applied

sciences
Article
Condition Monitoring of Industrial Equipment Based
on Multi-Variables State Estimate Technique
Dongteng Long 1, * , Heng Zheng 1 and Feng Hong 2
1 China Astronautics Standards Institute, Beijing 100071, China; [email protected]
2 State Key Laboratory of Alternate Electric Power System with Renewable Energy Sources, School of Control
and Computer Engineering, North China Electric Power University, Beijing 102206, China;
[email protected]
* Correspondence: [email protected]; Tel.: +86-10-8810-8251




Received: 15 July 2020; Accepted: 10 August 2020; Published: 14 August 2020


Abstract: Unexpected failures commonly occur in industrial equipment, and condition monitoring
could significantly improve the efficiency of maintenance and failure of early alarm. A condition
monitoring method using multi-variables state estimate technique (MSET) is proposed, and an
improved multi-variables memory matrix construction method is employed, furthermore, an analysis
of comprehensive similarity index that considering variable weights is accomplished, and incipient
failure alarm thresholds are determined, which lead to effective early detection of failure. The method
proposed in this paper is validated using actual data for blower fan in a thermal power plant, and the
simulation and comparison results are discussed. The verification results reveal that the proposed
method is effective for failure monitoring modeling and achieve a superior accuracy, and incipient
failure could be accurately detected.

Keywords: condition monitoring; vibration; multi-variables state estimate technique

1. Introduction
Industrial equipment is usually complicated and features multiple failure modes. As a large-scale
rotate machine, once failure occurs, a long-time repairing, even shutdown, is required, resulting
in a significant economic loss. On the other hand, industrial field characterized harsh operational
environment, equipment would experience a high failure rate, therefore, accurate condition monitoring
for industrial equipment is necessary, and plays a pivotal role in condition-based maintenance, which
would be more beneficial than corrective and preventive maintenance [1,2].
Tavner [3] concluded the general condition monitoring for industrial equipment, including
temperature, chemical emission, vibration, electrical current, and electrical discharging. Furthermore,
the paper illustrated the statistics distribution of failure modes in industrial equipment in the past
decades, according to the research, rotation related failure occupies nearly a half (44%), higher than
stator related (35%) and bearing related failure (21%), and according to previous work, unexpected
vibration often occurs before serious failure events [4,5].
As a typical condition monitoring parameter, vibration is chosen for example and investigated
in our paper. The aim of condition monitoring is inferring the condition of equipment rapidly, and
giving a clear indication for incipient failure modes using minimum measurements. In industrial
fields, vibrations are directly measured, for example, vibrations data are recorded in supervisory
information system (SIS) in thermal power plants. In normal operating conditions, the vibrations of
different directions are controlled in a certain range. The unexpected increase of vibrations corresponds
to over-load, poor lubrication, bearing loosening, rotor unbalance, or shaft bending [6]. Therefore,

Appl. Sci. 2020, 10, 5637; doi:10.3390/app10165637 www.mdpi.com/journal/applsci

Appl. Sci. 2020, 10, 5637 2 of 17

effective condition monitoring could evaluate equipment effectively, and the failures would be detected
in early stage.
The state-of-the-art for rotate equipment vibration condition monitoring is concluded into directly
signal monitoring and model-based monitoring. In the aspect signal analysis, time domain analysis
and frequency domain analysis are effective methods for condition monitoring of rotate component
vibration [6–9]. However, for large rotating equipment, such as blower fan, there are some difficult
to realize signal condition monitoring: (1) blower fan working at variable power load and frequency.
Working condition of rotating equipment changes frequently, which leads to frequency band width of
vibration changed, resulting in poor accuracy of failure location. (2) The industrial environment is
complicated, the vibration sources of each component are multiple and seriously coupling, therefore, the
vibration analysis of components is difficult to reflect characteristic of blower fan. With the development
and application of information technology in industrial fields, operating data, including vibration
data, of equipment is recorded automatically in detail, which makes it possible to achieve condition
monitoring using data-driven analysis methods. Expert system is an effective implementation for
data-driven artificial intelligent method, an expert system for industrial machine vibration analysis is
designed for condition monitoring, and the knowledge set was commonly constructed using operating
data and laboratory and industry validation [10,11]. Neural network (NN) is an available method for
nonlinear vibration; the model between recorded parameters and vibration state is established; the
knowledge in historical data is acquired in the course of model training. The derivative neural network,
such as recurrent neural network and convolutional neural network are also applied in vibration
condition monitoring [12–16]. However, the weaknesses of NN are also obvious, the training process
is indispensable and it is always time-consuming. Furthermore, the adequate actual failure events data
are difficult to obtain in industrial fields, the labeled data are the key component for improving the
vibration monitoring accuracy in model training.
A concise and effective method is establishing an exact nonlinear normal state envelope space
using normal vibration data, normal operating condition is completely included in the normal state
envelope space, and the distance between current condition and boundary of success envelop space
is used to quantify the condition monitoring. Cempel [17] proposed vibration condition monitoring
method applying symptom observation matrix; discrete observation symptom vectors were obtained
by analyzing historical data, and the vectors create symptom observation matrix, using next the
singular value decomposition for the given symptom observation matrix, fault symptoms could be
extracted. MSET(multi-variables state estimate technique) was first proposed by Gross [18] and applied
in sensor and gearbox monitoring [19,20]. MSET is proved as an alternative method for constructing
normal state envelope space. In general, our paper focused on proposing an improved and universal
condition monitoring method and judging criteria, although the vibration condition monitoring for a
specific equipment is applied and discussed in our paper, obviously, the condition monitoring for the
other measurement parameters for different industrial equipment, such as temperature, current, etc.,
could be realized by reusing the proposed method.
In this paper, the normal state envelope space construction method for industrial equipment
condition monitoring is proposed based on MSET, and the variables selecting for normal state envelope
space are discussed. For a further feasibility discussion, a specific application in blower fan vibration
condition monitoring is proposed. The weighted-similarity between the observation vibration vector
and estimated vibration vector is investigated. While the potential failure occurred, there will be
distinguished difference in similarity distribution, which allows incipient failure to be detected at
early stage.
The rest of paper is organized as follows: In Section 2, introduction of a typical industrial blower
fan, which is applied for discussing in our research, is provided, and the parameters for blower fan
are exhibited. In Section 3, the structure for vibration condition monitoring model using MSET is
discussed. In Section 4, the vibration condition monitoring simulation is presented, including variables
selection, model construction, similarity calculation, and alarm threshold determination. In Section 5,
 Appl. Sci. 2020, 10, x FOR PEER REVIEW 3 of 17
Appl. Sci. 2020, 10, 5637 3 of 17
data for blower fan have been used for revealing the effectiveness of proposed method. Finally,
Section 6 concludes the work.
the actual industrial data are applied for validation, and actual failure data for blower fan have been
used for revealingofthe
2. Introduction effectiveness
Research Objectof proposed method. Finally, Section 6 concludes the work.
A typical of
2. Introduction industrial
Researchblower
Object fan is discussed in our research. It is manufactured by TLT
Cooperation, equipped in a thermal power plant in northern China, used for conveying air into
A typical industrial blower fan is discussed in our research. It is manufactured by TLT Cooperation,
furnace of boiler for combustion. The blades of blower fan are integrated with driving motor by
equipped in a thermal power plant in northern China, used for conveying air into furnace of boiler
shaft, and the shaft is sustained by bearings. While the motor is running, the blades rotate and air is
for combustion. The blades of blower fan are integrated with driving motor by shaft, and the shaft is
injected into the furnace of the boiler. The structure for blower fan is shown in Figure 1, and the
sustained by bearings. While the motor is running, the blades rotate and air is injected into the furnace
actual layout and location of measuring points for investigated blower fan are also exhibited.
of the boiler. The structure for blower fan is shown in Figure 1, and the actual layout and location of
measuring points for investigated blower fan are also exhibited.

