Study On Fatigue Reliability Of Metal Structure System Of Engineering Machine

Fatigue damage is the main destroy mode for the machinery parts and engineering structures under alternate loading over a long period of time. Fatigue is an important subject whether in theory or in engineering. After the Industrial Revolution in the early 19th century, the phenomenon of fatigue was taken into account and studied. The scholars all over the world have reaped many fruits of research; however, there are still many phenomena cannot be explained attracting the scholars to study. The fatigue life is determinate when determination methods are used to analyze fatigue. But the fatigue of structure is a much complicated phenomenon affected by many uncertainties. So it is necessary to analyze the structural fatigue from the point of view of probability and statistics. This is a rational new method to evaluate the structural fatigue security.The random analysis of structure under fatigue loading, the statistic finite element method (SFEM) for fatigue reliability, random fatigue damage accumulation theory and fatigue reliability of structure system were studied in this paper with regard to the metal structure of engineering machine.The main achievements in this paper are as following:1. The analysis of random structure is the base of structural reliability analysis. The structural randomness analysis under cycle loading is of important theoretical significance and practical significance. The analysis of randomness of local stress and strain in strain-based fatigue was studied, the approximate polynomial method was proposed. The approximate polynomial method was based on Neuber's method, the cyclic stress-strain responses and Neuber's rule were treated as probabilistic curves, and the statistic characteristic was obtained from the approximate polynomial. The method is fast and easy for engineering application. If the distribution of probability density of local stress and strain was known, the fatigue life for crack initiation can be predicted under any reliability by the local stress and strain method.2. The iterative formulas of elastoplastic stochastic finite element method under cyclic loading were deduced. The random responses of local multiaxial stress and strain were accurately calculated by the Taylor expansion elastoplasticstochastic finite element method under cyclic loading. According to the complication of cyclic loading the aeolotropism and Bauschinger effect resulted form plastic deformation were reflected by the kinematic hardening model, the transient stress-strain relation was described by Jhansale model in the iterative formulas. The elastoplastic finite element method overcomes the limitation that only the uniaxial local stress and strain could be calculated by the approximate method, establishing the base of multiaxial fatigue analysis. The strain-based fatigue reliability can be calculated if the proposed method was used with reliability theory.3. The SFEM reliability method applied to components' fatigue reliability was studied. The performance functions of fatigue reliability were set up and their partial derivative were deduced. The direct SFEM method and the partial derivative SFEM method were proposed. The SFEM results were used directly to calculate the fatigue reliability of components in the direct SFEM method. The method is simple and easy any SFEM could be used. The results are approximate unless the distribution of probability density of nominal stress was known. The accuracy of reliability calculation was improved in the derivative SFEM method. The local derivatives of FEM and numerical method were used to obtain the iterative formulas of the performance function's local derivatives. The procedure is relatively simple and adaptable to engineering application as the stiffness matrix formed and decomposed only once.4. The establishment principle and the applicability of the three improved Miner's rules were analyzed. The rules were examined by a constant amplitude loading. The results proved the rationality of the statistic Miner's rule. The approximate...