What is it about?
Predicting the number of failures within a time period is vitally important for asset management firms. Previous publications developed models, including the widely used NHPP (non-homogeneous Poisson process) model, do not perform well on many real-world time-between-failure datasets in terms of model performane measures such as the Akaike’s information criterion (AIC) or corrected Akaike’s information criterion (AICc). This paper developed a new model that outperforms nine other existing models on 11 out of 15 real-world datasets.
Featured Image
Why is it important?
Previous research on models for repairable systems seldom compared model performance on more than 5 real-world time-between-failure datasets. This is the first paper that compares model performance measures (eg, AIC, AICc, and BIC) of 10 models on 15 real-world datasets. The model developed in this paper provides more accurate fit and forecasting on the expected number of failures of a system. One may develop models based on historical data, based on which one may make more precise prediction on the number of failures of an asset in a given time period, and can therefore make a more precise fiscal plan on preparing capital for maintaining the asset.
Perspectives
Read the Original
This page is a summary of: A failure process model with the exponential smoothing of intensity functions, European Journal of Operational Research, November 2018, Elsevier,
DOI: 10.1016/j.ejor.2018.11.045.
You can read the full text:
Contributors
The following have contributed to this page