Figure 1. Layout of the blower fan.

Figure 1. Layout of the blower fan.
The Supervisory Information System (SIS) for the power plant records the parameters for blower
fan inThe
a 20-s sampling solution.
Supervisory Information In each
System sample,
(SIS) related
for theparameters
power plant include:
recordstime
the stamp, current
parameters for
air
blower fan in a 20-s sampling solution. In each sample, related parameters include: time motor,
volume, load of unit, vibration of bearings, temperature of bearings, current of driving stamp,
operation
current airfrequency
volume, of driving
load motor,
of unit, and power
vibration of driving
of bearings, motor. The
temperature parameters
of bearings, are shown
current in
of driving
Table 1. Particularly, the temperature values, including temperature of bearings
motor, operation frequency of driving motor, and power of driving motor. The parameters are and end-side bearings,
are measured
shown in Tableusing1.platinum resistance
Particularly, temperature values,
the temperature detector.including
Before variables selection
temperature of for condition
bearings and
monitoring, the 2 out
end-side bearings, areofmeasured
3 actions using
are accomplished, the absolute
platinum resistance errors between
temperature detector.each
Beforeof variables
the two
temperature
selection forvalues are compared,
condition monitoring,and the
the average
2 out ofof3temperatures whose absolutethe
actions are accomplished, error is smallest
absolute is
errors
regard
betweenas the
each actual
of thevalue
two of measurement,
temperature which
values are would improve
compared, the average
and the reliability
ofof the measurement.
temperatures whose
Except the parameters, the actual vibration alarms were also recorded in Supervisory
absolute error is smallest is regard as the actual value of measurement, which would improve the Information
System; while
reliability the measurement.
of the vibration values exceed the thresholds setting ahead, the time stamp is recorded. The
7-day Except
normal condition data are available
the parameters, the actual for vibration
modeling. alarms
Meanwhile,
wereactual
also failure alarms
recorded in are acquired
Supervisory
inInformation
our investigated blower fan, which is used for incipient failure detecting validation
System; while the vibration values exceed the thresholds setting ahead, the time in our papers.
stamp
is recorded. The 7-day normal condition data are available for modeling. Meanwhile, actual failure
Table 1. Parameter of blower fan.
alarms are acquired in our investigated blower fan, which is used for incipient failure detecting
validation inParameter
our papers. Remark Quantity Unit
Load Current Load of the Unit 1 %
Table 1. Parameter of blower fan.
Vol Current Air Volume of Unit 1 %
parameter I CurrentRemark
of Driving Motor 1
Quantity A Unit
P Power of Driving Motor 1 kW
Load f Current
Operation Load ofofthe
Frequency Unit Motor
Driving 1 1 Hz %
Vol t1 Current Air Volume of
Temperature of BearingsUnit 3 1 ◦C %
◦C
I t2 Current ofofDriving
Temperature End-SideMotor
Bearings 3 1 A
P v x Vibration
Power of Horizontal
of Driving Motor 1 1 mm/s kW
vy Vibration of Vertical 1 mm/s
f Operation Frequency of Driving Motor 1 Hz
p1 Inlet Air Pressure 1 Pa
t1 p2 Temperature of Bearings
Furnace Pressure 1 3 Pa ℃
t2 Temperature of End-Side Bearings 3 ℃
vx Vibration of Horizontal 1 mm/s
Appl. Sci. 2020, 10, 5637 4 of 17

3. Multi-Variable State Estimate Technique (MSET)

By applying MSET, considerable historical data for industrial equipment are used for constructing
the normal state envelope space, which is noted as memory matrix D. While the industrial equipment
operating, the measurement parameters are observed real-time, and the related n parameters construct
a vector noted as X obs . In the time sample j, the observation vector is written as,

X(tj ) = [x1 (tj ), x2 (tj ), . . . , xn (tj )]T (1)

It should be pointed out that for different equipment condition monitoring, only the parameters
and thresholds need to be determined if reusing the proposed method, and the complete preprocess of
the case study in our paper is posed in Section 4.
D is an m × n dimension matrix, which means m normal samples are strictly selected and contained
in D at last. Comprehensive operating conditions, including high load, low load, change load, and
other operating conditions should be contained. Memory matrix D could be written as,
 
 x1 (t1 ) x1 (t2 ) ··· x1 (tm ) 
i  x2 (t1 ) x1 (t2 ) ··· x2 (tm )
 
h 
D= X ( t1 ) X ( t2 ) ··· X(tm ) =  (2)

.. .. .. .. 

