Baixe o app para aproveitar ainda mais
Prévia do material em texto
An Updated Bibliography of Accelerated Test Plans Wayne B. Nelson, Fellow, IEEE Key Words: Accelerated testing, test plans, overstress, life, degradation, extrapolation, regression models, minimum variance. 1 I TRODUCTIO Purpose. This paper describes the following bibliography of statistical plans for accelerated tests. This introduction provides background on accelerated testing and test plans. The body of this paper briefly surveys the literature on theory for accelerated test plans and the considerations that go into choosing or developing such plans. Many plans need to be developed, especially for degradation models. The author welcomes additions to and corrections of the bibliography, which is available on request as a searchable updated Word file, which includes email addresses of authors. Overview. This introduction briefly describes accelerated testing, the history of test plans, accelerated test (AT) models, the simple model, AT plans, and considerations in developing such plans. Accelerated testing. Accelerated testing consists of high stress testing that shortens product life or hastens degradation of product performance. Here "stress" is used in a general sense to mean any accelerating variable. The usual purpose of such overstress testing is to estimate the life or degradation of the product at lower stress levels encountered in normal use. This is done by statistically fitting an acceleration model to the test data and extrapolating the fitted model to estimate product life or degradation at a normal stress level. Tested under normal use stress levels, such products last so long or degrade so little that their life or performance cannot be estimated. Accelerated testing yields such estimates in a much shorter time. In this article, the purpose of accelerated testing is to measure product reliability or degradation. Nelson [94,97] and Meeker and Escobar [82] present in detail many models, statistical methodologies, and applications of accelerated testing. An entirely different purpose of accelerated testing is to force the product to fail to discover failure modes that would occur in actual use. Then the product or process is improved to reduce those failure modes. Such testing is used during product development or debugging of the production process and includes HALT, HASS, and environmental stress screening. This bibliography does not include references to such techniques as they are nonstatistical and do not yield estimates of product life. History of test plans. Test plans for analysis-of-variance and regression models have a long history, starting well before the work of R.A. Fisher, and such plans are still being developed. A number of general references on such test planning and optimization are included in the bibliography. Virtually all of those test plans are intended to estimate the product "response" within the test region. That is, they are plans for interpolation. In contrast, AT plans are used for extrapolation outside the range of test stress levels. Chernoff [27] developed the first accelerated life test plans; these involved the simple exponential distribution, two life-stress relationships, and censored data. Nelson and coworkers [67,87,98,99] were first to develop plans for the more realistic lognormal and Weibull life distributions. Nelson [94,97] provided the first book chapter on AT plans and briefly summarized the limited literature then. Since then work on AT plans has produced the 245 references in this updated bibliography. AT plans are included in the bibliography only if they include the (possibly asymptotic) variance of some estimate and preferably reduce that variance. Also, some general references on optimum design are included. Since HALT, HASS, and environmental stress screening are used only to reveal failure modes and do not yield estimates of product life, references to them are excluded. AT models. The following view of accelerated life test (ALT) models is used throughout this article. Such a model consists of a theoretical life distribution whose parameters are functions of accelerating variables (and possibly other engineering variables) and unknown coefficients to be estimated from the test data. For example, the Arrhenius- lognormal model consists of the lognormal distribution, an Arrhenius temperature relationship for the lognormal median, and a constant parameter. An ALT plan is designed to perform well for a particular model and test situation and may perform poorer for other models. An accelerated degradation test (ADT) model consists of a distribution of performance whose parameters are a function of accelerating variables and the time on test. Comprehensive references on statistical models and analyses for AT data are Nelson [94,97] and Meeker and Escobar [82]. Simple model. Most ALT plans have been developed for the following simple model for constant stress testing. The life distribution for y = ln(t), the log of life t, has a location parameter and a scale parameter . That is, the cumulative distribution function (cdf) for the population fraction