Research On Structural Realiability Index Confidence Considering Multi-Error Factors

With the development of the structure reliability, the accuracy ofreliability index no longer demands with real engineering. This paper willlook for error factors in reliability calculating procedure. Then theinfluence of these error factors on reliability index is analyzed to makethe reliability index more significant.The data for calculating the reliability index are from measuringresults. By probability theory, a measuring result is a test, and repeatedmeasuring results are specimen. Generally, the sample result is regardedas parameter estimation of the population. Then, different sample resultwill get different parameter of the same population. Therefore, thereliability index is a random variable, not a fixed value in classicalstructure reliability. As a random variable, the reliability index must beconsidered confidence interval when using in engineering. The influenceof error factors on the reliability index will be researched in this paper.Firstly, the error in measuring process is inescapable. These errorscontain systematic error, random error and gross error. They can becontrolled by increasing the sample size and logical data processing. But,civil engineering standards and criterions demands are not strict and notagree with real engineering. Then, the calculated results of reliabilitywould be a castle in the air when standards and criterions do not work.Therefore, it is necessary to pay attention to data processing in structurereliability to prevent data processing error.Secondly, the normal distribution is applied extensively in structurereliability. It has mathematics advantage and agrees with real distributionof engineering data in a large part. The difference between the normaldistribution and engineering distribution is at two tails of their probabilitydensity curves. It is named end effect and can produce nonlinearcumulative error in structural system reliability. A new distribution is putforward in this paper. It inherits the advantage of the normal distributionand avoids the end effect.Thirdly, the whole process of a structure, which containing design, manufacturing and managing, is completed by human. Then, human erroris inevitable. An AHP-THERP model in created in this paper to analyzethe influence of human error on reliability index.Finally, a structure is made up of many construction members. Themanufacture size of a member is not equal to its design size. Thisdifference is structural geometrical error. It can bring up internal forceand influences structure reliability.In summary, a new calculating model of reliability index consideringall error factors is put forward in this paper. It is significant to structurereliability research.