基于相似机床信息的CXK5463的可靠性评估

发布时间：2018-04-03 18:37

本文选题：加工中心　切入点：可靠性评估　出处：《燕山大学》2015年硕士论文

【摘要】：可靠性评估技术是数控机床可靠性工程技术的重要组成部分,它可以验证机床可靠性是否达到预期目标,检验机床的设计是否合理,进而指出机床的薄弱环节,为改进机床性能提供科学依据。但由于高档数控机床具有数量少且故障数据匮乏的小样本特征,传统的评估方法无法对其可靠性进行有效地评估,这已成为当前高档数控机床可靠性工作的重点和难点。因此,本文以小样本的CXK5463车铣加工中心为研究对象,针对传统贝叶斯方法在利用相似机床信息时确定的先验分布不合理的问题,通过引入继承因子,提出了一种混和先验分布的贝叶斯可靠性评估方法。首先,通过对加工中心的故障数据进行再抽样得到了再生大样本,并用它与两型号相似机床的故障数据进行相容性检验,证实了两型号相似机床故障数据与加工中心现场故障数据均相容,得出加工中心服从双参数威布尔分布;根据相似机床的故障数据计算得到混合先验分布中历史先验分布参数的数学模型;根据Reference先验方法计算得到混合先验分布中更新先验分布参数的数学模型。其次,通过对两型号相似机床故障数据和加工中心故障数据进行再抽样处理,分别得到了它们的概率密度分布,进而利用相似机床与加工中心概率密度分布的差异情况计算得到了基于客观法的继承因子;通过分析处理两相似机床与加工中心在各指标的相似度专家评分,利用赋权法计算了各相似指标所占权重,进而根据权重和各相似指标值计算得到了基于主观法的继承因子;利用综合赋权法综合考虑客观法和主观法的继承因子值,得到了综合继承因子;根据信息融合原理将历史先验分布、更新先验分布和继承因子融合得到了混和先验分布。然后,根据贝叶斯公式将混和先验分布和加工中心现场试验样本进行了融合,得到了加工中心故障间隔时间的后验分布函数。最后,针对加工中心的后验分布为多重积分,难以计算的问题,以马尔科夫链蒙特卡罗法为理论基础,运用Open BUGS软件对加工中心的后验分布进行了模拟仿真,得到了加工中心MTBF的估计值。通过与传统贝叶斯方法和无信息贝叶斯方法对比,验证了本文方法的正确性。
[Abstract]:The reliability evaluation technology is an important part of the reliability engineering technology of NC machine tool, it can verify the reliability of the machine is expected to achieve its objectives, design of test machine is reasonable, and it is pointed out that the weak link of the machine, to provide scientific basis for improvement of the performance of machine tools. But because of the high-end CNC machine tool has the characteristics of small sample quantity is small and the lack of fault data, evaluation the traditional methods can not effectively assess its reliability, which has become the focus and difficulty of the current high-end CNC machine tool reliability. Therefore, this paper takes CXK5463 milling machining center small sample as the research object, according to the traditional Bias method determined a priori information in the use of the similar machine distribution is not reasonable, the inheritance factor introduction, it proposes a hybrid reliability evaluation method of Bias prior distribution. Firstly, based on the machining center Fault data re sampling has been regeneration samples, and use it with two types of fault data of similar machine compatibility test, confirmed the two types of similar machine fault data and processing center field failure data are compatible, the machining center obeys two parameter Weibull distribution; according to the fault data similarity calculated mathematical model of hybrid machine tool the distribution parameters of prior distributions in the prior; according to the Reference method to calculate a priori update mixed prior distribution parameters of the prior distribution mathematical model. Secondly, re sampling processing by similar machine fault data and fault data processing center on the two models, are obtained. The probability density distribution of them, and then use the similar machine tools and machining center the probability density distribution of the difference was calculated based on the method of objective inheritance factor; through the analysis of two phase like Machine tools and machining center in each index of similarity expert score, using weighting method to calculate the similarity index weights, and then according to the weight and the similarity index value is calculated based on the method of subjective factor inheritance; the comprehensive weighting method considering the inheritance factor objective method and subjective value of law, the comprehensive inheritance factor; according to the information fusion principle of history prior distribution, update the prior distribution and inheritance factor fusion has been mixed prior distribution. Then, according to the Bayesian formula mixing field test sample prior distribution and processing center for the integration, has been processing center fault interval posterior distribution function. Finally, the posterior distribution for machining Center for multiple integrals, difficult computational problem, using Markov chain Monte Carlo method as the theoretical basis, using Open BUGS software to the processing center of the posterior distribution. The MTBF value of the machining center is obtained by simulation. The correctness of the method is verified by comparing with the traditional Bias method and the information free Bias method.

【学位授予单位】：燕山大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TG659

【参考文献】