Search results

In this study, the authors investigate methods of sequential analysis to test prospectively for the existence of a unit root against stationary or explosive states in a p-th order autoregressive (AR) process monitored over time. Our sequential sampling schemes use stopping times based on the observed Fisher information of a local-to-unity parameter. In contrast to the Dickey–Fuller (DF) test statistic, the sequential test statistic has asymptotic normality. The authors derive the joint limit of the test statistic and the stopping time, which can be characterized using a 3/2-dimensional Bessel process driven by a time-changed Brownian motion. The authors obtain their limiting joint Laplace transform and density function under the null and local alternatives. In addition, simulations are conducted to show that the theoretical results are valid.

The purpose of this paper is to consider the general k level step-stress accelerated life test with the Rayleigh lifetime distribution for units subjected to stress under progressive Type-I censoring.

The parameter of this distribution is assumed to be a log-linear function of the stress, and a tampered failure rate model holds. The progressive Type-I censoring reduces the cost of testing. Due to constrained resources in practice, the test design must be optimized carefully. A numerical study is conducted to illustrate the optimum test design based on several four optimality criteria under the constraint that the total experimental cost does not exceed a pre-specified budget.

This paper compares unconstrained and constrained optimal k level step-stress test. Based on the results of the simulation study, the cost constraint reduces cost and time of the test and it also, in the most cases, increases the efficiency of the test. Also, the T-optimal design is lowest cost and time for testing and it is found more optimal in both conditions.

In this paper, various optimization criteria for selecting the stress durations have been used, and these criteria are compared together. Also, because of affecting the stress durations on the experimental cost, the author optimize under the constraint that the total experimental cost does not exceed a pre-specified budget. The efficiency of the unconstrained test in comparison with constrained test is discussed.

Accelerated life test is undertaken to induce early failure in high-reliability products likely to last for several years. Most of these products are exposed to several fatal risk factors and fail due to one of them. Examples include solar lighting device with two failure modes: capacitor failure, and controller failure. It is necessary to assess each risk factor in the presence of other risk factors as each one cannot be studied in isolation. The purpose of this paper is to explore formulation of optimum time-censored accelerated life test model under modified ramp-stress loading when different failure causes have independent exponential life distributions.

The modified ramp-stress uses one test chamber in place of the various chambers used in the normal ramp-stress accelerate life test thus saving experimental cost. The stress-life relationship is modeled by inverse power law, and for each failure cause, a cumulative exposure model is assumed. The method of maximum likelihood is used for estimating design parameters. The optimal plan consists in finding out relevant experimental variables, namely, stress rate and stress rate change point(s).

The optimal plan is devised using D-optimality criterion which consists in finding out optimal stress rate and optimal stress rate change point by maximizing logarithm of determinant of Fisher information matrix to the base 10. This criterion is motivated by the fact that the volume of joint confidence region of model parameters is inversely proportional to square root of determinant of Fisher information matrix. The results of sensitivity analysis show that the plan is robust to small deviations from the true values of baseline parameters.

The model formulated can help reliability engineers obtain reliability estimates quickly of high-reliability products that are likely to last for several years.

The purpose of this paper is to design a simple step-stress model under type-I censoring when the failure time has an extension of the exponential distribution.

The scale parameter of the distribution is assumed to be a log-linear function of the stress and a cumulative exposure model is hold. The maximum likelihood estimates of the parameters, as well as the corresponding Fisher information matrix are derived. Two real examples are given to show the application of an extension of the exponential distribution in reliability studies and a numerical example is presented to illustrate the method discussed here.

A simple step-stress test under cumulative exposure model and type-I censoring for an extension of the exponential distribution is presented.

The work is original.

Accelerated life tests (ALTs) are used to make timely assessments of the lifetime distribution of highly reliable materials and components. Life test under accelerated environmental conditions may be fully accelerated or partially accelerated. In fully accelerated life testing, all the test units are run at accelerated condition, while in partially accelerated life testing, they are both run at normal and accelerated conditions. The products can fail due to one of the several possible causes of failure which need not be independent. The purpose of this paper is to design constant-stress PALT with dependent competing causes of failure using the tampered failure rate model.

Gumbel–Hougaard copula is used to model and measure the dependence between the life times of competing causes of failure. The use of the copula simplifies the model specification and gives a general class of distributions with the same dependent structure and arbitrary marginal distributions.

The optimal plan consists in finding optimum allocation of test units in different chambers by minimizing the reciprocal of the determinant of Fisher Information Matrix. The confidence interval for the estimated values of the design parameters has been obtained and sensitivity analysis carried out. The results of sensitivity analysis show that the plan is robust to small deviations from the true values of baseline parameters.

The model formulated can help reliability engineers obtain reliability estimates quickly of high reliability products that are likely to last for several years.

Statistical methods of spatial analysis are often successful at either prediction or explanation, but not necessarily both. In a recent paper, Dearmon and Smith (2016) showed that by combining Gaussian Process Regression (GPR) with Bayesian Model Averaging (BMA), a modeling framework could be developed in which both needs are addressed. In particular, the smoothness properties of GPR together with the robustness of BMA allow local spatial analyses of individual variable effects that yield remarkably stable results. However, this GPR-BMA approach is not without its limitations. In particular, the standard (isotropic) covariance kernel of GPR treats all explanatory variables in a symmetric way that limits the analysis of their individual effects. Here we extend this approach by introducing a mixture of kernels (both isotropic and anisotropic) which allow different length scales for each variable. To do so in a computationally efficient manner, we also explore a number of Bayes-factor approximations that avoid the need for costly reversible-jump Monte Carlo methods.