 . . . . 

xn (t1 ) xn (t2 ) ··· xn (tm )
 
n×m

The vectors included in the memory matrix D construct a nonlinear normal state envelope
space. After the memory matrix D constructed, dynamic state estimation could be accomplished by
applying it.
The construction of estimation vector for the same time sample X est is constructed by the vectors
in memory matrix D, the different of X est and X obs is applied for condition monitoring. X est could be
written as,
T
Xest = D · W = D · [ w1 w2 · · · wm ]
(3)
= w1 X(t1 ) + w2 X(t2 ) + · · · + wm X(tm )
The weight matrix W stands for similarity between observation vector and components in matrix
D. The optimized weight matrix W could be obtained by minimizing the error between X est and X obs ,
which is,
e = Xest − Xobs (4)

According to the least square algorithm, for minimizing kek2 let,

n
e2i = eT e = (Xobs − Xest )T · (Xobs − Xest )
P
S(w) =
i=i
n 

m
2 (5)
T P P 
= (Xobs − D · W ) · (Xobs − D · W ) = Xobs (i) − w j Dij 

i=1 j=1

where, Dij is the element for memory matrix in i line j column.

Solving the partial derivative Equation for wi in S(w), which is,
n m
∂S X X
= −2 (Xobs (i) − w j Dij )Dik = 0 (6)
∂wk
i=1 j=1

That is,
n
X n X
X m n X
X m
Xobs (i)Dik = w j Dij Dik = ( Dij Dik )w j (7)
i=1 i=1 j=1 i=1 j=1
Appl. Sci. 2020, 10, 5637 5 of 17

The Equation (7) could be expressed as matrix form as,

DT · D · W = DT · Xobs (8)

The weight matrix could be solved, that is,

−1
W = ( DT · D ) · (DT · Xobs ) (9)

In particular, if components in memory matrix D are linear correlative, the matrix DT · D would
be irreversible, which leads the (9) unsolvable. In order to solve the problem properly, there are
many alternative operator; Argonne laboratory has also developed a unique algorithm for nonlinear
calculation [15]. A nonlinear operator ⊗ is introduced, and the Equation (9) could be written as,

−1
W = (DT ⊗ D) · (DT ⊗ Xobs ) (10)

Concise nonlinear operator that calculates the Euclidean distance arithmetic is chosen in our
paper, that is,
v
t n
X
⊗ (X, Y) = (xi − yi )2 (11)
i=1

The operator we employed is easy to understand physically, and the similarity of two vectors
is represented by spatial distance. If the observation vector is similar with the vectors in memory
matrix D, the corresponding spatial distances would be small, and while there is a distinguishable
different between observation vector and memory matrix, the distance would become large. Finally,
the estimation vector could be written as,
−1
Xest = D · (DT ⊗ D) · (DT ⊗ Xobs ) (12)

The memory matrix D established a normal nonlinear normal state envelope space, while the input
observation vector X obs is located in the normal space and similar with a vector in D, the corresponding
vector X est obtained by Equation (12) is rather accurate. However, if an incipient failure occurs, the
characteristic of X obs would change, and the observation vector X obs would exceed the boundary of
envelope space, as a result, X obs could not be represented by using the vectors included in memory
matrix D, and the current observation vector X obs and the estimate vector X est would be very different.
In conclusion, the incipient failure could be detected by analyzing similarity between X obs and X est .

4. Modeling Description

4.1. Data Preprocess and Variables Selection

Data preprocess is an important stage in the data-driven modeling. Normally, missing values and
outliers, should be removed or specially dealt with. In our research, the outliers and missing values
are replaced by the average of the adjacent samples to guarantee the data continuity. On the other
hand, the dimensional difference of input and output variables is eliminated by data standardization
using Equation (13), which transforms different variables to same scale.

x−x
x0 = (13)
σ
Before constructing the memory matrix, the variables need to be selected. Vibration variables
are target vectors, and the relationship between the other variables is listed in Table 1 and target
vectors need to be discussed. Grey relational analysis is an effective statistical method quantifying
the relevance degrees, the “grey relational degree” of variables is calculated using consecutive data
Appl. Sci. 2020, 10, 5637 6 of 17

of variables, and the method is proved effective in many fields. Therefore, grey relational analysis is
applied in variables selection in our paper.
Consecutive 1000 samples are used for obtaining grey relational degree. Vibration vectors are set
as target sequences, and the rest 9 variables are set as comparing vectors. The grey relational degree
sequence for variable could be calculated as,

min min y(k) − xi (k) + ρ max max y(k) − xi (k)

i k i k
ξi ( k ) = (14)
y(k) − xi (k) + ρ max max y(k) − xi (k)
i k

where, ρ is discrimination coefficient, and is set as 0.6 in our research.

The average grey relational degrees for compared vector are calculated as,
n
1X
ri = ξi ( k ) (15)
n
k =1

The results of grey relational analysis between the horizontal vibration, vertical vibration, and the
compared vector are shown in Tables 2 and 3, respectively. According to the grey relational analysis
results, except the vibration of horizontal and vertical itself, the temperature of end-side bearings,
current air volume of unit, operation frequency of driving motor and current of driving motor (gray
relational degree > 0.6), are chosen for vibration condition monitoring.

Table 2. Results of grey relational analysis for horizontal vibration.

Parameter Remark Gray Relational Degree

t2 Temperature of End-Side Bearings 0.913
Vol Current Air Volume of Unit 0.844
f Operation Frequency of Driving Motor 0.819
I Current of Driving Motor 0.789
Load Current Load of The Unit 0.595
p2 Furnace Pressure 0.549
P Power of Driving Motor 0.417
t1 Temperature of Bearings 0.355
p1 Inlet Air Pressure 0.191

Table 3. Results of grey relational analysis for vertical vibration.

