【作者】 马明阳；

【导师】 王义强；

【作者基本信息】 吉林大学 ， 机械制造及其自动化， 2012， 硕士

【摘要】 数控加工是现代机械制造的关键技术。随着计算机和自动化技术的飞速发展，数控机床在工业生产中占据着越来越重要的地位，是衡量一个国家制造业发展水平的重要标志。数控系统作为数控机床的核心部位，对整个数控机床的性能起着至关重要的作用。国产数控系统与进口数控系统存在的主要差距就是在可靠性方面。因此，国家投入了大量的科研力量对数控系统的可靠性进行研究。本文结合国家科技重大专项“数控系统可靠性技术及重型机床运行可靠性研究（2009ZX04014-013）”，主要研究了数控系统可靠性预计和软件可靠性两方面内容。对数控系统的可靠性进行准确的预计能够为后续的使用和设计提供度量，作为今后改进设计的依据。本文通过现场试验采集某型号数控系统的故障数据，建立三层神经网络模型对该型号数控系统的可靠性进行了成功的预测。对于今后指导数控系统的研发、升级有着积极的作用。现代数控系统正逐步向高速高精高效化、多功能化、智能化的方向发展，软件作为数控系统的重要组成部分，其规模、计算量和功能都越来越强大，可靠性也就越来越难以保证。因此，提高软件部分的可靠性成为提高整个数控系统可靠性的关键问题。本文对某型号数控系统软件部分进行了FMECA分析，找出了该型号数控系统软件的薄弱环节、多发故障模式和主要故障原因，并有针对性的提出了有效的改进措施。从失效机理上来看，数控系统软件的故障可以描述为“软件错误→软件缺陷→软件故障”，即：程序设计人员在软件设计过程中由于设计等原因在软件内引入一些缺陷，当软件程序的执行路径经过某一个缺陷，并激活该缺陷时，就会导致软件发生故障。因此，当数控系统软件发生故障时，快速准确的定位故障、排除故障是减少损失的重要手段。本文提出了基于结点差异的相似路径的确定和基于模糊算法的语句怀疑率的计算方法，从代码层对数控系统软件的故障进行定位，并以NURBS曲线插补为例进行实验验证。实验结果表明，本文提出的故障定位方法是行之有效的，为今后数控系统软件故障定位技术的研究提供了新思路。更多还原

【Abstract】 The CNC processing is the key technology of modern machinery manufacturing. Alongwith the development of the technology of computer and automation, CNC machine plays amore and more important role in the modern industry, and becomes a major mark for thelevel of a country’s manufacturing. The CNC system as the central part of CNC machineplays an important part in the function of the whole CNC machine. The main difference ofCNC system between domestic products and the import products is the reliability. So thenation inputs a great deal of research on the reliability of CNC system. This paper mainlydiscussed the reliability prediction of the CNC system and the reliability of CNC software,combining with the Major National Science and Technology Projects—“Research onreliability technology of CNC system and reliability on the operation of heavy machine（2009ZX04014-013）”.Predicting the reliability of CNC system accurately will provide a metric for thefollowing use and design, and be the foundation for improving the design in the future. Thispaper collected the fault datum of a certain type CNC system according to the field tests.Rebuilt a neural network model containing three tiers to predict the reliability of the certaintype CNC system, the results indicated that the model was effective for the reliabilityprediction. This will do great well for directing the upgrade, research and development ofthe CNC system.Nowadays, the modern CNC system is moving to the direction of high-speed,high-precision, high-efficiency, multi-functional and intelligent. Software is one of theimportant part of the CNC system. The scale, calculation and function of the software arebecoming so powerful that it is more and more difficult for ensuring the reliability of thesoftware. So how to improve the reliability of the software is the key to improve thereliability of the whole CNC system. This paper gave the FMECA to a certain type softwareof CNC system. Found out the weak parts, main mode and the major cause of faults. Andproposed effective countermeasures to improve the reliability of the CNC software.From the perspective of failure mechanism, the fault of the CNC software can bedescribed “software-error→software-defect→software-fault”, that is to say, the designersbrought in some defects during the processing of designing, when the execution trace wentthrough a defect, and activated it, the software would be failure. So locating the faultsquickly and accurately when the CNC system becomes failure is the most efficient way toreduce the loss. This paper proposed a new method to locate the fault of CNC software.This method contained tow parts—generating the similar execution traces set based on thedifference of nodes and calculating the suspiciousness of the statements executed by these traces based on the fuzzy algorithm. This could locate the fault of the CNC software into thecode layer. Finally, gave an experiment on the NURBS interpolation for fault location. Theresult showed that this method was very effective in fault location and it provided a newidea for the fault location of CNC software.更多还原