Small Sample Algorithm For Structural Fatigue And Reliabilty Analysis Based On Historical Data

Aeronautic and astronautic products are small in quantity and high in cost while they need high safety and reliability. So it leads to lots of limits in experimental time and expense. Thus, algorithms for small samples are necessary and popular. Algorithms for small samples are different from statistic methods which are widely used today. The ones are based on the infinite collectivity assumption that we can not get the results as accurate as we need with small samples. So it has important significance to find out an algorithm from the finite information.In this paper, recent studies on small samples of structural fatigue are reviewed, classified as definition and solution. Methods put forward recently are discussed. Based on that, algorithms for fatigue and reliability analysis are proposed which are based on the historical data.The algorithms proposed in this paper are all in the situation of small samples. One study of fatigue life analysis is made on the assumption that fatigue life follows three-parameter Weibull distribution. Based on S-N curve, the empirical data were translated to the current test status by the weighted least square method. Estimated parameters modified with current test data, an approach to determine the fatigue life distribution is given. Moreover, a reliability model of structural components under constant amplitude loading is presented, which is based on the distribution of instantaneous crack length. The model proposed the steps for computing parameters of the Paris-Erdogan's crack propagation equation and established the expectation and variance of the crack length at a given time. Consequently, a crack length-based reliability model is set up.Methods introduced in this paper are efficient and convenient for calculation. Several numerical examples are employed to demonstrate their suitability. The results show that they all have good agreements with the experiment data.