Parameter Remark Gray Relational Degree

f Operation Frequency of Driving Motor 0.847
I Current of Driving Motor 0.739
t2 Temperature of End-Side Bearings 0.733
P Power of Driving Motor 0.719
Vol Current Air Volume of Unit 0.627
p1 Inlet Air Pressure 0.514
t1 Temperature of Bearings 0.419
Load Current Load of The Unit 0.261
p2 Furnace Pressure 0.167

4.2. Vector Selection for Memory Matrix

While the MSET was first proposed, the D is constructed by the extreme vectors for each parameter,
and the boarder of normal state envelope space is defined by the extreme vectors. In latest works,
vectors of normal condition are added to memory matrix according to constant interval and access
criteria, indeed, the method improved the memory matrix D and enhanced the ability for describing
normal conditions. However, special normal condition vector might be omitted in sample process,
Appl.
Appl.Sci. 2020,
Sci. 10,10,
2020, x FOR
5637 PEER REVIEW 7 of 1717
7 of

process, which would seriously affect the prediction performance. Thus, a novel normal vectors
which would
selection method seriously affect the prediction
using self-adaption performance.
access criteria is proposed. Thus,
Aftera standardization,
novel normal vectors selection
the deviation
method using self-adaption access criteria is proposed. After standardization,
degree is reflected by the distance between normal vector with origin point in hyperplane. Because the deviation degree is
reflected
the vectorsby forthe distance
memory between
matrix normalin
is selected vector
normal with origin point
condition, in hyperplane.
the vectors Because
are relatively the vectors
intensive in
for memory matrix is selected in normal condition, the vectors are relatively intensive
the area that is close to the origin point in hyperplane, and in the area relatively far from the origin in the area that
is close
point, to theare
vectors origin point
sparse andinspecial
hyperplane,
normal and in the area
condition relatively
often appears. farThe
from the origin
special normal point, vectors
condition
are sparse
vector mightand notspecial normal
be selected bycondition
using the often appears.
constant Theand
interval special
accessnormal condition
criteria sampling vector might not
method.
be selected by using the constant interval and access criteria sampling method.
In order to select representative normal vector and retain some special normal vector, the
sampleInaccess
order criteria
to selectchange
representative normal vector
with probability densityand retain some
function, whilespecial normal
the value vector, the
of function sample
is small,
accessmeans
which criteriathechange
vectorwith probability
is located density
in sparse area function, while the
in hyperplane, the access
value of function
criteria is small, loose
is relatively which
means the vector is located in sparse area in hyperplane, the access
for introducing more special normal vectors. As shown in Figure 2, the maximum of probabilitycriteria is relatively loose for
introducing
density function more special
is 0.62. Thenormal
number vectors. As shown
of vector in memoryin Figure
matrix2, the
D ismaximum
set as 200.ofThe
probability
procedure density
for
function is 0.62. The number of vector in memory
constructing memory matrix D is formulated in Figure 3 matrix D is set as 200. The procedure for constructing
memory matrix D is formulated in Figure 3.

Figure 2. Fitting for probability density function.

Figure 2. Fitting for probability density function.
Appl. Sci. 2020, 10, 5637 8 of 17
Appl. Sci. 2020, 10, x FOR PEER REVIEW 8 of 17

Figure3.3.Procedure
Figure Procedurefor
forconstructing
constructingmemory
memorymatrix
matrixD.
D.

Where,
Where,
k
k ε= (16)
ϕ
ε = j)( (16)
ϕ ( j)
k is an empirical parameter, and is set as 0.0015 in our simulation.
k is an empirical parameter, and is set as 0.0015 in our simulation.
4.3. Improved Similarity Calculation
4.3. Improved Similarity
The residual Calculation
of vibrations component in X obs and X est , including the average value for residual
sequence,
The are usually
residual of used for judging
vibrations whether
component inthe
Xobsincipient
and Xestfailures occur
, including in average
the previousvalue
works.for
However,
residual
the
sequence, are usually used for judging whether the incipient failures occur in previouscurrent
information contained in the other variables, including temperature of end-side bearings, works.
air volume of unit, operation frequency of driving motor, and current of driving
However, the information contained in the other variables, including temperature of end-side motor, would be
neglected, but some vibrations failure might be first reflected by indirect parameters.
bearings, current air volume of unit, operation frequency of driving motor, and current of driving Therefore, except
the analysis
motor, the residual
would of vibrations
be neglected, directly,
but some a comprehensive
vibrations similarity
failure might calculation
be first based
reflected byon other
indirect
components
parameters. inTherefore,
observation vectorthe
except is proposed,
analysis theandresidual
the difference between directly,
of vibrations components is considered.
a comprehensive
The similarity could be formulated as,
similarity calculation based on other components in observation vector is proposed, and the
difference between components is considered. 1
S(X
The similarity could , Xest ) = as,s
beobsformulated (17)
n
Ri [Xobs,i (t(1k)) − Xest,i (t(k))]2
P
1+
S ( X obs , X est ) = i=1
n
(17)
1+  R [X
i =1
i obs ,i (t (k )) − X est ,i (t (k ))]2

where Xobs,i and Xest,i are the component i in observation vector and estimation vector,
respectively, and Ri is corresponding weight coefficient for i component in vectors, which could be
obtained according to the result of grey relational analysis in Table 4.
Appl. Sci. 2020, 10, 5637 9 of 17

where X obs,i and Xest,i are the component i in observation vector and estimation vector, respectively, and
Ri is corresponding weight coefficient for i component in vectors, which could be obtained according
to the result of grey relational analysis in Table 4.
ri
Ri = n
(18)
P
rj
j

where n is the number of components that are selected in to memory matrix.

Table 4. Comparison of RMSE and MAPE for condition I.

