Study For Comparing The Parameters Of Several Life Distributions

The comparison of several life distributions has significant meaning in both theory field and application field. For instance, the chief problem is whether shape parameters could be equal in the comparison of Weibull distributions. Different groups of Weibull distributions have the same shape parameter is an important hypothesis in many application problems. When shape parameters are equal, the ratio of scale parameters of two Weibull distributions is equivalent to the ratio of p-quantiles and also the ratio of mean values, and two hazard functions are in proportion, survival functions can also be derived with each other.This paper, by using two method, generalized value method and likelihood ratio test study life distribution of comparison, namely test multiple life distribution parameters is the same hypothesis. Likelihood ratio test is a kind of common thinking and method, its theory applicable scope is very extensive, the main reason is the unknown general probability distribution does not require too much limitation, and in large sample cases, inspection effect is more ideal. Generalized value test was first in order to solve the unilateral hypothesis testing because of the existence of extra parameters to obtain the accurate inspection problems and proposed. A large amount of research shows that when can not use conventional precise inspection method, using generalized value and general confidence interval method can get effective inspection and confidence interval.Generalized exponential distribution is relatively new a life distribution, it in the geophysical signal and heart sound signal processing has a big role. This paper considers the two generalized index distribution of the scale parameter, shape parameters and reliability comparison, using the parameter of the pivot quantity structure generalized inspection variables, and gives corresponding generalized value and general confidence interval. Secondly is deduced more generalized exponential distribution compared likelihood ratio statistic.Distribution is the most widely used one of the life distribution, it can be used for various types of service life of product distribution model, such as valve, ball bearing and electric insulation material, etc. Considering it is based on destiny successive truncated samples, shape parameter and the comparison of the scale parameter, and give more parameter comparison of the likelihood ratio method. Successive truncated life test is used in actual life test types, there has been discussed based on the literature of successive truncated life data comparison of distribution. Therefore, in view of the present research status and theoretical research needs, this paper first studies the distribution based on destiny successive truncated sample parameter comparison.Finally this paper through the numerical simulation compared the generalized test variable and likelihood ratio statistic effect. Simulation results show that the two methods have advantages and disadvantages, and the generalized value determined inspection made the first kind of wrong probability and given significance level more close to, and has convenient calculation, easy used in small sample characteristics. Therefore, this paper holds that the generalized value method is worth us to further discuss.