Engineering machinery is one of the most important equipment in modern industry, and its small fault may bring great harm to the society, due to the large working strength and frequency. On the other hand, the enterprise economic benefit is directly related to the safety and stability operation of the engineering machinery. It will cause a great waste of resources, if all the equipment are deactivated according to a fixed number of years, without considering the health status of engineering equipment. At the same time, the structural damage originates from the material defect, the local stress concentration and other factors. The structural damage process experiencing from microscopic initiation to expansion, forming from the hole to the empty, and experiencing from unknown to considerable physical evolution. The influence factors of the structural damage across the time and space, and also including known and unknown. The composite effect of these multiple level factors will affect the accuracy of prediction results. Therefore, the fatigue life prediction and reliability assessment for the engineering machinery structure need to be considered as a complex problem, which is also need to be solved in the modern engineering practice.
In this paper, aiming to solve the problem existed in the current fatigue life prediction and reliability assessment for structure, the prediction approach for fatigue life in the crack initiation phase and the prediction approach for crack propagation are studied based on the theory of multiple factors correction for fatigue phenomenology. And the Bayesian theory is used to fuse the prediction result with the test data, in order to provide a new approach for the accurate assessment of structural reliability. The main work of this paper can be shown as follows:
(1)The influence factors of fatigue life are selected as the core of research, and based on which a new prediction approach for fatigue life is proposed by combine the quantitative correction factors with the fatigue phenomenological model. And the quantitative expression between structure factor, stress ratio, processing approach and the fatigue life is achieved, which further provide a relevant approach for the prediction of structural crack initiation life. At the same time, the feasibility and the accuracy of proposed approach are verified by the case study. And the sensitivity of correction factors to the fatigue life and its change tendency are analyzed. It can be found from the results that the proposed prediction approach is feasible for the prediction of structural fatigue life, and the prediction result is consistent with the test result.
(2)An prediction approach for crack propagation is proposed by correction with the extracting parameters from loading process and the vector feature of crack propagation, which also considerate the structural factors, stress ratio, the quality of processing and the interaction of loading. The proposed approach can be used for the prediction of crack propagation under constant load and random load. The crack propagation of case studies are predicted, and the acoustic emission equipment is applied for the monitoring experiment research. It can be clearly found from the results that the prediction accuracy of proposed model based on multiple factors correction can be guaranteed. Also the relationship between the correction factors and the crack length is analyzed by the proposed approach. It can be shown from the results that the crack propagation length is inversely proportional to the loading factor and the surface quality factor.
(3)A dynamic model of residual stress redistribution with cracks or damage extending is proposed by combine the damage theory with the calculation method of residual stress redistribution, from the perspective of micro defects derived. And an analysis approach for crack propagation with consideration of the residual stress is proposed, by using the increment selection criterion. It can be clearly found from the results that the release of residual tensile stress will decline the crack propagation rate. The neglecting of the change in residual stress distribution will overestimate the crack propagation rate, and to achieve a more conservative result.
(4) In view of the exponential distribution and Weibull distribution widely exists in the data of structural reliability assessment, the assessment approach for reliability is established by using the Bayesian theory to combine the theoretical prediction model with the test data. The proposed approach implements the component reliability in the process of crack propagation, and further effectively improves the information fusion strategy between the priori information and the test data. The posterior information is solved by using the MCMC algorithm, and the structural reliability value is calculated. It can be clearly found from the results that the distribution of posterior parameter is concentrated, and the confidence is 90%. So the evaluation accuracy of proposed approach can be guaranteed.
(5) Aiming to solve the reliability assessment for same structure under different initial damage status, a kind of reliability evaluation model under different initial state is proposed by combine the strategy of reliability ordering model with the Bayesian theory. The effective integration between the process parameter and test data is achieved by using the progressive factors between different data. The structural reliability under different crack initial states are accessed by the test information of case studies. It can be clearly found from the results that the average prediction accuracy is 92%.