The Improvement of Remaining Useful Life Prediction for Aero-engines by Injecting Noise

The remaining useful life (RUL) prediction methods for rotating machines have been extensively studied in recent decades. In most of the existing RUL prediction methods, noise is always regarded as the disgusting disturbance and tried to remove from the observation data. In contrast, noise can be beneficial if properly used. This paper proposes a new RUL prediction method which is boosted by the injection of intentionally-added noise into a long short-term memory (LSTM) neural network. The intentionally-added noise satisfies the noisy-expectation maximization (EM) positivity condition, and boosts the RUL prediction results. Therefore, the proposed method is referred as a noise-boosted RUL prediction method. This paper is a preliminary attempt when predicting RUL for aero-engines by using intentionally-added noise. The popular C-MAPSS lifetime dataset for aero-engines is used to validate the effectiveness of the new method. The prediction results show the improvement of prediction accuracy by the injection of intentionally-added noise.