Situation RMSE (mm/s) MAPE (%)

Proposed method 0.0045 0.45
Horizontal vibration x1
Traditional method 0.0265 2.12
Proposed method 0.0017 0.11
Vertical vibration x2
Traditional method 0.0177 2.79

4.4. Incipient Failure Threshold Determination

4.4.1. Residual Threshold

Residual threshold is determined for residuals between X obs and X est , and it is a direct vibration
monitoring parameter, information contained in vibration component of observation vector is applied.
In order to reflect the long-term trend of vibration monitoring and eliminate the influence of
measurement errors, residuals sequence is discussed and sliding window is applied. The threshold of
average value Taver is defined as,
Taver = |c1 · AVER| (19)

where AVER is the average residual in a sliding window. The current residual is compared with Taver .
The coefficient c1 is determined by operation experience and validation, in this paper, c1 = 2.5.

4.4.2. Similarity Threshold

As a supplement for direct vibration monitoring, information contained in the other component
in X obs and X est are applied to calculate a comprehensive similarity, which is significant for detecting
faults reflecting in indirect variables. Furthermore, as shown in Figure 1 and Table 1, there are
no redundancies for vibration measurement, therefore, similarity judgement is an effective way in
situations where vibrating measurement fails, and it is significant for improving the accuracy of
condition monitoring. The similarity threshold could be formulated as,

S(Xobs , Xest )min

Ts = (20)
c2

where S(Xobs , Xest )min is the minimum of similarity in a sliding window, and coefficient c2 is set as 1.1
in this paper.

5. Validation and Discussion

5.1. Validation of Blower Fan MSET Modeling

The sample solution of data is 20 seconds, and the variables we discussed in Section 4 is available,
which are vibration of horizontal x1 , vibration of vertical x2 , the temperature of end-side bearings x3 ,
current air volume of unit x4 , operation frequency of driving motor x5 , current of driving motor x6 ,
respectively. The typical distribution for normal data in 10/25/2015 is shown in Figure 4, the maximum
Appl. Sci. 2020, 10, 5637 10 of 17

of vibration
Appl. in xhorizontal
Sci. 2020, 10, and vertical
direction are 2.725 mm/s and 5.195 mm/s, respectively. In 10
FOR PEER REVIEW totally,
of 17
7 days normal samples were used for contrasting memory matrix D.

3
x 1 (mm/s)

2
1
10/25 00:00 10/25 06:00 10/25 12:00 10/25 18:00 10/26 00:00
10
x 2 (mm/s)

5
0
10/25 00:00 10/25 06:00 10/25 12:00 10/25 18:00 10/26 00:00
100
x3 (℃)

80
60
10/25 00:00 10/25 06:00 10/25 12:00 10/25 18:00 10/26 00:00
100
x4 (%)

90
80
10/25 00:00 10/25 06:00 10/25 12:00 10/25 18:00 10/26 00:00
100
x5 (Hz)

90
80
10/25 00:00 10/25 06:00 10/25 12:00 10/25 18:00 10/26 00:00
300
x6 (A)

200
100
10/25 00:00 10/25 06:00 10/25 12:00 10/25 18:00 10/26 00:00

Figure 4. Layout of variables data.

Figure 4. Layout
After using the method mentioned of variables
in Section data.matrix D is constructed, and 211
4, memory
vectors areusing
After concluded, therein, mentioned
the method 11 vectors are
in selected bymemory
Section.4, extreme condition
matrix D ofis 6constructed,
variables andand
remove
211
duplication, and the rest 200 vectors are obtained by method mentioned in Figure 3.
vectors are concluded, therein, 11 vectors are selected by extreme condition of 6 variables and Consecutive
normal samples,
remove including
duplication, and thesteady
rest and
200 dynamic working
vectors are condition
obtained data, are
by method used for validating
mentioned in Figure the
3.
accuracy of MEST
Consecutive normalmodel.
samples, including steady and dynamic working condition data, are used for
validating the accuracy of MEST model.
5.1.1. Steady Working Condition
5.1.1. In
Steady
orderWorking Condition
to validate the effectiveness of the proposed method, traditional constant interval and
access
In criteria
order tomethod
validate[18]
theare applied forofcomparison,
effectiveness the proposed which is shown
method, as traditional
traditional constantmethod
intervalinand
the
figures, and the size of memory matrix is also set as 200. It should be pointed out
access criteria method [18] are applied for comparison, which is shown as traditional method in the that in traditional
method,and
figures, the access
the sizecriteria ε is a constant,
of memory matrix isand
alsoextreme condition
set as 200. vectors
It should are also out
be pointed included
that inintraditional
traditional
method, the access criteria ε is a constant, and extreme condition vectors are also included
method. The comparisons for proposed method and traditional method of horizontal and vertical
in
vibration are shown in Figures 5 and 6, respectively, and the residual and similarity
traditional method. The comparisons for proposed method and traditional method of horizontal and are compared.
Thevibration
vertical dataset for
arevalidation
shown ininFigures
Figures 55 and
and 66,isrespectively,
relative steady andthe
and chosen fromand
residual a normal working
similarity are
condition
compared. for blower fan, named normal condition I.
The dataset for validation in Figures 5 and 6 is relative steady and chosen from a normal
working condition for blower fan, named normal condition I.
 Appl. Sci. 2020, 10, x FOR PEER REVIEW 11 of 17
Appl. Sci. 2020, 10, 5637 11 of 17
Appl. Sci. 2020, 10, x FOR PEER REVIEW 11 of 17

Figure 5. Comparison for proposed method and traditional method for horizontal vibration in
condition
Figure
Figure 5.5. I.Comparison
Comparison for
for proposed
proposed method
method and
and traditional
traditional method
method for
for horizontal
horizontal vibration
vibration in
in
condition I.
condition I.

