Research On Stress Concentration, Size And Surface Impacting On Metal Fatigue

With worldwide energy crisis and price increases, competition is increasingly fierce, forcing most of technical personnel to use fatigue theory and methods to create a safe and economical products. Today, people of the fatigue failure analysis has been more profound understanding of, but on its strength in terms of influencing factors, the current research results there is still a certain gap, but also can not provide for the practical application. Therefore, the fatigue factor affecting the issue is still a need to be on the subject.The S-N curve and fatigue limit, the only representative sample of the standard smooth fatigue limit and the actual parts of the size, shape and surface is a wide range of the standard sample are very different. Therefore, the fatigue strength in parts of the design must take these factors into account the effects of fatigue. Fatigue strength of the mechanical parts affected by a lot of factors, stress concentration, parts size, surface, environmental media, loading sequence and frequency, before the three most important.This article analyzed the impact of fatigue from stress concentration, size effect and surface finished. After summing up and comment on the relationship of Kf and Kt, found that the existing relations between the two all-linear relationship. In a lot of fatigue test data analysis, this paper presented the fatigue notch factor and the theoretical stress concentration factor had polynomial relations, that related to the heat treatment methods. Use different test methods in the course of the size of the impact of fatigue on the production of detailed analysis. It analyzed the relationship of surface morphology and micro-stress concentration. In considering the surface processing of fatigue, the formula for the sensitivity of fatigue notch factor had been amended. Improved BP neural network analyzed the impact of stress concentration, size and surface finished on fatigue. Established a dual-layer BP neural network model. Traininged it and certificated it in MATLAB.