Investigating different structures for mapping sensor information of a complex mechanical system to its health status


One of the most important challenges of Prognosis and Health Monitoring (PHM) is how to map the sensor information to its health status. In this paper, different patterns of mapping are discussed. An appropriate pattern should be able to show the actual condition, degradation and remaining life of the machine for the maintenance engineer. Furthermore, in a complex system, several monitoring sensors are used, and a huge volume is needed for storage of all of this information. Therefore, reducing the dimensions of the system data is also considered. In the present study, a gas turbine engine is considered for case study. Various patterns for multi-sensors data mapping to health status have been evaluated due to their ability to estimate Remaining Useful Life (RUL) of the engine. The results of this study showed that the fusion of sensors through Principal Component Analysis (PCA) has a satisfying performance. The result is valuable for maintenance engineers to achieve accurate estimate from the health status with access to a one-dimensional signal requiring minimum memory. Another important point is that data fusion actually maintains all useful information.  The results of this article can be used to select the appropriate structure for health monitoring of complex mechanical systems.