Predictive Maintenance of Aircraft Engine using Empirical Mode Decomposition based Long Short Term Memory

In this modern era of Industry 4.0, efficiency and safety of industrial machines are a primary concern in trade sectors and prediction of Remaining Useful Life (RUL) of an engine helps to plan effectively maintenance operation. However, the Predictive Maintenance is to acquire reliable prognostics and estimation RUL in turbofan engines is a difficult task. To solve this problem, an Empirical Mode Decomposition based Long Short-Term Memory (EMD-LSTM) for prediction of RUL in aircraft engine utilizing C-MAPSS dataset. Initially, the most relevant sensor signals are selected by Correlation analysis and filtered using Moving Median Filter (MMF). After, the signals are normalized by Z-score technique to train the prediction model and signals are fed as input to proposed EMD-LSTM approach for prediction of RUL span of the engine. The evaluation results of the proposed method using performance metrics are Root Mean Square Error (RMSE), Score, Accuracy and Precision. The presented EMD-LSTM model achieved great accuracy of 99.3%, Precision of 98.9%, RMSE of 13.02 and Score of 282.0 which is greater than the previous models like Deep Convolutional Neural Network (DCNN)-LightGBM, LSTM, Graph CN (GCN), DCNN and Denoising Online Sequential Extreme Learning Machine (DOSELM).