Proposed method Traditional method

1.4 1.4
Proposed method Traditional
observationmethod
observation
1.4
1.2 1.4
1.2
estimation estimation
observation observation
1.21 1.21 estimation
estimation
1
0.8 1
0.8

0.8
0.6 0.8
0.6
0 50 100 150 200 0 50 100 150 200
0.6 -3 0.6
10
15 0 50 100 150 200 0.1 0 50 100 150 200
-3
10
15
10 0.1
0.05
105
0.05
0
50
0
0
-5 -0.05
0 50 100 150 200 0 50 100 150 200
-5 -0.05
10 50 100 150 200 10 50 100 150 200

1 1
0.99
0.95
0.99
0.98
0.95
0.98
0.97 0.9
0 50 100 150 200 0 50 100 150 200
0.97 samples/[20s] 0.9 samples/[20/s]
0 50 100 150 200 0 50 100 150 200
samples/[20s]
Figure 6. Comparison for samples/[20/s]
proposed method and traditional method for vertical vibration in condition I.
Figure 6. Comparison for proposed method and traditional method for vertical vibration in
condition
As shown
Figure 6. I.Comparison
in Figures 5for
andproposed
6, the ranges forand
method horizontal andmethod
traditional verticalforvibrations are 0.589 in
vertical vibration mm/s
to 0.752 mm/s I.and 0.717 mm/s to 1.253 mm/s, respectively, which are normal vibration ranges. The
condition
As shown
vibrations in theindataset
Figures
are5 and 6, the
steady, andranges
in thatfor horizontal
condition, theand verticalcomparing
residuals vibrations results
are 0.589 mm/s
reveal to
the
0.752 mm/s
As shown
advantage and 0.717
of theinproposedmm/s to
Figures 5method, 1.253
and 6, the mm/s,
the ranges respectively,
for horizontal
horizontal which
vibration orand are normal
vertical
vertical vibration
vibrations
vibration. Theare ranges. The
0.589absolute
average mm/s to
vibrations
0.752 mm/s
values in the dataset
and 0.717
of residuals are
mm/s
for the steady, and
to 1.253method
proposed in that condition,
mm/s, respectively, the
for horizontalwhichresiduals comparing
are normal
and vertical results
vibration
vibrations reveal
ranges.
are 0.003 the
The
mm/s
advantage of the proposed method, the horizontal vibration or vertical vibration. The
vibrations in the dataset are steady, and in that condition, the residuals comparing results reveal the average
advantage of the proposed method, the horizontal vibration or vertical vibration. The average
Appl. Sci. 2020, 10, 5637 12 of 17

and 8.73 × 10−4 mm/s, respectively, that for traditional method are 0.0143 mm/s and 0.0238 mm/s,
respectively. The similarity obtained according to Equation (17) also reflects the advantage of the
proposed method, the similarity for the proposed method is above 0.97, including horizontal and
vertical vibrations, that for most of samples are over 0.999, and the results for similarity comparisons
prove the advantage of proposed method.
For comprehensively evaluating the ability of characterizing normal state envelope space, the root
mean square error (RMSE) and MAPE (mean absolute percentage error) are applied for comparison,
the calculating formula are,
s
n
( yi − ŷi )2
P
i=1
RMSE = n (21)
n yi − ŷi
1 P
MAPE = n yi
i=1

where yi and ŷi are observation and estimation values, respectively.

The RMSE and MAPE comparison of proposed method and traditional method for normal
condition I is shown in Table 4. In normal condition I, the RMSE and MAPE for proposed method of
horizontal vibration x1 are 0.0045 mm/s and 0.45%, respectively, that for traditional method are 0.0265
mm/s and 2.12%, respectively. For vertical vibration x2 , the RMSE and MAPE for proposed method
are 0.0017 mm/s and 0.11%, that for traditional method are 0.0177 mm/s and 2.79%, respectively. The
results reveal that while in a relative steady normal condition, the proposed method achieves a better
estimation performance.

5.1.2. Dynamic Working Condition

In order to further validate the effectiveness of the proposed method, the horizontal and vertical
vibrations for blower fan in dynamic work condition, named condition II, are applied. The comparison
Appl. Sci. 2020, 10, x FOR PEER REVIEW 13 of 17
for proposed method and traditional method in condition II is shown in Figures 7 and 8.

Figure 7.7.Comparison
Figure Comparisonfor
forproposed
proposedmethod
methodand traditional
and method
traditional for for
method horizontal vibration
horizontal condition II.
vibration
condition II.
 Figure 7. Comparison for proposed method and traditional method for horizontal vibration
Appl. Sci. 2020, 10, 5637 13 of 17
condition II.

Figure 8. Comparison for proposed method and traditional method for vertical vibration condition II.
Figure 8. Comparison for proposed method and traditional method for vertical vibration condition
II.
As shown in Figure 7, the horizontal vibration of test data set is change in the range of 1.144 mm/s
to 1.767 mm/s, which is the normal vibrating range according to engineering handbook. Comparing
the residual results, the proposed method achieves a higher accuracy, the residual is contained in
−0.0243 mm/s to 0.0274 mm/s, which for traditional method is −0.0411 mm/s to 0.0756 mm/s.
It should be pointed out that the condition changed obviously at certain samples, such as sample
A and sample B marked in the Figure, and the condition changes do not reflect in residuals obviously,
however, the similarity reveals the condition changes accurately, at the sample A and sample B, the
vibrate condition changes and similarity descends correspondingly. Owning to the calculation of the
other variables in observation vector, information except direct vibration variables is contained in
similarity calculation, which is significant for a comprehensive vibration condition monitoring.
As shown in Figure 8, the vertical vibration of test dataset is changed in the range of 1.573 mm/s
to 3.273 mm/s, which is also a normal vibrating range, and the trend for vertical vibration is similar
with horizontal vibration. The ranges for residual for proposed method and traditional method are
−0.048 mm/s to 0.1434 mm/s, −0.5495 mm/s to 0.192 mm/s, respectively, and the results reveal that the
proposed method achieve a higher estimation accuracy.
The RMSE and MAPE for condition II is shown in Table 5. For horizontal vibration x1 , the RMSE
and MAPE of the proposed method are 0.0077 mm/s and 0.36%, respectively, that for traditional
method are 0.0189 mm/s and 0.85%, respectively. For vertical vibration x2 , the RMSE and MAPE for
the proposed method are 0.0293 mm/s and 0.94%, that for the traditional method are 0.1323 mm/s and
3.33%, respectively. The results reveal that while in dynamic normal condition, the proposed method
keeps a superior estimation performance.

Table 5. Comparison of RMSE and MAPE for condition II.

