Uncertainty of performance requirements for IVHM tools according to business targets

Abstract

Operators and maintainers are faced with the task of selecting which health monitoring tools are to be acquired or developed in order to increase the availability and reduce operational costs of a vehicle. Since these decisions will affect the strength of the business case, choices must be based on a cost benefit analysis. The methodology presented here takes advantage of the historical maintenance data available for legacy platforms to determine the performance requirements for diagnostic and prognostic tools to achieve a certain reduction in maintenance costs and time. The
effect of these tools on the maintenance process is studied using Event Tree Analysis, from which the equations are derived. However, many of the parameters included in the formulas are not constant and tend to vary randomly around a mean value (e.g.: shipping costs of parts, repair times), introducing uncertainties in the results. As a consequence the equations are modified to take into account the variance of all variables. Additionally, the reliability of the information generated using diagnostic and prognostic tools can be affected by multiple characteristics of the fault, which are never exactly the same, meaning the performance of these tools might not be constant either. To tackle this issue, formulas to determine the acceptable variance in the performance of a health monitoring tool are derived under the assumption that the variables considered follow Gaussian distributions. An example of the application of this methodology using synthetic data is included.