failed by log age y is F(y) = G[(y )/ ] ; here G[ ] is the standardized cdf with = 0 and = 1. This includes the exponential, Weibull, lognormal, and some other distributions as special cases. The location parameter is a simple linear function of the (possibly transformed) accelerating variable x. That is, (x) = 0 + 1 x ; here 0 and 1 are unknown coefficients to be estimated from test data. Such a life-stress relationship is usually based on physical theory; examples include the Arrhenius, inverse power, exponential, and other widely used relationships. The cdf can then be expressed as 978-1-4799-6703-2/15/$31.00©2015 IEEE F(y; x, 0, 1, ) = G[(y 0 1x)/ ]. Usually one seeks to accurately estimate a specified F fractile yF of the distribution at a specified low stress level x0 (an actual use level below the test stress levels), namely, yF(x0) = (x0) + zF = 0 + 1 x0 + zF ; here zF is the F fractile of the standardized distribution G[ ]. yF(x0) is a function of the three unknown model parameters 0, 1, and , which all must be estimated from the test data. This simple model and its special cases are presented in detail in Nelson [94,97] and Meeker and Escobar [82]. Test plans. An accelerated life test (ALT) plan with constant stress consists of the chosen test conditions, the number of specimens at each test condition, and the prechosen censoring time of each specimen. For example, for the simple model, there may be three test stress levels x1, x2, and x3, corresponding numbers of specimens n1, n2, and n3, and corresponding chosen censoring times 1, 2, and 3. An ALT test can involve accelerating and other engineering variables (e.g., materials, designs, environments, process variables, etc.). In a test with constant accelerating stresses, a test condition is a specified set of constant values for the accelerating and engineering variables. In a test with time- varying accelerating stresses, the test condition is a specified varying pattern for each stress and specified constant values for the engineering variables. The references present such plans that have good performance, which has been evaluated using analytic theory or simulation. Test plans for traditional design of experiments for analysis-of-variance models and for least-squares regression models usually have equal allocation of specimens to test conditions. This is appropriate to get accurate estimates within the test region. In contrast, AT plans are used for extrapolation outside the test region, and unequal allocation of specimens yields more accurate extrapolated estimates. Then more specimens are tested at low stress than at high. The bibliography has a few references on plansfor accelerated degradation tests (ADT). Considerations. Such plans deal with considerations including life distributions (exponential, Weibull, lognormal, etc.), relationships (life-stress or degradation-stress), optimization criteria (minimum variance, determinant, cost, etc.), data censoring (right and interval), estimation method (maximum likelihood, least squares, BLUEs, etc.), types of stress loading (constant. step, ramp, cyclical, specified profile, stochastic, field use), constraints (on test region, nonsimultaneous testing, etc.) robustness (to model errors, loss of specimens, etc.) specimens (size and geometry), The following section briefly discusses these considerations and cites some relevant articles. 2 CO SIDERATIO S Purpose. This section briefly discusses the preceding considerations for AT plans. Distributions. Most ALT plans have been developed for the exponential, Weibull, and lognormal life distributions. References on a particular life distribution are easy to locate in the bibliography, since the distribution usually appears in the title. In practice, the exponential distribution is rarely suitable, but plans for it may provide insight into plans suitable for more realistic distributions like the Weibull and lognormal. There is a need for plans suited to other life distributions used in practice. Relationships. Most ALT plans have been developed for models with a simple relationship between the life distribution mean (or a percentile) and a single accelerating stress (no other engineering variables). Such simple relationships are widely used in practice, and special cases include the Arrhenius, inverse power, exponential, and other relationships. Most of these are log-linear relationships between the distribution typical life and the (possibly transformed) accelerating variable. Few references deal with test plans for relationships with more than a single accelerating variable. These few include Nelson [97, Ch. 6, Sec. 3] and Escobar and Meeker [82]. More complicated ALT models can involve a second relationship for the other distribution parameter, such as the Weibull shape or the lognormal . Meeter and Meeker [88] and Kim and Bai [68] deal with AT plans for such relationships. As most accelerated degradation models have relationships that are unique to the application, and the relationship is rarely appears in the title of a reference. Optimization criteria. Various optimization criteria have been used to design ALT plans. Most