Situation RMSE (mm/s) MAPE (%)

Proposed Method 0.0077 0.36
Horizontal vibration x1
Traditional Method 0.0189 0.85
Proposed Method 0.0293 0.94
Vertical vibration x2
Traditional Method 0.1323 3.33
Appl. Sci. 2020, 10, 5637 14 of 17

5.2. Incipient Failure Detection

According to the operation handbook, the vibration conditions often evaluated by measuring
values directly, the vibration alarm is not triggered until the vibration measure values exceed the
threshold, therefore, an early alarm could not be realized, and the condition judgment criteria is shown
in Table 6, which means once the vibration measurements exceed 5.1 mm/s, the alarm occurs.

Table 6. The vibration threshold for investigated blower fan.

Alarm Level Vibration Measurement Condition

/ X ≤ 5.1 mm/s Normal
I 5.1 mm/s < x ≤ 7.1 mm/s Carry out operations, maintenance if necessary
II X > 7.1 mm/s Immediately shutdown and maintenance

The exceed-threshold events could be detected by an effective failure early detection, and the
incipient failure thresholds for residual and similarity proposed in 4.4 are applied. As shown in
Figure 9, the typical vertical vibration alarm events are discussed, and the process is exhibited, residual
and similarity are shown correspondingly. There are two vibration alarms in the process, noted as
alarm 1 and alarm 2, respectively. According to the data records, the timelines for the event are:

• Sample 1~Sample 123: Normal

•Appl.Sample 124 (2015/11/25 14:01:00): vertical vibration exceeds threshold (5.75 mm/s)
Sci. 2020, 10, x FOR PEER REVIEW 15 of 17
• Sample 139 (2015/11/25 14:06:00): check vibration status and adjust the air volume
• Sample
Sample151 151(2015/11/25
(2015/11/2514:10:00):
14:10:00):vertical
verticalvibration
vibrationrecovery
recovery
• Sample 152~275:
Sample 152~275: NormalNormal
• Sample
Sample276 276(2015/11/25
(2015/11/2514:51:40)
14:51:40)vertical
verticalvibration
vibrationexceeds
exceedsthreshold
threshold(5.20
(5.20mm/s)
mm/s)

6
obsevation
x 2 /[mm/s]

alarm threshold
4

2 estimation

Alarm 1 Alarm 2
0
0 50 100 150 200 250 300
0.5
residual/[mm/s]

-0.5
Alarm 1 Alarm 2
-1
0 50 100 150 200 250 300
1
similarity

0.9

0.8
Alarm 1 Alarm 2
0.7
0 50 100 150 200 250 300
samples/[20/s]

Figure9.9.Process
Figure processof
oftypical
typicalvertical
verticalvibration
vibrationalarm
alarmevent.
event.

The size of sliding window is set as 10, and the effectiveness of residual threshold and similarity
The size of sliding window is set as 10, and the effectiveness of residual threshold and similarity
threshold are validated in the alarms. The thresholds, current value, and corresponding sample for
threshold are validated in the alarms. The thresholds, current value, and corresponding sample for
alarms are compared in Table 7. In the alarm 1, the actual alarm is NO.124 sample, and the failure alarm
alarms are compared in Table 7. In the alarm 1, the actual alarm is NO.124 sample, and the failure
is earlier detected in NO.120 sample and NO.121 sample by using residual threshold and similarity
alarm is earlier detected in NO.120 sample and NO.121 sample by using residual threshold and
threshold, respectively, and the failure could be detected 1 minute ahead. Furthermore, in the alarm 2,
similarity threshold, respectively, and the failure could be detected 1 minute ahead. Furthermore, in
the alarm 2, the actual alarm is NO.276 sample, in comparison, the failure alarm is detected in
NO.263 sample and NO.260 sample, respectively, and the failure is detected over 5 minutes ahead.
For a further discussion, the other variables in vector used for similarity calculation are
investigated, and the results indicated that the deviations of the temperature of end-side bearings x3
Appl. Sci. 2020, 10, 5637 15 of 17

the actual alarm is NO.276 sample, in comparison, the failure alarm is detected in NO.263 sample and
NO.260 sample, respectively, and the failure is detected over 5 minutes ahead.
For a further discussion, the other variables in vector used for similarity calculation are investigated,
and the results indicated that the deviations of the temperature of end-side bearings x3 and current
of driving motor x6 lead to the similarity descend in the early stage of failure, as shown in Figure 10.
The results indicate that failure information reflected by the other variables indicates the prognosis
of failure.

Table 7. Comparison of different failure detection method.

Method Factor Alarm 1 Alarm 2

Threshold 5.1 mm/s 5.1 mm/s
Vibration
Measurements Current Value 5.75 mm/s 5.20 mm/s
Sample Number 124 276
Threshold ±0.2385 mm/s ±0.00625 mm/s
Residual Current Value 0.2492 mm/s 0.007 mm/s
Sample Number 120 263
Threshold 0.908 0.901
Similarity
Current Value 0.832 0.848
Sample Number 121 260
Appl. Sci. 2020, 10, x FOR PEER REVIEW 16 of 17

100
obsevation
90 estimation

80

70

NO.121 sample NO.260 sample

60
0 50 100 150 200 250 300

220

200

180

160

140
0 50 100 150 200 250 300
sample/[20s]

Figure 10.The
Figure10. trendofofx3x3and
Thetrend andx6x6while
whilealarms.
alarms.

