Non Coding of Big Dataset and the use of Neural Network Regression Artificial Intelligence Model in Azure for Predicting the Remaining Useful Life (RUL) of Bearing

摘要

In this paper a non-coding method for analyzing big data without the use of Hadoop, Hive, Pig, etc. was demonstrated with the use of Neural Network regression (NNR) for predicting the remaining useful life (RUL) of a bearing. Using these nonlinear nonparametric approach to predict the RUL of a bearing is intuitively appealing. The applications of NNR model have surged over the years that it has now been recognized as a major forecasting technique in a forecaster's toolbox hence, its application here on the NASA FEMTO BEARING DATASET. The motivation for this paper is to check if without using the normal big dataset, structuring the NNR models would add value in its applications. Although comparing this results is worth doing, the ultimate decision remains with NASA FEMTO as we are checking our result against their obtained experimental value which is in their archive. In the end, for the time series concerned, we clearly showed that NNR models do indeed add value in the forecasting process and we hope that our predicted RUL for the dataset of NASA FEMTO provided which they want a predicted result is accepted within a close range.