Research On Reliability Evaluation For Mechanical Products Based On Degradation Modelling

Reliability evaluation is fundamental technique in reliability engineering, and plays an important role in the product life cycle, including the stages of R & D, production, storage, use and maintenance. The traditional modellings and methods of reliability evaluation are mainly based on the classical statistics method which often utilizes lots of failure data.The development of new technology on design for reliability has greatly enhanced reliability and quality of mechanical products. It will take an extremely long time for a mechanical product to fail, even if it is operated under severe conditions. The traditional methods of reliability evaluation can hardly be used for high-reliability and long-life mechanical products. A promising way for reliability modelling of highly reliable products is to make use of degradation data that reflects the health conditions of a product. Bayesian approach has a notable advantage in the field of information fusion, and has been widely used in reliability engineering in recent years. Most of the new methods based on degradation models and Bayesian theory are developed based on electronic products. So, they may not very suitable or effective for mechanical products. Thus, in this dissertation, some relevant work has been done on the problem of researching on the reliability evaluation methods for mechanical products based on the degradation models, and the main contents and achievements of this dissertation are summarized as follows:(1)Reliability assessment of mechanical products is suffered from the difficulty induced by no failure dada and small sample size. The general degradation path model is used to generate to deal with the problem of no failure data. The Bayesian method is implemented to construct the reliability assessment framework by fusing the available field data,expert information and pseudo-lifetime data. Finally, reliability assessment is carried out based on this Bayesian model. To illustrate the approach, an application to a milling head of a gantry machining center is investigated.(2)The basic Wiener process degradation model, gamma process degradation model and inverse Gaussian degradation process are introduced. Heterogeneity among different samples is studied and handled by introducing unit-specific random effects into the gamma process degradation model and inverse Gaussian process model. The method of estimating parameters for the two extended stochastic process models are proposed based on Bayesian approach. Then, the extended gamma process is adopted to describe the internal structure degradation of a spool valve based on the failure mechanism analysis. A classic example of a GaAs Laser device is presented to demonstrate the applicability of the extended inverse Gaussian process model and the Bayesian method of estimating parameters.(3)Combining a nonlinear damage accumulation model, a probabilistic S-N curve, and a one-to-one probability density functions transformation technique, a general probabilistic methodology for modeling damage accumulation is developed to analyze the time-dependent fatigue reliability. The damage accumulation is characterized as a distribution in a general degradation path, which captures a nonlinear damage accumulation phenomenon under variable-amplitude loading conditions; its mean and variability change with time. The proposed methodology is then validated by experimental data obtained for a railway axle(45 steel and LZ50 steel). The time-dependent fatigue reliability is analyzed and demonstrated through probabilistic modeling of cumulative fatigue damage, and good agreement between the predicted results and the experimental measurements under different variable amplitude loadings is obtained.(4)A general approach for reliability evaluation based on degradation modeling, considering multiple degradation measures is introduced in this dissertation. Inverse Gaussian process model is incorporated with Copula function to construct a new general multiple degradation process model. Previous research of multiple degradation analysis based on competing risk model. However, the competing risk model is not always suitable for mechanical products. This dissertation presents a new reliability model for multiple degradation processes analysis to handle this non-competing relationship. The proposed models and methods are validated through the illustrative examples.(5)A Bayesian approach for the optimal design of degradation test is proposed in this dissertation. Other than an optimal design with pre-estimated planning values of model parameters, we handle the situation with uncertainty in the planning values using the Bayesian method. An average pre-posterior variance of reliability is used as the optimization criterion. A trade-off between sample size and number of degradation observations is investigated in the degradation test planning. The effects of priors on the optimal designs and the value of prior information are also investigated and quantified. Based on the inverse Gaussian process model, an application to the degradation test planning of a GaAs Laser device is used to demonstrate the proposed method.