references develop plans that reduce or minimize the asymptotic variance of the maximum likelihood (ML) estimator of a specified fractile at a specified low stress condition. Some plans minimize the asymptotic variance of an ML estimator for a model parameter or a function of the parameters, such as the population fraction failing by a specified age at a specified low design stress. In contrast, for a model with two variables, Escobar and Meeker [44] minimize the determinant of the covariance matrix of the model parameter estimates, an attempt to globally estimate well. Yum and Kim [157] optimize the operating characteristic function of a reliability demonstration test. Little work has been done on cost optimization of AT plans. Some references optimize plans subject to constraints discussed below. Censoring. Some ALT plans require that all specimens be run to failure, which is poor practice for reasons stated by Nelson [96]. Most recent references deal with exact failure times and Type I (prechosen fixed) censoring times. That is, the units at each test condition start at the same time, run together, and have a prechosen common censoring time. Some deal with exact failure times and Type II (failure) censoring times; that is, the units running together at test condition i are stopped at the random time yr of the r-th failure. Type II censoring is rare in practice but has easier and exact statistical theory compared with Type I censoring. A few references deal with interval data (also called read-out data), which result from periodic inspection of the units for failure. These include Yum and Choi [156], Schatzoff and coworkers [116-118], Bai, Kim, and Lee [16], and Seo and Yum [119]. Estimation method. The statistical method used to fit an AT model to test data determines the accuracy of estimators of model parameters and functions of the parameters. Consequently, test plans are designed to provide accurate estimators with a particular method of fitting. Of course, such a plan generally provides good estimators when other fitting methods are used. Various fitting methods are described by Nelson [94,97] and include maximum likelihood, least squares, best linear unbiased, and graphical methods. Most recent references use ML fitting, as it is straightforward, applies to most AT models, has asymptotically normally distributed estimators, and yields approximate variances and confidence limits for estimators. Also, many statistical packages perform ML fitting of AT models to test data; Nelson [97, Chap. 5] describes such current packages. A drawback of ML theory is that the approximate variances and confidence limits for estimators are accurate only for tests with many specimen failures. Small sample properties of ML estimators have been investigated using simulation by McSorley, Lu, and Li [80], Nelson and Kielpinski [98], and Zhang and Meeker [158]. Stress loading. Most AT plans involve constant-stress testing. Work on step-stress plans includes Miller and Nelson [90], Nilsson [100], Schatzoff and Lane [118], DeGroot and Goel [34], Bai, Kim, and Lee [15,16], Bai and Chun [6], and Bai, and Kim [12,13] among many others. For a ramp test (linearly increasing stress), Bai, Cha, and Chung [5] developed plans. In practice, constant-stress tests are easier to carry out but may need to run a long time at low stress to yield failures. Step-stress and ramp tests quickly yield failures, which provide information in a short time. However, the cumulative damage models have not been confirmed in most applications, the accuracy of estimates is usually poor, as test time is short and the failures generally occur at very high stress far from the design stress level, requiring much extrapolation. In general, the standard error of AT estimates is roughly inversely proportional to the total test time, and step-stress and other short tests yield poor accuracy. Test constraints. AT plans generally involve constraints on the censoring times and on the test region (allowed values of accelerating and engineering variables). Most plans have a single common censoring time for all test conditions. This usually presumes that all specimens can be started and run together. For example, in a temperature- accelerated test, this requires a separate oven for each test temperature. An exception, Disch [37] develops a plan where there is one test oven, and different temperature groups of specimens must be run one after another. He optimizes the censoring time for each group, subject to a specified total oven time. Another exception is Bai, Chun, and Kim [8]. Most plans require that the maximum test stress level must not exceed a specified value, determined from practical considerations. There can be constraints on the engineering variables in the AT model; Escobar and Meeker [44] deal with this. In practice, there may be cost constraints on a test. Little cost modeling has been done for AT plans. Robustness. Researchers have investigated plans that are robust to certain practical problems. Distribution error. Any theoretical distribution used in an AT model differs from the actual physical distribution. In many