Prognostic will be an essential and integral component of the future advanced turbine engine operations and is the main focus of this paper. In the paper we have developed, implemented and tested a prognostic software system for estimating the remaining useful life (RUL) of failing components of a commercial grade turbofan engine. The software system was developed in two stages - prognostic stage for predicting engine component's remaining useful life (RUL) and decision-fusion stage for combining the different decisions predicted by the prognostic software. A graphical user interface (GUI) was also developed for testing and ease of operation. Two different prediction approaches were used in the development of the prognostic software - a deterministic and a probabilistic prediction approaches. A nonlinear exponential prediction method was used to develop the deterministic prognostic algorithm and the cloud computation theory was used to develop a probabilistic prediction algorithm. The data used were the estimated efficiency and flow of the engine derived by a diagnostic software from the sensor measurement of temperature and pressure of an engine component. The decision fusion stage was used to fuse all predicted RUL values by the prognostic algorithms to produce a final decision about the most accurate RUL prediction of an engine's failed component. Fuzzy logic inference system was used in the development of the decision-fusion algorithm. The overall performance of the developed prognostic software system was satisfactory and the test results produced accuracy above the ninety percent when the input efficiency and flow data were free of measurement noise. The prediction accuracy of the component's RUL was dropped to about eighty percent when the input data were corrupted with ten percent measurement noise. However, during the fusion stage some of the low RUL prediction accuracy due to noise was regained. In this paper, the development steps along with descriptions of the technical approaches used in the development of the prognostics and fusion algorithms are included. Test results are also provided and explained with adequate illustrations along with comparison of the major differences between the two prediction approaches used to develop the prognostics algorithms.
[1]
Abhinav Saxena,et al.
Damage propagation modeling for aircraft engine run-to-failure simulation
,
2008,
2008 International Conference on Prognostics and Health Management.
[2]
Jonathan S. Litt,et al.
User's Guide for the Commercial Modular Aero-Propulsion System Simulation (C-MAPSS)
,
2007
.
[3]
李德毅.
Knowledge Representation in KDD Based on Linguistic Atoms
,
1997
.
[4]
Deyi Li,et al.
Artificial Intelligence with Uncertainty
,
2004,
CIT.
[5]
Jonathan S. Litt,et al.
User's Guide for the Commercial Modular Aero-Propulsion System Simulation (C-MAPSS): Version 2
,
2012
.
[6]
Lotfi A. Zadeh,et al.
Fuzzy Sets
,
1996,
Inf. Control..
[7]
Hai Qiu,et al.
Modeling Propagation of Gas Path Damage
,
2007,
2007 IEEE Aerospace Conference.