A Multi-attribute Information Based Method of Material Strength Distribution Fitting

摘要

Information fusion technique has been widely applied to a variety of subjects such as fault diagnosis and image identification.Bayes estimation is a special type of information fusion technique applied to parameter estimation for probability distribution of random variable.The present paper presents a new type of information fusion technique for material strength distribution estimation in the situation of small size sample.To precisely describe material strength,three-parameter Weibull distribution is used.To find out a reasonable location parameter in the situation that only a few experimental observations are available,the knowledge and information from different aspects are utilized.First,an empirical shape parameter is chosen with reference to the strength distribution of similar material.Then,a location parameter is assigned to make the estimated material strength variation at a realistic level,by judging the rationality of the location parameter through the strength probability distribution thus estimated.At last,big data technique is applied to further verify the rationality of the estimated material strength distribution by testing the relation between location parameter and the minimum observation in a sample of particular size for a special three-parameter Weibull distribution.