Search results

The purpose of this paper is to analyze the dependent competing failure reliability of harmonic drive (HD) with strength failure and degradation failure.

Based on life tests and stiffness degradation experiments, Wiener process is used to establish the accelerated performance degradation model of HD. Model parameter distribution is estimated by Bayesian inference and Markov Chain Monte Carlo (MCMC) and stiffness degradation failure samples are obtained by a three-step sampling method. Combined with strength failure samples of HD, copula function is used to describe the dependence between strength failure and stiffness degradation failure.

Strength failure occurred earlier than degradation failure under high level accelerated condition; degradation failure occurred earlier than strength failure under medium- or low-level accelerated condition. Gumbel copula is the optimum copula function for dependence modeling of strength failure and stiffness degradation failure. Dependent competing failure reliability of HD is larger than independent competing failure reliability.

The reliability evaluation method of dependent competing failure of HD with strength failure and degradation failure is first proposed. Performance degradation experiments during accelerated life test (ALT), step-down ALT and life test under rated condition are conducted for Wiener process based step-down accelerated performance degradation modeling.

The purpose of this paper is to analyze the time-dependent reliability of harmonic drive.

The transient finite element analysis (FEA) of harmonic drive is established to calculate the stress under different loads. Combined with the residual strength model and random variables, the time-dependent reliability model of harmonic drive is deduced by the stochastic perturbation method and Edgeworth series. Based on accelerated life tests, the degradation parameters are estimated by maximizing likelihood function. Under variable load, the key stress from transient FEA is transformed into probability density function by kernel density estimation, and the residual strength model is modified by adding adjustment factors to deal with strength degradation under different loads.

The critical position of stress concentration from transient FEA is consistent with the fatigue fracture position at the accelerated life test sample. Compared with the time-dependent reliability method with equivalent circular-shell static stress or empirical degradation parameters, the proposed method has the smallest prediction error of failure life. Under variable load, the state function should be expanded to second-order series for avoiding error items relevant to variance. The failure life expectation under random variable load is smaller than that under constant load.

The time-dependent reliability method of harmonic drive is firstly proposed under constant and variable load. The transient FEA of harmonic drive is established to calculate the stress for strength analysis. The accelerated life test of harmonic drive is conducted for degradation parameters estimation. The adjustment factor is added to the residual strength model for strength degradation under different loads.