To demonstrate the effectiveness of this Variable Length Scale (VLS) model in terms of both predictions and local marginal analyses, we employ selected simulations to compare VLS with Geographically Weighted Regression (GWR), which is currently the most popular method for such spatial modeling. In addition, we employ the classical Boston Housing data to compare VLS not only with GWR but also with other well-known spatial regression models that have been applied to this same data. Our main results are to show that VLS not only compares favorably with spatial regression at the aggregate level but is also far more accurate than GWR at the local level.

Zero-failure reliability testing aims at demonstrating whether the product has achieved the desired reliability target with zero failure and high confidence level at a given time. Incorporating accelerated degradation testing in zero-failure reliability demonstration test (RDT) facilitates early failure in high reliability items developed within short period of time to be able to survive in fiercely competitive market. The paper aims to discuss these issues.

The triangular cyclic stress uses one test chamber thus saving experimental cost. The parameters in model are estimated using maximum likelihood methods. The optimum plan consists in finding out optimum number of cycles, optimum specimens, optimum stress change point(s) and optimum stress rates.

The optimum plan consists in finding out optimum number of cycles, optimum specimens, optimum stress change point(s) and optimum stress rates by minimizing asymptotic variance of estimate of quantile of the lifetime distribution at use condition subject to the constraint that total testing or experimental cost does not exceed a pre-specified budget. Confidence intervals of the design parameters have been obtained and sensitivity analysis carried out. The results of sensitivity analysis show that the plan is robust to small deviations from the true values of baseline parameters.

For some highly reliable products, even accelerated life testing yields little failure data of units in a feasible amount of time. In such cases accelerated degradation testing is carried out, wherein the failure termed as soft failure is defined in terms of performance characteristic of the product exceeding its critical (threshold) value.

The step-stress accelerated test is the most appropriate statistical method to obtain information about the reliability of new products faster than would be possible if the product was left to fail in normal use. This paper presents the multiple step-stress accelerated life test using type-II censored data and assuming a cumulative exposure model. The authors propose a Bayesian inference with the lifetimes of test item under gamma distribution. The choice of the loss function is an essential part in the Bayesian estimation problems. Therefore, the Bayesian estimators for the parameters are obtained based on different loss functions and a comparison with the usual maximum likelihood (MLE) approach is carried out. Finally, an example is presented to illustrate the proposed procedure in this paper.

A Bayesian inference is performed and the parameter estimators are obtained under symmetric and asymmetric loss functions. A sensitivity analysis of these Bayes and MLE estimators are presented by Monte Carlo simulation to verify if the Bayesian analysis is performed better.

The authors demonstrated that Bayesian estimators give better results than MLE with respect to MSE and bias. The authors also consider three types of loss functions and they show that the most dominant estimator that had the smallest MSE and bias is the Bayesian under general entropy loss function followed closely by the Linex loss function. In this case, the use of a symmetric loss function as the SELF is inappropriate for the SSALT mainly with small data.

Most of papers proposed in the literature present the estimation of SSALT through the MLE. In this paper, the authors developed a Bayesian analysis for the SSALT and discuss the procedures to obtain the Bayes estimators under symmetric and asymmetric loss functions. The choice of the loss function is an essential part in the Bayesian estimation problems.

Degradation measurement of some products requires destructive inspection; that is, the degradation of each unit can be observed only once. For example, observation on the mechanical strength of interconnection bonds or on the dielectric strength of insulators requires destruction of the unit. Testing high-reliability items under normal operating conditions yields a small amount of degradation in a reasonable length of time. To overcome this problem, the items are tested at higher than normal stress level – an approach called an accelerated destructive degradation test (ADDT). The present paper deals with formulation of constant-stress ADDT (CSADDT) plan with the test specimens subject to stress induced by temperature and voltage.

The stress–life relationship between temperature and voltage is described using Zhurkov–Arrhenius model. The fractional factorial experiment has been used to determine optimal number of stress combinations. The product's degradation path follows Wiener process. The model parameters are estimated using method of maximum likelihood. The optimum plan consists in finding out optimum allocations at each inspection time corresponding to each stress combination by using variance optimality criterion.

The method developed has been explained using a numerical example wherein point estimates and confidence intervals for the model parameters have been obtained and likelihood ratio test has been used to test for the presence of interaction effect. It has been found that both the temperature and the interaction between temperature and voltage influence the quantile lifetime of the product. Sensitivity analysis is also carried out.

Most of the work in the literature on the design of ADDT plans focusses on only a single stress factor. An interaction exists among two or more stress factors if the effect of one factor on a response depends on the levels of other factors. In this paper, an optimal CSADDT plan is studied with one main effect and one interaction effect. The method developed can help engineers study the effect of elevated temperature and its interaction with another stress factor, say, voltage on quantile lifetime of a high-reliability unit likely to last for several years.

In this paper, a mixture of exponential and Rayleigh distributions in the proportions α and 1 − α and all the parameters in the mixture distribution are estimated based on fuzzy data.

The methods such as maximum likelihood estimation (MLE) and method of moments (MOM) are applied for estimation. Fuzzy data of triangular fuzzy numbers and Gaussian fuzzy numbers for different sample sizes are considered to illustrate the resulting estimation and to compare these methods. In addition to this, the obtained results are compared with existing results for crisp data in the literature.

The application of fuzziness in the data will be very useful to obtain precise results in the presence of vagueness in data. Mean square errors (MSEs) of the resulting estimators are computed using crisp data and fuzzy data. On comparison, in terms of MSEs, it is observed that maximum likelihood estimators perform better than moment estimators.

Classical methods of obtaining estimators of unknown parameters fail to give realistic estimators since these methods assume the data collected to be crisp or exact. Normally, such case of precise data is not always feasible and realistic in practice. Most of them will be incomplete and sometimes expressed in linguistic variables. Such data can be handled by generalizing the classical inference methods using fuzzy set theory.