靠性综合评价框架，实现融合机床历史故障数据、运行状态、精度劣化信息、零件加工精度等多源信息的可靠性评价。为实现数控机床运行状态实时性能评估，数控机床的运行可靠提供理论和方法。