Definition of the Term “Prognostics”:



Prognostics is a “forecast of future performance and / or condition”.

Prognostic Engineering is therefore the discipline, body of knowledge, procedures, computation programs, and other resources, which supports Prognostics development and implementation.

Prognostic Technology is then the body of underlying science and mathematics, which supports Prognostic Engineering.

Prognostic Technique is a specific approach to Prognostics that may include development of specific features, feature vectors, models, statistical sets, etc. to be implemented in hardware / software and applied to a given component, subsystem, system, etc. These techniques may be broadly grouped in four classes:

  • Those based upon statistical treatment of data or information.

  • Those based upon a model of the application and the responses of the model to introduced faults or degradation.

  • Those based upon direct sensor measurement, feature and feature vector development, and analysis and prediction based upon these parameters.

  •  A combination of any or all of the preceding three.

Prognostic Accuracy or Confidence Level – the accuracy in terms of difference between the future forecast of performance or condition and the actual future value achieved expressed as  +/- an amount or as a percentage of the forecast. It may also be applied to the accuracy of the predicted time to failure, time to a given performance degradation point or percentage, remaining useful life, etc.

Prognostic Horizon – the maximum time or related parameter (such as number of missions, etc.) for which a given Prognostic Technique will achieve a set accuracy or confidence level.  For example, technique “A” may achieve a 90% prognostic accuracy with a horizon of 200 operating hours, or Prognostic Technique “B” may achieve a 75% prognostic accuracy with a prognostic horizon of 3 missions.

Prognostic Data – raw or processed sensor outputs and/or visual observations used directly or indirectly as a basis for prognostics.

Prognostic Information – the information derived from data and / or other sources used directly or indirectly as a basis for prognostics.

Prognostic Features – measures (or Figures of Merit) derived from data or information that are used by a given prognostic technique.

Prognostic Feature Vector – a specific combination of features that are used to generate a partial or total output for a specific prognostic technique.

Prognostic Feature Vector Set – those combined feature vectors that are used to generate a partial or total output for a specific prognostic technique.

Prognostic Model – A model of a component or system that allows the introduction of faults and degradation to develop or test prognostic techniques.

Prognostic Metrics – those measures of performance of a prognostic technique or system that characterize the performance and predictive reliability of that technique or system for a specific application. These metrics may include:

  • Demonstrated versus design prognostic accuracy / confidence level.

  • Demonstrated versus design prognostic horizon.

  • Demonstrated reliability of the prognostic system versus the system it monitors.

  • Applicability or robustness of the prognostic technique or system – how many other applications can the technique be applied to with commensurate accuracy, reliability and horizon attributes.

Previous Page