基于通用生成函数的系统寿命可靠性分析  

Reliability Analysis for System Life Based on Universal Generating Function

作　　者：任博[1] 吕震宙[1] 李贵杰[1] 唐樟春[1] 

机构地区：[1]西北工业大学航空学院,陕西西安710072

出　　处：《航空学报》2013年第11期2550-2556,共7页Acta Aeronautica et Astronautica Sinica

基　　金：国家自然科学基金(51175425);航空科学基金(2011ZA53015)

摘　　要：提出一种基于通用生成μ函数的液压系统可靠性分析新方法,并对某液压系统可靠性进行分析,为改进系统提供依据。首先从系统失效数据出发,将部件工作状态分类,根据失效数据得到部件处于各状态的概率,建立部件的μ函数模型,描述部件性能的不确定性。然后依据系统的结构功能逻辑关系,借助μ函数特性和逻辑算子,得到系统μ函数。最后基于所提μ函数的系统可靠性指标,完成系统可靠性分析。与传统Monte Carlo(MC)方法相比,所提方法直接从失效数据出发,确定部件状态并构造反应其不确定性性能分布的μ函数,通过μ函数将部件不确定性传递到系统,并基于系统μ函数直接进行可靠性分析,避免了传统MC方法拟合部件级不确定性概率分布的误差。以某液压系统为例验证所提方法的正确性和高效性。A novel technique for the reliability analysis of a hydraulic system is presented based on the universal generating μ function and useful information can be obtained for system analysts and design engineers to guide effective system improve- ment both in the design phase and during operation. First, according to the fault data, the states and their corresponding probabilities for the constitutive elements of the system are determined by the histogram technique. Then, based on the de- termined state probabilities for all elements, an individual p function for each element is defined by means of p function. Fi- nally, by using the composition operators over the p function of individual elements and their logic combinations in the entire system structure, the μ function of the entire system can be obtained by simple algebraic operations, and this μ function ex- presses the output performance distribution for the system. Due to the definition of reliability measures about the μ function, reliability analysis can be easily completed from this output performance distribution compared with traditional Monte Carlo （MC） method. As the method is performed without estimating the distribution of the elements, the errors of truncation and numerical fitting are reduced. The example of a certain hydraulic system is used to illustrate the correctness and effective- ness of the novel method.

关 键 词：μ函数 液压系统 Z变化 可靠性分析 MONTE CARLO方法 

分 类 号：V221[航空宇航科学与技术—飞行器设计] TB114.3[理学—概率论与数理统计]