【作者】 吴军；

【导师】 邵新宇； 邓超；

【作者基本信息】 华中科技大学 ， 工业工程， 2008， 博士

【摘要】 国产数控装备的可靠性是制约数控装备行业发展的重大共性和关键性工程问题之一。作为一种对数控装备可靠性进行定量控制的必要手段,数控装备服役可靠性评估的主要任务是衡量数控装备是否达到预期的设计目标及使用要求,指出数控装备运行中的薄弱环节,为改进数控装备的设计、制造、工艺与维护等指明方向,从而实现数控装备的可靠性增长,保证数控装备长期安全可靠运行。本文根据数控装备的固有特征,分别从理论方法、工程应用、软件系统三个层面对数控装备服役可靠性评估进行了较为深入系统地研究与探索。首先,综述了目前数控装备可靠性研究及其相关领域的理论方法和研究现状,指出传统的基于二元状态(正常和故障)的可靠性评估研究的诸多缺陷。在此基础上,定义数控装备服役可靠性评估的内涵,给出数控装备服役可靠性指标,提出数控装备服役可靠性评估的核心内容与方法。构建数控装备服役可靠性评估的三层结构,并论述数控装备的性能状态监测与加工过程动力学建模方法。围绕着数控装备服役可靠性评估的三层结构,开展基于性能参数的数控装备服役可靠性评估研究,具体包括:(1)分析了数控装备使用过程中的性能劣化现象,考虑性能状态变化特征,提出基于性能状态监测的数控装备性能劣化建模与分析的原理。建立基于非平稳ARIMA时间序列法的数控装备性能劣化模型,探索数控装备的性能劣化时变性规律。引入离散隐Markov链分析多观测序列下数控装备可靠性变动状况,较好地解决了数控装备使用过程中所观测到的性能特征参数不能与性能状态——对应的问题。在此过程中,采用Lloyd算法实现性能特征参数的矢量量化,建立性能劣化的状态变迁模型及其可靠性变动计算模型,该方法能够准确地感知数控装备隐状态变迁过程,并能推断数控装备完成规定功能的概率以及处于不同状态下的概率。(2)分析了传统方法应用于数控装备服役可靠性评估时的困境,考虑数控装备结构特性和加工工艺特征等因素,采用动力学建模手段,分析数控装备处于某一性能状态时不同因素对数控装备输出性能的影响,提出小样本条件下数控装备服役可靠性评估的新思路。引入统计学习理论(SLT),建立数控装备服役可靠性评估模型,并给出相应的模型选择算法和模型求解算法,得到数控装备在某一性能状态下的概率型可靠性指标。在此基础上,将Bootstrap与SLT相结合进行可靠性置信区间估算,并对评估结果进行可靠性敏感度和重要度分析,以发现影响服役可靠性的最敏感因素以及不同单元对系统服役可靠性影响的重要度顺序等。该方法较好地解决了小样本条件下数控装备服役可靠性评估中非线性、评估准确度低等问题,提高了数控装备服役可靠性评估模型的稳健性和准确性。(3)以数控DM4600型立式铣床为研究对象,通过结构综合动刚度实验和铣削力实验,得到相应的动力学模型参数和铣削力模型参数,建立该装备的铣削加工过程非线性动力学模型,并进行实验校验。在此基础上,采用仿真手段分析数控装备机械结构与工艺系统的动态特性对数控装备铣削加工输出性能的影响,进而对数控装备在一个加工周期内的服役可靠性进行了评估。(4)自主设计与开发了数控装备可靠性建模与评估(MRME)原型软件系统,介绍系统的功能模块及特点,并以具体的工程应用实例为例演示MRME系统的运行过程。最后,对全文进行总结,指出进一步的研究方向。更多还原

【Abstract】 The reliability of Made-in-China CNC equipment is one of the key problems to restrict the rapid development of CNC equipment industry. The operational reliability estimation for a CNC equipment is an indispensable approach to control the reliability level of the equipment, and its tasks are to ensure that the CNC equipment will achieve desired objectives and requirements, and to identify those underlying faults, and then to indicate the required improvements in the CNC equipment’s design, manufacture, operation, maintenance and etc., in order to implement reliability growth and improve reliability level. According to current deficiency research on operational reliability estimation for CNC equipment, methods, applications and a developed prototype system for the operational reliability estimation are discussed in detail based on the inherent characteristics of CNC equipment.Firstly, this thesis provides a systematic literature review on the theories and methods for CNC equipment reliability and its related fields, and points out the limitations of traditional reliability estimation based on binary states: working or failed. Therefore, the concept of operational reliability estimation is brought forward. The reliability index and main research content of operational reliability estimation for the CNC equipment are built. And two failure modes such as catastrophic failure and drift failure are analyzed. And then, the three-tier system structure of operational reliability estimation is set up. Meanwhile, the methods of performance monitoring and machining processes’ dynamic modeling in CNC equipment are presented.According to the three-tier system structure of operational reliability estimation, the operational reliability estimation for CNC equipment reliability is implemented as follows:(1) The symptoms of performance degradation related to CNC equipment are illuminated, and the methods of performance degradation modeling and analysis based on performance monitoring are proposed. The performance degradation models based on nonsteady ARIMA time series are set up, which might correctly describe performance degradation process. Furthermore, discrete hidden Markov chain model is introduced to analyze the CNC equipment reliability under conditions of several observation sequences. The result shows that the discrete hidden Markov chain model might recognize the hidden states change of degradation process.(2) After indicating the difficulties in applying traditional reliability estimation methods to the CNC equipment, it is brought forward that the approach of the operational reliability estimation for the CNC equipment with small sample are based on statistical learning theory. The reliability estimation models are built using v-SVR and AIS arithmetic. And the confidence interval of calculated reliability index is deduced by integrating the Bootstrap with v-SVR arithmetic. Furthermore, the sensitivity and importance level of calculated reliability index are counted.(3) The foregoing methods are adopted to estimate the operational reliability of a machining process in DM4600 vertical milling machining center. The nonlinear dynamics model of the machining process is built based on the dynamic characteristic testing for the DM4600 machining center. Meanwhile, the model is verified by milling testing. The result shows that the nonlinear dynamics model is feasible. Furthermore, it is discussed how the dynamic characteristics of mechanical structure and machining system impact the operational reliability for the machining center. And the effect of machining parameter varieties on machining performance is studied by means of emulation. And then, the operational reliability for this equipment within a machining cycle is estimated.(4) The prototype system of machine reliability modeling and estimation called MRME is designed and developed. Its function modules and adoptive technologies are presented. A case is given to demonstrate the operation process of MRME system.Finally, a conclusion is drawn and the trend on operational reliability estimation is anticipated.更多还原