6. Conclusions
6. Conclusions
An industrial equipment condition monitoring method based on MEST is proposed in this paper,
An industrial equipment condition monitoring method based on MEST is proposed in this
and a novel method for construction of the memory matrix is put forward. The improved similarity
paper, and a novel method for construction of the memory matrix is put forward. The improved
that combines relative information in other variables is applied in monitoring. For different equipment
similarity that combines relative information in other variables is applied in monitoring. For
and monitoring parameters, the method could be applied. The actual operational data for a blower fan
different equipment and monitoring parameters, the method could be applied. The actual
in thermal power plant is used for validation, and the condition estimation performances are discussed.
operational data for a blower fan in thermal power plant is used for validation, and the condition
The actual data for failure event are used for validating the effectiveness of incipient failure detection
estimation performances are discussed. The actual data for failure event are used for validating the
effectiveness of incipient failure detection employing proposed condition monitoring and early
alarm method, the performance is compared with traditional exceed threshold method.
The method has advantages such as, it is concise and easy to understand conceptually, which
should be attractive to industrial engineers. Irrespective of the steady or dynamic work condition,
the proposed memory matrix construction method improves the accuracy of condition monitoring.
Appl. Sci. 2020, 10, 5637 16 of 17

employing proposed condition monitoring and early alarm method, the performance is compared
with traditional exceed threshold method.
The method has advantages such as, it is concise and easy to understand conceptually, which
should be attractive to industrial engineers. Irrespective of the steady or dynamic work condition, the
proposed memory matrix construction method improves the accuracy of condition monitoring. The
proposed method using residual and similarity enhances the alarm timeliness, and proposed method
could identify the incipient failure ahead.
Although the proposed method has been employed in a specific blower fan, the model can be
further explored with suitable parameters selection and threshold determination. The method should
be further proved for different industrial equipment, and further research is planned for taking the
method into account in other industrial equipment.

Author Contributions: D.L. proposed the idea and carried out the validation; H.Z. guided the research; F.H.
collected the data. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Hameed, Z.; Hong, Y.S.; Cho, Y.M.; Ahn, S.H.; Song, C.K. Condition monitoring and fault detection of wind
turbines and related algorithms: A review. Renew. Sustain. Energy Rev. 2009, 13, 1–39. [CrossRef]
2. Wang, T.; Han, Q.; Chu, F.; Feng, Z. Vibration based condition monitoring and fault diagnosis of wind turbine
planetary gearbox: A review. Mech. Syst. Sig. Process. 2019, 126, 662–685. [CrossRef]
3. Tavner, P.J. Review of condition monitoring of rotating electrical machines. IET Electr. Power Appl. 2008, 2,
215–247. [CrossRef]
4. Rusiński, E.; Moczko, P.; Odyjas, P.; Pietrusiak, D. Investigation of vibrations of a main centrifugal fan used
in mine ventilation. Arch. Civ. Mech. Eng. 2014, 14, 569–579. [CrossRef]
5. Carden, E.P. Vibration Based Condition Monitoring: A Review. Struct. Health Monit. 2004, 3, 355–377.
[CrossRef]
6. Ristivojevic, M.; Mitrovic, R.; Lazovic, T. Investigation of causes of fan shaft failur. Eng. Fail. Anal. 2010, 17,
1188–1194. [CrossRef]
7. Ruiliang, Z.; Xi, G.; Fengshou, G.; Wang, T.; Ball, A.D. Gear Wear Process Monitoring Using a Sideband
Estimator Based on Modulation Signal Bispectrum. Appl. Sci. 2017, 7, 274.
8. Qingjun, W.; Chen, H.; Chen, X.; Yang, H.; Wang, G. Early fault detection of gearbox using weak vibration
signals. Eksploat. I Niezawodn. Maint. Reliab. 2011, 49, 11–15.
9. Bolaers, F.; Cousinard, O.; Marconnet, P.; Rasolofondraibe, L. Advanced detection of rolling bearing spalling
from de-noising vibratory signals. Control Eng. Pract. 2004, 12, 181–190. [CrossRef]
10. Singh, G.K.; Kazzaz, S.A. Induction machine drive condition monitoring and diagnostic research—a survey.
Electr. Power Syst. Res. 2003, 64, 145–158. [CrossRef]
11. Ebersbach, S.; Peng, Z. Expert system development for vibration analysis in machine condition monitoring.
Expert Syst. Appl. 2008, 34, 291–299. [CrossRef]
12. Dai, J.; Tang, J.; Shao, F.; Huang, S.; Wang, Y. Fault Diagnosis of Rolling Bearing Based on Multiscale Intrinsic
Mode Function Permutation Entropy and a Stacked Sparse Denoising Auto-encoder. Appl. Sci. 2019, 9, 2743.
[CrossRef]
13. Harrison, G.A. Neural Network Based Analysis System for Vibration Analysis and Condition Monitoring.
J. Acoust. Soc. Am. 2001, 111, 1969. [CrossRef]
14. Kong, Z.; Tang, B.; Deng, L.; Liu, W.; Han, Y. Condition monitoring of wind turbines based on spatio-temporal
fusion of SCADA data by convolutional neural networks and gated recurrent units. Renew. Energy 2020, 146,
760–768. [CrossRef]
15. Bangalore, P.; Letzgus, S.; Karlsson, D.; Patriksson, M. An artificial neural network-based condition monitoring
method for wind turbines, with application to the monitoring of the gearbox. Wind Energy 2017, 20, 1421–1438.
[CrossRef]
Appl. Sci. 2020, 10, 5637 17 of 17

16. Toh, G.; Park, J. Review of Vibration-Based Structural Health Monitoring Using Deep Learning. Appl. Sci.
2020, 10, 1680. [CrossRef]
17. Cempel, C. Optimization of symptom observation matrix in vibration condition monitoring. In Proceedings
of the Reliability, Maintainability and Safety, 2009. ICRMS 2009. 8th International Conference on IEEE,
Chengdu, China, 20–24 July 2009.
18. Gross, K.C.; Singer, R.M.; Wegerich, S.W. Application of a model-based fault detection system to nuclear
plant signals. In Proceedings of the Office of Scientific & Technical Information Technical Reports, Seoul,
Korea, 6–10 July 1997.
19. Bockhorst, F.K.; Gross, K.C.; Herzog, J.P.; Wegerich, S.W. MSET modeling of Crystal River-3 venturi flow
meters. In Proceedings of the Office of Scientific & Technical Information Technical Reports, San Diego, CA,
USA, 10–15 May 1998.
20. Guo, P.; Infield, D.; Yang, X. Wind Turbine Generator Condition-Monitoring Using Temperature Trend
Analysis. IEEE Trans. Sustain. Energy 2012, 3, 124–133. [CrossRef]

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).