The reliability prognosis model, developed by the University of Wuppertal and an important automotive supplier, has been proved to be suitable in practice to answer reliability questions in the automotive industry. The model is based on failure data, which is recorded during the warranty period. To consider the load of automotive components, the model uses the mileage of the car as a substitute to the operating time, which is mostly not recorded. To adapt the prognosis on real behaviour the mileage distribution is used to estimate further failure-candidates. Due to this fact the mileage distribution is one fundamental element of the prognosis model. The empirical failure function Fk(s) is being fit by a theoretical function e.g. a Weibull distribution. The associated parameters a and ß are estimated by the method of least squares. Suitable starting values can be given by the Linear Regression. It is also suitable using generic algorithms to estimate the parameters as shown in Kazeminia. Reliability characteristics such as the failure probability or the hazard rate are used answering several questions regarding quality, economics, safety or reliability. Generally these questions depend on time e.g. costs of warranty period or end of life calculation. It is obvious that km-dependent reliability characteristics have to be converted into time-dependent.
Influence of the mileage distribution on reliability prognosis models
2008
6 Seiten, 4 Bilder, 3 Tabellen, 7 Quellen
Conference paper
English
Engineering Index Backfile | 1929
|Engineering Index Backfile | 1899
|Engineering Index Backfile | 1896
European Patent Office | 2022
|