applications, the erroneous fitted distribution is extrapolated to estimatethe lower tail of the physical distribution, usually to estimate low percentiles at a design stress level. To minimize distribution error, some plans have early censoring times at each test condition. Then data come from the lower tail of the distribution at each stress level, and data from the upper tail do not distort the fit. Nelson [96] explains distribution error in detail. Relationship error. Any theoretical relationship used in an AT model differs from the actual physical relationship. In applications, the erroneous fitted relationship is extrapolated to a design stress level. To minimize relationship error, some plans use lower test stress levels. Barton [18] develops such a plan. Nelson [96] explains relationship error in detail. Loss of specimens. Test specimens can be damaged and lost when a test has problems, say, an oven temperature runs amok. For example, when this happens and there just two test stress levels, then the simple life-stress relationship cannot be estimated. Also, no specimens may fail at the lower of just two test stress levels; then, too, the life-stress relationship cannot be estimated. To be robust, many plans use three or more test stress levels to assure that there are failures at two stress levels. For discussions of this issue, see Nelson [94,97 Ch. 6] and Meeker and Hahn [86]. Distribution and parameter uncertainty. In some applications, it is not known whether a Weibull or lognormal distribution should be used to design the test plan. Bayesian work on this problem involves using a prior distribution (a weighted average) of the two distributions. Also, test plans require knowledge of the true unknown model parameter values. Bayesian work replaces the unknown values with a priori distributions for the parameter values. Bayesian references include DeGroot and Goel [34,35], Chaloner and Larntz [24,25], Verdinelli, Polson, and Singpurwalla [137], and Pascual and Montepiedra [110,111]. ReliaSoft [114] provides sensitivity analysis for uncertain parameter values. Compromise plans. Various authors have developed compromise plans that are robust and have good accuracy. Examples include Meeker and Hahn [86], Yang and Jin [144], Pascual and Montepiedra [110,111]. Specimen size. For some products, the test specimens of different sizes may be used. Examples include lengths of cable insulation and lengths of aluminum conductors in microcircuit test structures. Bai and Yun [17] provide some plans for this situation. Software. There is little commercially available software for evaluating or optimizing AT plans. Software of Meeker and Escobar [83] in S-Plus evaluates plans for exact failure data with right (Type I) censoring at specified times. Their software optimizes plans for models with a single accelerating variable and evaluates the performance of completely specified plans for models with more than one variable. ReliaSoft [114] handles similar situations and provides sensitivity analysis of the plans. Software of Schatzoff and coworkers [116,118] evaluates plans for the simple model, inspection data, and step-stress testing. JMP (2015, Ch. 15) has features for evaluating plans specified by the user and for basic optimum plans. Concluding remarks. As stated above, this bibliography is intended to aid practitioners in choosing test plans and to stimulate researchers to develop needed plans. An updated bibliography is available in a searchable Word file from WNconsult@aol.com and includes email addresses of many authors. Please provide corrections and missing references and author email addresses. A search reveals that few references or their abstracts appear on the Web. ACK OWLEDGME T Many authors in the bibliography kindly provided information on their publications and those of others. The author is indebted to Dr. Guangbin Yang and Prof. Hoang Pham, who stimulated him to publish the original bibliography, and to Prof. J.C. Lu for helpful suggestions. BIOGRAPHY Dr. Wayne Nelson is a leading expert on analysis of reliability and accelerated test data. He consults on applications, gives training courses for companies and professional societies, and works as an expert witness. For 24 years he consulted across the General Electric Co. and received the Dushman Award of GE Corp. R&D for developments and applications of product reliability data analysis. He was elected a Fellow of the Amer. Statistical Assoc. (1973), the Amer. Soc. for Quality (1983), the Institute of Electrical and Electronics Engineers (1988) for his innovative developments. He was awarded the 2003 Shewhart Medal and the 2010 Shainin Medal of ASQ and the 2005 Lifetime Achievement Award of IEEE for outstanding developments of reliability methodology and contributions to reliability education. He authored three highly regarded books Applied Life Data Analysis (Wiley 1982, 2004), Accelerated Testing (Wiley 1990, 2004), Recurrent Events Data Analysis (SIAM 2003), two ASQ booklets, and 130 journal articles. Contact him at WNconsult@aol.com ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ A UPDATED BIBLIOGRAPHY OF ACCELERATED