Techniques for estimating the parameters of the 2-parameter Weibull distribution from data obtained from uncensored tests are compared. This will allow the most convenient method to be chosen by considering the data's characteristics and the level of algorithm complexity. It is shown that common techniques, such as least squares regression and maximum likelihood, may give rise to very significant errors in terms of the bias of the estimated Weibull parameters. The estimator suggested by Jacquelin for α and the correction factor suggested by Ross for β usually give errors of < 5 and < 1%, respectively and are to be recommended, because they are straightforward to implement. Techniques are available which will eliminate virtually all bias in the estimation of the parameters, but these may be at the expense of considerable complexity in implementation. © 1997 IEEE.
In search of convenient techniques for reducing bias in the estimation of Weibull parameters for uncensored tests / Montanari G.C.; Mazzanti G.; Cacciari M.; Fothergill J.C.. - In: IEEE TRANSACTIONS ON DIELECTRICS AND ELECTRICAL INSULATION. - ISSN 1070-9878. - STAMPA. - 4:3(1997), pp. 306-313. [10.1109/94.598287]
In search of convenient techniques for reducing bias in the estimation of Weibull parameters for uncensored tests
Montanari G. C.;Mazzanti G.;
1997
Abstract
Techniques for estimating the parameters of the 2-parameter Weibull distribution from data obtained from uncensored tests are compared. This will allow the most convenient method to be chosen by considering the data's characteristics and the level of algorithm complexity. It is shown that common techniques, such as least squares regression and maximum likelihood, may give rise to very significant errors in terms of the bias of the estimated Weibull parameters. The estimator suggested by Jacquelin for α and the correction factor suggested by Ross for β usually give errors of < 5 and < 1%, respectively and are to be recommended, because they are straightforward to implement. Techniques are available which will eliminate virtually all bias in the estimation of the parameters, but these may be at the expense of considerable complexity in implementation. © 1997 IEEE.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
