The Application Of Bayesian Updating In Reliability Analysis

Structural reliability analysis, a branch of reliability engineering, plays an important role in guaranteeing the safety of products. Bayesian approach has been rapidly in theory due to the solution of the calculation of posterior distribution. As a result, it has been widely accepted by reliability researchers and used to solve a great many structural reliability problems, which were unsettled by classical statistics. In general, because of reliability technology to deal with the test object that you can do less, how to use the prior information to reduce the testing time is crucial which promoted the application of Bayesian method in the field of reliability.Based on a combination of Bayesian approach and fatigue crack propagation model, a new method is established for identifying the distribution of fatigue life of the structure when uncertainties in fatigue crack propagation are modeled as random variables. By utilizing the presented method, we can research fatigue life distribution under a given crack length and the remaining life distribution of fatigue crack propagation, which is important for produce the best possible structures potential.As to materials fatigue life distribution, a new computational method is proposed for identifying the probability distribution of fatigue life based on maximum entropy principle in this paper, which use of the flexibility and optimal unbiased estimation of the maximum entropy principle. The sample statistics are used to estimate the first four statistical moments of fatigue life. Then, based on all those statistical moment constraints, the maximum entropy principle is employed to identify a rational probability distribution for fatigue life.Fatigue crack propagation is a time-dependent process. Due to the stochastic nature of fatigue crack propagation, fatigue crack growth is not a smooth and stable process. Therefore additional information such as usage information from health monitoring system and crack size measures from inspections can be used to Bayesian update various quantities of interest. By performing continuous updating, uncertainties associated with system reliability and crack size prognosis can be reduced for decision-making. It is significant for make full use of structure potential.