TEST PLA S Sep. 20, 2014 This is an updated version of the bibliography published in IEEE Trans. on Reliability . You can request an updated bibliography at WNconsult@aol.com. References with ** were added after publication of the original article. ** Ahmad, N. and Islam, A. (1996), " Optimal Accelerated Life Test Designs for Burr Type xii Distributions under Periodic Inspection and Type I Censoring," aval Research Logistics 43, 1049-1077. [1] Ahmad, N., Islam, A., Kumar, R., and Tuteja, R.K. (1994), "Optimal Design of Accelerated Life Test Plans under Periodic Inspection and Type I Censoring: The Case of Rayleigh Failure Law," South African Statistical Jour. 28, 27-35. [2] Ahmad, N., Islam, A., and Salam, A. (2006), "Analysis of Optimal Accelerated Life Test Plans for Periodic Inspection: The Case of the Exponentiated Weibull Failure Model," Internat'l J. of Quality and Reliability Management 23, 1019-1046. [3] Alhadeed, Abdulla A. and Yang, Shei-Shein. (2002), "Optimum Simple Step-Stress Plan for Khamis-Higgins Model," IEEE Transactions on Reliability R-51, 212- 215.A.Alhadeed@uaeu.ac.ae, SSYang@stat.ksu.edu. ** Alhadeed, Abdulla A. and Yang, Shei-Shein. (2005), "Optimum Simple Step-Stress Plan for Cumulative Exposure Model Using Log-Normal Distribution," IEEE Transactions on Reliability R-54, 64-68. ** Atkinson, A.C. and Donev, A.N. (1993), Optimal Experimental Designs, Oxford Science Publishers. [4] Bagdonavicius, V. and Nikulin, M. (2002), Accelerated Life Models - Modeling and Statistical Analysis, Section 5.6, Chapman & Hall/CRC, Boca Raton, FL. Merely formulates a simple test plan. vilius@sm.u- bordeaux2.from, M.S.Nikouline@u-bordeaux2.fr [5] Bai, D.S., Cha, M.S, and Chung, S.W. (1992), "Optimum Simple Ramp Tests for the Weibull Distribution and Type-I Censoring," IEEE Trans .on Reliability 41, 407- 413. mscha@star.ks.ac.kr [6] Bai, D.S. and Chun, Y.R. (1991), "Optimum Simple Step- Stress Accelerated Life-Tests with Competing Causes of Failure," IEEE Trans. on Reliability 40, 622-627. dsBai@ei1.kaist.ac.kr, yrchun@kyungnam.ac.kr [7] Bai, D.S., Chun, Y.R., and Cha, M.S. (1997), "Time- Censored Ramp Tests with Stress Bound for Weibull Life Distribution," IEEE Trans. on Reliability 46, 99-107. [8] Bai, D.S., Chun, Y.R., and Kim, J.G. (1995), "Failure- Censored Accelerated Life Test Sampling Plans for Weibull Distribution under Expected Test Time Constraint", Reliability Engineering and System Safety 50, 61-68. [9] Bai, D.S. and Chung, S.W. (1991), "An Optimal Design of Accelerated Life Test for Exponential Distribution," Reliability Engineering and System Safety 31, 57-64. swchung@chonnam.chonnam.ac.kr [10] Bai, D.S.and Chung, S.W. (1992), "Optimal Design of Partially Accelerated Life Tests for the Exponential Distribution under Type-I Censoring," IEEE Trans. on Reliability 41, 400-406. swchung@chonnam.chonnam.ac.kr [11] Bai, D.S., Chung, S.W., and Chun, Y.R. (1993), "Optimal Design of Partially Accelerated Life Tests for the Lognormal Distribution under Type I Censoring," Reliability Engineering and System Safety 40, 85-92. [12] Bai, D.S. and Kim, M.S. (1993a), "Optimum Simple Step-Stress Accelerated Life Tests for the Weibull Distribution and Type I Censoring," aval Res. Logistics 40, 193-210. mskim@suwon.ac.kr [13] Bai, D.S. and Kim, M. S. (1993b), "A Comparison of Constant-Stress and Step-Stress Accelerated Life Tests for Weibull Distribution", Bulletin of the 49th Session of the International Statistical Institute, Contributed Papers, Book 1, 65-66. [14] Bai, D.S., Kim, J.G. and Chun, Y.R. (1993), "Design of Failure-Censored Accelerated Life-Test Sampling Plans for Lognormal And Weibull Distributions", Engineering Optimization 21, 197-212. jgkim@yurim.skku.ac.kr [15] Bai, D.S., Kim, M.S., and Lee, S.H. (1989a), "Optimum Simple Step-Stress Accelerated Life Tests with Censoring," IEEE Trans. on Reliability 38, 528-532. dsBai@ie1.kaist.ac.kr [16] Bai, D.S., Kim, M.S., and Lee, S.H. (1989b), "Optimum Simple Step-Stress Accelerated Life Tests under Periodic Observation," J. Korean Statist. Soc. 18, 125-134. [17] Bai, D.S. and Yun, H.J. (1996), "Accelerated Life Tests for Products of Unequal Size," IEEE Trans. on Reliability 45, 611-618. stmyhji@lgcns.com [18] Barton, R.R. (1991), "Optimal Accelerated Life-Time Plans that Minimize the Maximum Test Stress," IEEE Trans. on Reliability 40, 166-172. rbarton@psu.edu [19] Bessler, S., Chernoff, H., and Marshall, A.W. (1962), "An Optimal Sequential Accelerated Life Test," Technometrics 4, 367-379. chernoff@stat.Harvard.edu [20] Boulanger, M. and Escobar, L.A. (1994), "Experimental Design for a Class of Accelerated Degradation Tests," Technometrics 36, 260-272. luis@stat.lsu.edu [21] Box, G.E.P. and Lucas, H.L. (1959), "Design of Experiments in Non-Linear Situations," Biometrika 47, 77-99. [22] Carey, M.B. and Escobar, L.A. (1992), "Experimental Design for a Class of Accelerated Degradation Tests," personal comm. from Dr. M.B. Carey, Quality Assurance Ctr.