3. Failure diagnosis and prognosis based on physical models

Failure diagnosis and prognosis based on physical models are based either on empirical models deduced from endurance tests carried out with a series of samples, or on knowledge models where parameter values are often not known with certainty, or on representation models. To optimize performance indicators and confidence in residual life prediction, it is essential to implement techniques for identifying and estimating the model parameters used. A large number of identification and estimation methods have been developed over the decades. This section briefly presents the principles of diagnostic and prognostic methods based on identification techniques, and the reader can refer to numerous specialized books and publications for further details