Journal of The Japanese Society for Quality Control
Online ISSN : 2432-1044
Print ISSN : 0386-8230
Search
Advanced search
OR
Browse
Search
Volume 22, Issue 1
Displaying 1-13 of 13 articles from this issue
Show all abstracts
    • |<
    • <
    • 1
    • >
    • >|
Message
Technical Note
  • On a Comparison between Sudden Death Life Testing and Type II Number Fixed Life Testing －The Precisions and the Testing using the Maximum Likelihood Estimators－
    Kazuyuki SUZUKI, Kumi OHTSUKA, Masumi ASHITATE
    Article type: Technical Note
    1992 Volume 22 Issue 1 Pages 5-12
    Published: January 15, 1992
    Released on J-STAGE: February 15, 2019
    DOIhttps://doi.org/10.20684/quality.22.1_5
    JOURNAL RESTRICTED ACCESS
    Show abstractHide abstract
    "Sudden Death" life testing is a technique of reducing testing time[3]. It consists of grouping test specimens into 2 or more sets and then looking upon each set as an assembly of specimens in series and running all the specimens in a set simultaneously and waiting for the first specimen in each set to fail. The precision of estimators using the above testing is not known until now. This paper considers both the precisions of the estimators and the testing times based on the sudden death life testing and type II fixed number life testing. The comparisons between two life testing methods are done using the maximum likelihood estimators of Weibull distribution .The following results are obtained ; if there is no restriction on the number of the sets or the number of specimens in a set, type II fixed number life testing is superior to sudden death life testing. If there is the restriction, sudden death life testing becomes important and how to group test specimens into sets and when the test should be terminated are examined.
    View full abstract
    Download PDF (670K)
  • Maximum Entropy Distribution Based on Incomplete Lifetime Data
    Tsunenori ISHIOKA, Yasuo NONAKA
    Article type: Technical Note
    1992 Volume 22 Issue 1 Pages 13-25
    Published: January 15, 1992
    Released on J-STAGE: February 15, 2019
    DOIhttps://doi.org/10.20684/quality.22.1_13
    JOURNAL RESTRICTED ACCESS
    Show abstractHide abstract
    Siddall let the theoretical moments μ_k derived from the pdf (probability density function ) based on maximum entropy principle be equal to the sample moments M_k ; he proposed the estimation method of the pdf by minimizing the total sum of the squared errors, where the error is 1-μ_k/M_k. This paper presents a technique to obtain the sample moments for randomly censored data ; by using its sample moments, we can utilize Siddall's method for randomly censored data. That means we can estimate the pdf for the incomplete data when there is no previous information for the model nominee. We illustrate the property with numerical examples. Moreover we indicate how to determine the optimum number of the moments.
    View full abstract
    Download PDF (1058K)
Features
Tutorial
    • |<
    • <
    • 1
    • >
    • >|
feedback
 Top