-2K.503, AT&T Bell Labs, Crawfords Corner Rd., Holmdel, NJ 07733, (201)949-2542. luis@stat.lsu.edu [23] Ceci, C. (1991), "A Stochastic Control Problem Arising in Accelerated Life Tests," research report, Dipartimento di Matematica, Univ. delgi Studi di Roma "La Sapienza," Piazzale Aldo Moro 2, I-00185 Rome, Italy. claudia.ceci@uniroma1.it [24] Chaloner, K. and Larntz, K. (1989), "Optimal Bayesian Experimental Design Applied to Logistic Regression Experiments," J. of Statist. Planning and Inference 21, 191-208. kathryn-chaloner@uiowa.edu [25] Chaloner, K. and Larntz, K. (1992), "Bayesian Design for Accelerated Life Testing," J. of Statist. Planning and Inference 33, 245-259. ** Chaloner, K. and Verdinelli, I. (1995), "Bayesian Experimental Design: A Review," Statistical Science 10, 273-304. [26] Chernoff, H. (1953), "Locally Optimum Designs for Estimating Parameters," Ann. Math. Statist. 24, 586-602. chernoff@stat.harvard.edu [27] Chernoff, H. (1962), "Optimal Accelerated Life Designs for Estimation," Technometrics 4, 381-408. [28] Chernoff, H. (1972), Sequential Analysis and Optimal Design, Conference Board of the Mathematical Sciences Regional Conference Series in Applied Mathematics, No. 8, Soc. for Industrial and Applied Mathematics (SIAM), 3600 Univ. City Science Ctr., Philadelphia, 19104-2688, (800)447-SIAM. [29] Chung, S. W. and Bai, D. S. (1998), " Optimal Designs of Partially Accelerated Life Tests for Weibull Distributions", J. of the Korean Inst. of Indus. Engineers 24, 367-379. swchung@chonnam.chonnam.ac.kr [30] Chung, S. W. and Bai, D. S. (1998), "Optimal Designs of Simple Step-Stress Accelerated Life Tests for Lognormal Lifetime Distributions", Int. J. of Reliability, Quality and Safety Engineering 5, 315-336. [31] Clark, J.A., Garganese, U.S. and Swarz, R.S. (1997). “An Approach to Designing Accelerated Life-testing Experiments,” Proc. Reliability and Maintainability Symposium, 242-248. ** Clyde, M., Müller, P., and Parmgiani, G. (1995), "Optimal Design for Heart Defibrillators," in Case Studies in Bayesian Statistics 2, pp. 278-292, eds. C. Gatsonis, J. Hodges, R.E. Kass, ande N. Singpurwalla, Springer, New York. [32] Condra, L.W. (2001), Reliability Improvement with Design of Experiments, 2nd ed., Marcel Dekker, New York. lloyd.w.condra@boeing.com [33] Daniel, C. and Heerema, N. (1950), "Design of Experiments for the Most Precise Slope Estimation or Linear Extrapolation," J. Amer. Statist. Assoc. 45, 546- 556. [34] DeGroot, M.H. and Goel, P.K. (1979), "Bayesian Estimation and Optimal Design in Partially Accelerated Life Testing," aval Res. Logistics Quar. 26, 223-235. goel@stat.ohio-state.edu [35] DeGroot, M.H. and Goel, P.K. (1988), "Bayesian Design and Analysis of Accelerated Life Testing with Step Stress," Accelerated Life Testing and Experts' Opinion in Reliability: Proceedings of the International School of Physics "Enrico Fermi", eds. D.V. Lindley and C.A. Clarotti, 193-202, North Holland, Amsterdam. [36] Del Grosso, G., Gerardi, A., and Koch, G. (1992), "Optimality in Accelerated Life Tests," Mathematics of Operations Research 17, 866-881. anna.gerardi@uniroma1.it, san_lorenzo@libero.it [37] Disch, Damon (1983), "Optimum Accelerated Sequential Life Tests When Total Testing Time Is Limited," private communication, Lilly Corp. Center, Eli Lilly and Co., Indianapolis, IN 46285. DamonDisch@Lilly.com [38] Elfving, G. (1952), "Optimum Allocation in Linear Regression Theory," Ann. Math. Statist. 23, 255-262. [39] Elfving, G. (1954), "Geometric Allocation Theory," Skandinavisk Akt. Tidskrift 37, 170-190. [40] Elfving, G. (1959), "Design of Linear Experiments," in Probability and Statistics, U. Grenander, ed., Wiley, New York, (800)879-4539, www.wiley.com. [41] Elsayed, E.A. and Jiao, Lixia (2002), “Optimal Design of Proportional Hazards Based Accelerated Life Testing Plans,” International Journal of Materials & Product Technology 17, 411-424. elsayed@rci.rutgers.edu ** Elsayed, E.A. and Zhang, H. (2007), "Design of Optimum Simple Step-Stress Accelerated Life Testing Plans," in Recent Advances in Stochastic Operations Research, pp. 23-28, eds. T Dohi, S. Osaki, and K. Sawaki, World Scientific, Singapore. [42] Erkanli, A. and Soyer, R. (2000), "Simulation-Based Designs for Accelerated Life Testing," J. of Statistical Planning and Inference 90, 335-348. soyer@gwu.edu [43] Escobar, L.A. and Meeker, Jr., W.Q. (1986), "Optimum Accelerated Life Tests with Type II Censored Data," J. of Statistical Computation and Simulation 23, 273-297. luis@lsu.edu [44] Escobar, L.A. and Meeker, Jr., W.Q. (1995), "Planning Accelerated Life Test Plans with Two or More Experimental Factors," Technometrics 37, 411-427. [45] Escobar, L.A. and Meeker, Jr., W.Q. (1998), "Fisher Information Matrices with Censoring, Truncation, and Explanatory Variables, Statistica Sinica 8, 221-237. www.stat.sinica.edu.tw/statistica/oldpdf/A8n112.pdf [46] Escobar, L.A. and Meeker, Jr., W.Q. (2003), "Test Planning for Accelerated Destructive Degradation Tests," talk at the International Conf. on Reliability and Survival Analysis, Univ. of South Carolina. ** Fard, N. and Li, C. (2009), "Optimal Simple Step-Stress Accelerated Life Test Design for Reliability Prediction," J. Statistical Planning and Inference 139, 1799-1808. [47] Fedorov, V.V. (1972), Theory of Optimal Experiments, Academic Press, New York. valeri_v_fedorov@sbphrd.com [48] Ford, I., Titterington, D.M., and Kiss, C.P. (1989), "Recent Advances in NonlinearExperimental Design," Technometrics 31, 49-60. ian@stats.gla.ac.uk , mike@stats.gla.ac.uk [49]Gaylor, D.W. and Sweeny, H.C. (1965), "Design for Optimal Prediction in Simple Linear Regression," J. Amer. Statist. Assoc. 