| This paper takes a kind of discrete degradation data into consideration. We suppose that the degradation variables are the number of defects appear on the products. We establish homogeneous markov chain based on the new defects occur in each unit time interval. The markov chain can be modified to a stationary one so that we can get the asymptotic distribution of the degradation variables. The critical threshold value plays a very important role in the estimation of product reliability. In practice, the critical threshold value can vary appreciably among users. In this case, a probabilistic, rather than a deterministic threshold value is more appropriate. So we do some comparisons between these two kind of critical threshold value. Meanwhile, we consider that: the failure of product comes from the competition of paroxysmal failure and degenerative failure. We establish exponential regression model between these two failure modes, and estimate parameters through Newton-Raphson method. Then reliability of this kind of product can be obtained. |
