Prediction of remaining useful life of packing sets in plunger-type
high-pressure compressor based on PCA/SVD analysis and NN model
Abstract
A machine-learning-based prognostic strategy is developed in this paper
for predicting the remaining useful life (RUL) of high-pressure packing
in plunger-type hyper compressors. The proposed strategy applies
principal component analysis (PCA) to identify three most important
sensors out of 33 that seem relevant to the high-pressure packing.
Singular value decomposition (SVD) is then performed with respect to
chronological Hankel matrices reconstructed from one of these three
sensor data, leakage flow. Normalized correlation coefficient between
SVD eigenvalue vectors of chronological data is defined to come up with
a health state assessment measurement. In order to enhance the
prediction accuracy of RUL of the high-pressure packing, a
linear-regression and two-term power series regression algorithms are
both integrated into the NN (Neural Network) model. The effectiveness of
the method is examined using the averaged difference (over thirteen data
set) between the predicting and real failure events. The results showed
that a maximum prediction RUL error of the model is less than 15 days
and an averaged prediction RUL error is 7.23 days for 13 run-to-failure
events. Furthermore, a more recent test was performed using the on-line
data to examine the health states of four identical packing.