Benchmark of methods applied to PHM08 Challenge Dataset



Aircraft Maintenance is one of the main causes of aircraft accidents and so improvements in this field should be performed. One way of improving aircraft maintenance is using Condition Based Monitoring (CBM) in aircraft with the goal of predicting when a component failure will occur. Using a Prognostics and Health Management (PHM) system it is possible to monitor the degradation behaviour of the equipment, proceeding to its replacement at the right time, i.e., before it can produce any damage by failing, and extending its usage lifetime to the maximum. This technical report presents three approaches that can be applied to the PHM08 Challenge Dataset [1] in order to predict the Remaining Usage Lifetime (RUL) of turbofan engines in aircraft. Based on a benchmark conceived by NASA [2] the three approaches are: Neural Network based methods, Extrapolation based methods and Similarity based methods. These categories don’t represent concrete methods or techniques but generalized approaches to be followed. For each one of them there are advantages and drawbacks, which are explained in this report. In order to improve their performance, some pre-processing should be done, including feature selection, data normalization and data partitioning.


Condition Based Monitoring (CBM), Prognostics and Health Management (PHM), Remaining Usage Lifetime (RUL), Aircraft systems


Prognostics and Health Management (PHM)

Related Project

H2020-REMAP – Real-time Condition-based Maintenance for Adaptive Aircraft Maintenance Planning

TechReport Number


Cited by

No citations found