18-07-2013, 02:29 PM
Multi component fault diagnosis of rotational mechanical system based on decision tree and support vector machine
Multi component fault diagnosis.pdf (Size: 875 KB / Downloads: 56)
ABSTRACT
The shaft and bearing are the most critical components in rotating machinery. Majority of problems arise
from faulty bearings in turn affect the shaft. The vibration signals are widely used to determine the con-
dition of machine elements. The vibration signals are used to extract the features to identify the status of
a machine. This paper presents the use of c-SVC and nu-SVC models of support vector machine (SVM)
with four kernel functions for classification of faults using statistical features extracted from vibration
signals under good and faulty conditions of rotational mechanical system. Decision tree algorithm was
used to select the prominent features. These features were given as inputs for training and testing the
c-SVC and nu-SVC model of SVM and their fault classification accuracies were compared.
Introduction
In last two decades, due to the increase in production capabili-
ties of modern manufacturing systems, plants are expected to run
continuously for extended hours. As a result, unexpected down-
time due to machinery failure has become more costly than ever
before. Therefore, condition monitoring is gaining importance in
industry because of the need to increase machine availability and
health trending, to warn of impending failure. It is required to
detect, identify and then classify different kinds of faults that can
occur within a machine. Often several different kinds of sensors
are employed at different positions to sense a variety of possible
faults. Features are then computed to analyze the signals from all
these sensors to assess the health of the machine or its compo-
nents. Thus, the application of a condition monitoring-based main-
tenance policy can help to minimize unnecessary costs and delays
caused by unscheduled repairs.
Experimental procedure
In the present study, four SKF 6206 ball bearings were used. Out
of four bearings two were new bearings and free from defects. In
the other two ball bearings, defects were created using EDM in or-
der to keep the size of the defect under control. The size of inner
race defect is 0.552 mm wide and 0.782 mm deep and that of outer
race defect is 0.625 mm wide and 0.974 mm deep. Fig. 3 show the
Inner race fault and Outer race fault bearing. Before installing, each
bearing was properly lubricated with grease.
Support vector machine (SVM)
The next logical step is classification using a classifier. Support
vector machines is used as the classifier here. It is a new generation
learning system based on statistical learning theory. SVM belongs
to the class of supervised learning algorithms in which the learning
machine is given a set of features (or inputs) with the associated
labels (or output values). Each of these features can be looked upon
as a dimension of a hyper-plane. SVMs construct a hyper-plane
that separates the hyper-space into two classes (this can be ex-
tended to multi-class problems). While doing so, SVM algorithm
tries to achieve maximum separation between the classes (see
Fig. 6). Separating the classes with a large margin minimizes the
expected generalization error. By ‘minimum generalization error’,
we mean that when a new set of features (that is data points with
unknown class values) arrive for classification, the chance of mak-
ing an error in the prediction (of the class to which it belongs)
based on the learned classifier (hyper-plane) should be minimum.
Intuitively, such a classifier is one, which achieves maximum
separation-margin between the classes. The above process of
maximizing separation leads to two hyper-planes parallel to the
separating plane, on either side of it. These two can have one or
more points on them. The planes are known as ‘bounding planes’
and the distance between them is called as ‘margin’.
Conclusion
Fault diagnosis of shaft and bearings is one of the core research
areas in the field of condition monitoring of rotating machines.
Many researchers reported the fault diagnosis of either shaft or
bearing, but here both the shaft and the bearing have been consid-
ered. The statistical features for 12 different conditions were ex-
tracted from the vibration signals. Decision tree is used to select
the best features. The best features were classified using four dif-
ferent SVM kernel functions of two SVM model in support vector
machine. The RBF of c-SVC model gives the better classification
efficiency for four different speeds. From the above result one
can conclude that c-SVC model of SVM classifier with RBF kernel
function is a good candidate for fault diagnosis of rotational
mechanical system.