Study on Parameter Distribution in Structure Reliability Analysis: Machine Learning Algorithm and Application

Abstract

The discrimination of parameter probability distribution type is the key to structure reliability analysis. A support vector machine (SVM) intelligent recognition model of probability distribution law is presented aiming at traditional method disadvantage. The intelligent recognition model of probability distribution is constructed by SVM algorithm realization, network design and feature extraction, inward stress probability distribution type of a stem structural member is recognized by the model, recognition result is Weibull distribution, SVM has a good generalization ability and clustering ability by comparison between network recognition result and regression analysis, the experiment result shows total recognition rate achieved 98.25%, it provides a good new method for structure reliability analysis.