The Failure Rate Modeling using Neural Network

The failure rate is one of the most important features in the theory of reliability, defined as the probability of failure in unit time of a device that is still working. This paper presents the possibility of the failure rate modeling using neural networks and comparison with classical models, i.e. modeling functions such as Exponential, Normal or Weibull. The paper is based on real examples shown statistical selection of optimal failure rate function and was proposal model and structure of neural networks as an improvement over the classical models.