A Data-Fusion Based Prognostic Method for Complex Degrading System

This paper develops a method where A composite indicator constructed by fusing sensors-data for modeling and life prediction is used to Characterize the degradation state of devices. During constructing the composite indicator based on multi-sensors information, the objective is to minimize the mean squared life prediction error when the composite constructed indicator is used in prognostics. Based on the constructed composite indicator, Wiener process with Bayesian updating for random-effect parameter is used to predict the life of devices. Finally, the C-MAPSS data set demonstrates that the proposed method make the improvement of the life prediction accuracy.