60, 205-216. dgaylor@nctr.fda.gov ** Ge, Z.Z., Li, X.Y., Zhang, J.R., and Jiang, T.M. (2010), "Planning of Step-Stress Accelerated Degradation Test with Stress Optimization," Advanced Materials Research 118, 404-408. [50] George, L.L. (2004), "Design of Accelerated Reliability Tests," Reliability Review 24, 6-28. [51] Ginebra, J. and Sen, A. (1998), "Minimax Approach to Accelerated Life Tests," IEEE Trans. on Reliability 47, 261-267. ginebra@eio.upc.es, amamdas@med.umich.edu ** Gouno, E., Sen, A. and Balakrishnan, N. (2004), "Optimal Step-stress Test under Progressive Type-I Censoring," IEEE Trans. on Reliability 53, 388-393. Correction: (2006) 55, 613-614. evans.gouno@univ-ubs.fr, amamdas@med.umich.edu, bala@mcmaster.ca ** Guan, Q. and Tang, Y. (2013), "Optimal Desighn of Accelerated Degradation Test Based on Gamma Process Models," Chinese J. of Applied Probability 29 (2). ** Guo, H. and Pan, Rong (2007), "D-Optimal Reliability Test Design for Two-Stress Accelerated Life Tests," Proc. IEEE International Conf. on Industrial Engineering and Engineering Management, 1236-1240. rong.pan@asu.edu ** Guo, H. and Pan, Rong (2008), "On Determining Sample Size and Testing Duration," Proc. of the 2008 Reliability and Maintainability Symp. ** Guono, E., Sen, A., and Balakrishnan, N. (2004), "Optimal Step-Stress Test Under Progressive Type-I Censoring," IEEE Trans. on Reliability 53, 388-406. ** Haynes, J., Simpson, J., Krueger, J., and Callahan, J. (1984), "Optimization of Experimental Designs for Temperature Stability Studies," Drug Devel. and Indus. Pharmacy 10, 1505-1526. [54] Herzberg, A.M. and Cox, D.R. (1972), "Some Optimal Designs for Interpolation and Extrapolation," Biometrika 59, 551-561. David.Cox@nuf.ox.ac.uk [55] Hoel, P.G. (1965), "Optimum Designs for Polynomial Extrapolation," Ann. Math. Statist. 36, 1483-1493. << /ASCII85EncodePages false /AllowTransparency false /AutoPositionEPSFiles false /AutoRotatePages /None /Binding /Left /CalGrayProfile (Gray Gamma 2.2) /CalRGBProfile (sRGB IEC61966-2.1) /CalCMYKProfile (U.S. Web Coated \050SWOP\051 v2) /sRGBProfile (sRGB IEC61966-2.1) /CannotEmbedFontPolicy /Warning /CompatibilityLevel 1.4 /CompressObjects /Off /CompressPages true /ConvertImagesToIndexed true /PassThroughJPEGImages true /CreateJobTicket false /DefaultRenderingIntent /Default /DetectBlends true /DetectCurves 0.0000 /ColorConversionStrategy /LeaveColorUnchanged /DoThumbnails false /EmbedAllFonts true /EmbedOpenType false /ParseICCProfilesInComments true /EmbedJobOptions true /DSCReportingLevel 0 /EmitDSCWarnings false /EndPage -1 /ImageMemory 1048576 /LockDistillerParams true /MaxSubsetPct 100 /Optimize false /OPM 0 /ParseDSCComments false /ParseDSCCommentsForDocInfo false /PreserveCopyPage true /PreserveDICMYKValues true /PreserveEPSInfo false /PreserveFlatness true /PreserveHalftoneInfo true /PreserveOPIComments false /PreserveOverprintSettings true /StartPage 1 /SubsetFonts false /TransferFunctionInfo /Remove /UCRandBGInfo /Preserve /UsePrologue false /ColorSettingsFile () /AlwaysEmbed [ true /Arial-Black /Arial-BoldItalicMT /Arial-BoldMT /Arial-ItalicMT /ArialMT /ArialNarrow /ArialNarrow-Bold /ArialNarrow-BoldItalic /ArialNarrow-Italic /ArialUnicodeMS /BookAntiqua /BookAntiqua-Bold /BookAntiqua-BoldItalic /BookAntiqua-Italic /BookmanOldStyle /BookmanOldStyle-Bold /BookmanOldStyle-BoldItalic /BookmanOldStyle-Italic /BookshelfSymbolSeven /Century /CenturyGothic /CenturyGothic-Bold /CenturyGothic-BoldItalic /CenturyGothic-Italic /CenturySchoolbook /CenturySchoolbook-Bold /CenturySchoolbook-BoldItalic /CenturySchoolbook-Italic /ComicSansMS /ComicSansMS-Bold /CourierNewPS-BoldItalicMT /CourierNewPS-BoldMT /CourierNewPS-ItalicMT /CourierNewPSMT /EstrangeloEdessa /FranklinGothic-Medium /FranklinGothic-MediumItalic /Garamond /Garamond-Bold /Garamond-Italic /Gautami /Georgia /Georgia-Bold /Georgia-BoldItalic /Georgia-Italic /Haettenschweiler /Impact /Kartika /Latha /LetterGothicMT /LetterGothicMT-Bold /LetterGothicMT-BoldOblique /LetterGothicMT-Oblique /LucidaConsole /LucidaSans /LucidaSans-Demi /LucidaSans-DemiItalic /LucidaSans-Italic /LucidaSansUnicode /Mangal-Regular /MicrosoftSansSerif /MonotypeCorsiva /MSReferenceSansSerif /MSReferenceSpecialty /MVBoli /PalatinoLinotype-Bold /PalatinoLinotype-BoldItalic /PalatinoLinotype-Italic /PalatinoLinotype-Roman /Raavi /Shruti /Sylfaen /SymbolMT /Tahoma /Tahoma-Bold /TimesNewRomanMT-ExtraBold /TimesNewRomanPS-BoldItalicMT /TimesNewRomanPS-BoldMT /TimesNewRomanPS-ItalicMT /TimesNewRomanPSMT /Trebuchet-BoldItalic /TrebuchetMS /TrebuchetMS-Bold /TrebuchetMS-Italic /Tunga-Regular /Verdana /Verdana-Bold /Verdana-BoldItalic /Verdana-Italic /Vrinda /Webdings /Wingdings2 /Wingdings3 /Wingdings-Regular /ZWAdobeF ] /NeverEmbed [ true ] /AntiAliasColorImages false /CropColorImages true /ColorImageMinResolution 200 /ColorImageMinResolutionPolicy /OK /DownsampleColorImages true /ColorImageDownsampleType /Bicubic /ColorImageResolution 300 /ColorImageDepth -1 /ColorImageMinDownsampleDepth 1 /ColorImageDownsampleThreshold 1.50000 /EncodeColorImages true /ColorImageFilter /DCTEncode /AutoFilterColorImages false /ColorImageAutoFilterStrategy /JPEG /ColorACSImageDict << /QFactor 0.76 /HSamples [2 1 1 2] /VSamples [2 1 1 2] >> /ColorImageDict << /QFactor 0.76 /HSamples [2 1 1 2] /VSamples [2 1 1 2] >> /JPEG2000ColorACSImageDict << /TileWidth 256 /TileHeight 256 /Quality 15 >> /JPEG2000ColorImageDict << /TileWidth 256 /TileHeight 256 /Quality 15 >> /AntiAliasGrayImages false /CropGrayImages true /GrayImageMinResolution 200 /GrayImageMinResolutionPolicy /OK /DownsampleGrayImages true /GrayImageDownsampleType /Bicubic /GrayImageResolution 300 /GrayImageDepth -1 /GrayImageMinDownsampleDepth 2 /GrayImageDownsampleThreshold 1.50000 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages false /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict << /QFactor 0.76 /HSamples [2 1 1 2] /VSamples [2 1 1 2] >> /GrayImageDict << /QFactor 0.76 /HSamples [2 1 1 2] /VSamples [2 1 1 2] >> /JPEG2000GrayACSImageDict << /TileWidth 256 /TileHeight 256 /Quality 15 >> /JPEG2000GrayImageDict << /TileWidth 256 /TileHeight 256 /Quality 15 >> /AntiAliasMonoImages false /CropMonoImages true /MonoImageMinResolution 400 /MonoImageMinResolutionPolicy /OK /DownsampleMonoImages true /MonoImageDownsampleType /Bicubic /MonoImageResolution 600 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.50000 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict << /K -1 >> /AllowPSXObjects false /CheckCompliance [ /None ] /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile (None) /PDFXOutputConditionIdentifier () /PDFXOutputCondition () /PDFXRegistryName () /PDFXTrapped /False /CreateJDFFile false /Description << /CHS<FEFF4f7f75288fd94e9b8bbe5b9a521b5efa7684002000410064006f006200650020005000440046002065876863900275284e8e55464e1a65876863768467e5770b548c62535370300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c676562535f00521b5efa768400200050004400460020658768633002> /CHT <FEFF4f7f752890194e9b8a2d7f6e5efa7acb7684002000410064006f006200650020005000440046002065874ef69069752865bc666e901a554652d965874ef6768467e5770b548c52175370300260a853ef4ee54f7f75280020004100630072006f0062006100740020548c002000410064006f00620065002000520065006100640065007200200035002e003000204ee553ca66f49ad87248672c4f86958b555f5df25efa7acb76840020005000440046002065874ef63002> /DAN <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> /DEU <FEFF00560065007200770065006e00640065006e0020005300690065002000640069006500730065002000450069006e007300740065006c006c0075006e00670065006e0020007a0075006d002000450072007300740065006c006c0065006e00200076006f006e002000410064006f006200650020005000440046002d0044006f006b0075006d0065006e00740065006e002c00200075006d002000650069006e00650020007a0075007600650072006c00e40073007300690067006500200041006e007a006500690067006500200075006e00640020004100750073006700610062006500200076006f006e00200047006500730063006800e40066007400730064006f006b0075006d0065006e00740065006e0020007a0075002000650072007a00690065006c0065006e002e00200044006900650020005000440046002d0044006f006b0075006d0065006e007400650020006b00f6006e006e0065006e0020006d006900740020004100630072006f00620061007400200075006e0064002000520065006100640065007200200035002e003000200075006e00640020006800f600680065007200200067006500f600660066006e00650074002000770065007200640065006e002e> /ESP <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> /FRA <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> /ITA (Utilizzare queste impostazioni per creare documenti Adobe PDF adatti per visualizzare e stampare documenti aziendali in modo affidabile. I documenti PDF creati possono essere aperti con Acrobat e Adobe Reader 5.0 e versioni successive.) /JPN <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> /KOR <FEFFc7740020c124c815c7440020c0acc6a9d558c5ec0020be44c988b2c8c2a40020bb38c11cb97c0020c548c815c801c73cb85c0020bcf4ace00020c778c1c4d558b2940020b3700020ac00c7a50020c801d569d55c002000410064006f0062006500200050004400460020bb38c11cb97c0020c791c131d569b2c8b2e4002e0020c774b807ac8c0020c791c131b41c00200050004400460020bb38c11cb2940020004100630072006f0062006100740020bc0f002000410064006f00620065002000520065006100640065007200200035002e00300020c774c0c1c5d0c11c0020c5f40020c2180020c788c2b5b2c8b2e4002e> /NLD (Gebruik deze instellingen om Adobe PDF-documenten te maken waarmee zakelijke documenten betrouwbaar kunnen worden weergegeven en afgedrukt. De gemaakte PDF-documenten kunnen worden geopend met Acrobat en Adobe Reader 5.0 en hoger.) /NOR <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> /PTB <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> /SUO <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>/SVE <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> /ENU (Use these settings to create PDFs that match the "Required" settings for PDF Specification 4.01) >> >> setdistillerparams << /HWResolution [600 600] /PageSize [612.000 792.000] >> setpagedevice
Compartilhar