Fatigue Life Classification And Prediction Of Welding Structure Based On Probabilistic Neural Network

With the development of artificial neural networks, Researchers both home and abroadproposed neural network technology is applied to the prediction of the fatigue life, socombined with the references and research results. In this paper, the probabilistic neuralnetwork is used to classify and predict the fatigue life of welded structures. the fatigue life ofwelded joints is estimated which is based on the stress-life calculation model that is presentedby forerunners.The probabilistic neural network is mainly used for classification, based on this feature, inthis study the model is improved, the system of the fatigue life of the classification andprediction is established. Based on the other researcher work, In this paper, the main worksare summarized as follows: First of all, according to the principle and procedure ofprobabilistic neural network to establish this system, the fatigue experimental data of buttjoints and corner joints collect is trained and learned, the cycle times2×106of fatigue limit istreated the boundary of classification, It is divided into limited life and infinite life, then, getthe classification results; Secondly, In the quantitative calculation layer of the model use anew fatigue reliability calculation method which is be putted forward by predecessors, Use itto estimate the life of boom parts of the pump,Stress-life model is treated as computationmodel, we can obtained the fatigue life of structures when know all kinds distribution of stressthat load in components and which is in the required reliability. It makes the calculation moreaccurate. To facilitate simple understand and operate this system for users, In this paper, InMATLAB language write the corresponding user operation interface (GUI), when user clickson the specific button, can get the results accordingly.