Discriminant Analysis And Prediction Of Concrete Compressive Strength Based On Order Statistic And Support Vector Machines

In the paper，we have discussed the gradually approaching normality of orderstatistic and inquired into a problem in GB/T50107－2010Standard for evaluation ofconcrete compressive strength by the precise distribution function of order statistic inmathematical statistics，the problem concerns the qualified boundaries of concretetest-components strength’s minimum value.According to the confidence level of therisk control by producers and users，using the density function of a minimum orderstatistic to solve the range of concrete test-components strength’s minimum valuef_cu,min，we have advanced a new discriminant inequality f_cu,min≥f_{cu, k} λ₂^*S_fcu，andthe numerical table of the coefficient λ₂^*.So it can not only resolve the confusionbrought by the mere experience judgment，but also improve the credibility of dataprocessing.Meanwhile, in this paper,we have discussed a density function of an extensionof Laplace distribution,which corresponding to Support Vector Machines, and theMLEs of the parameters have been calculated.The problem of prediction of concretecompressive strength is the hot topics about building’s construction, when we analyzeit’s influence factors in statistical methods,the collected data are basically smallsample and multivariate.In this paper,we establish models to predict concretecompressive strength by SVM, the result shows that SVM has certain advantages forthe research of small sample.