摘要
依靠寿命试验对整机进行可靠性评定受到多种制约,单元为不同分布的复杂系统可靠性综合一直是可靠性工程中的难点问题。根据可靠度一阶矩与二阶矩符合相等原则,基于工程背景,研究了无信息先验下求解含并联冗余的串并联系统可靠度Bayes置信限的方法。在已知单元寿命分布及试验信息的基础上,基于Bayes方法,分别用Beta分布和LΓ分布拟合各单元可靠度的验后密度,计算各单元及并联冗余子系统可靠度一阶矩和二阶矩。利用并联系统可靠性模型,将并联冗余子系统等效为二项或指数单元,使系统简化成一般串联系统。再根据串联系统可靠度一阶矩和二阶矩折算出系统试验信息,相当于替代了整机可靠性试验,并据此求解系统可靠度的二项近似与指数近似2种Bayes置信限。该方法在实际工程中具有明显的应用前景,以某型自动平衡记录仪为例,验证了其有效性和实用性。
The reliability assessment based on life testing is limited in the whole machine, and it is a difficult to synthesize the reliability of the complex system which consists of units with different life distributions. Bayes reliability confidence limit for series-parallel system under non-information prior is investigated for engineering applications, according to the equal-principle of first and second moment of reliability. When the life distribution and testing information are obtained, the posterior density functions for different units are fitted with Beta distribution and LГdistribution, based on Bayes statistic inference. The first and second moment of each unit and parallel redundant subsystem reliability are computed. The parallel redundant subsystem is equivalent to binomial or exponential distribution unit according to parallel subsystem reliability model. The system is simplified to be general series system, which can be folded into the test information according to first and second moment of reliability. It is used to replace reliability test of the whole machine. Hence the binomial and exponential approximate confidence limits of the system reliability can be obtained. The method has potential application in the practical engineering. The experimental result for certain automatic balanced recorder shows it efficient.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2013年第2期428-433,共6页
Chinese Journal of Scientific Instrument
基金
国家863计划(2009AA04Z407)
国家自然科学基金(79970045)
国土资源部公益性行业科研专项(201011082-06-1)
黑龙江省教育厅2011年度面上项目(12511093)
哈尔滨理工大学青年科学研究基金(2011YF040)资助项目
关键词
复杂系统
可靠性综合
一阶矩
二阶矩
二项置信限
指数置信限
complex system
reliability synthesis
first moment
second moment
binomial approximate confidence limit
exponential approximate confidence limit
