Proceedings of the 2010 14th International Conference on Computer Supported Cooperative Work in Design
Load Identification in Nonintrusive Load Monitoring
Using Steady-State and Turn-on Transient Energy
Algorithms
Hsueh-Hsien Chang
Ching-Lung Lin, Jin-Kwei Lee
Dept. of Electronic Engineering
Jin-Wen University of Science and Technology
Taipei, Taiwan
[email protected]
Dept. of Electrical Engineering
Ming-Hsin University of Science and Technology
Hsinchu, Taiwan
[email protected], [email protected]
addressed the load identification of power signatures in NILM.
Hart [1] proposed a load identification method that examined
the steady-state behavior of loads. Hart conceptualized a finite
state machine to represent a single appliance in which power
consumption varied discretely with each step change. The
method performs well. However, it has the limitations of the
method. For example, small appliances and appliances, which
are always on or non-discrete changes in power, should not be
chosen as targets for the method [1], [4]. Robertson [5]
employed a wavelet transformation technique to classify
several unknown transient behaviors for load identification.
This technique, however, is expensive for the detection of
transients. In addition, the detection of transient behavior can
be obscured by the simultaneous transient of other loads [6].
Cole [6], [7] examined a data extraction method and a steadystate load identification algorithm for NILM. The algorithm
developed by Cole can be employed for load switching
between individual appliances when one or more appliances
are switched on or off. This algorithm, however, requires an
extended period of time to accumulate real power (P) and
reactive power (Q) for sample data. In addition, any appliance
power consumption that does not change cannot be recognized
[7].
Abstract— Non-intrusive load monitoring (NILM) techniques are
based on the analysis of load energy signatures. With
characterizing associated transient energy signature, the
reliability and accuracy of recognition results can be accurately
understood or ascertained. In this study, the computer supported
cooperative work techniques (CSCW), artificial neural networks
(ANN), in combination with turn-on transient energy analysis,
are used to identify loads and to improve recognition accuracy
and computational speed of NILM results. The experimental
results indicated that the incorporation of turn-on transient
energy signature analysis into NILM revealed more information
than traditional NILM methods, and the resulting recognition
accuracy and computational speed were improved. In addition,
in combination with computer supported cooperative work in
electromagnetic transient program (EMTP) simulation,
calculations of turn-on transient energy facilitated load
identification that had significant effect on NILM results.
Keywords- load monitoring; pattern recognition; artificial
neural networks; CSCW
I.
INTRODUCTION
Traditional load-monitoring instrumentation systems
employ meters for each load to be monitored because they
tend to be comprehensive, systematic, and convenient. These
meters may incur significant time and costs to install and
maintain. Furthermore, increasing numbers of meters may
influence system reliability. Some research also indicate that
the utility of load-monitoring systems have been questioned
by load-monitoring system practitioners, and future studies of
load-monitoring systems will focus on more significant issues,
such as strategies for minimizing the number of instruments
using non-intrusive load monitoring (NILM) system [1]-[3].
Figure 1 shows the NILM system used to monitor voltage and
current waveforms in an electrical service entry powering
loads representative of different important load classes.
Recently, several papers have proposed new power
signature analysis algorithms [8]-[12], load identification
methods [13]-[16], and feature selection approaches [17]-[19]
to recognize loads and to solve classification problems. For
the load identification methods, many papers have been
published to improve the performance of recognition using
artificial neural networks for the NILM system. For example,
Roos et al. [2] proposed a detailed analysis of steady-state
appliance signatures to recognize industrial electrical loads.
This method, however, requires complicated computations for
accurate data of power signatures. In addition, Srinivasan et
al. [16] proposed a neural-network-based approach to identify
non-intrusive harmonic source. The method performs well.
However, it does not incorporate the various operational
modes of each load and operation under different voltage
sources. In a practical power system, there exist many
harmonics. How harmonics affect the results of the proposed
method has been demonstrated by authors in [20]. However,
Due to the importance and difference of recognition
accuracy of power signatures, several previous studies have
H. H. Chang is with the Department of Electronic Engineering, Jin-Wen
University of Science and Technology, HsinTien, Taipei, 231, TAIWAN
(e-mail: [email protected]).
978-1-4244-6763-1/10/$26.00?©2010?IEEE
27
harmonic content is very small for constant linear loads [10],
especially for commercial buildings and residences. Therefore,
another feature besides harmonics is necessary for power
systems, commercial buildings and residences.
of each cycle is sufficient and hence the sampling frequency is
approximately 15 kHz.
In the sinusoidal steady state or under linear time-invariant
loads, complex power is calculated from voltage, current, and
respective phase angles measured as in Eq. 2. In Eq. 2, the real
number is real power (P) or average power and the imaginary
number is reactive power (Q) in the complex power. In [21],
they can be computed by
To solve the disadvantages for the previously published
research, a new method for load identification of the NILM
system is proposed in this paper. This method uses the turn-on
transient energy (UT) analysis, traditional steady-state power
signatures, and artificial neural networks to improve the
recognition accuracy and to reduce computational
requirements. The proposed improvement technique is
unrelated to operational mode of loads, operation under
different voltage sources, and power consumption change. The
proposed method can be applied for commercial loads and
industrial loads. Moreover, the proposed method can be
applied for different loads with the same real power and
reactive power. Experimental results show that the proposed
method for the NILM system allows efficient recognition of
commercial or industrial loads as well as improvement of
computational requirements. Moreover, the turn-on transient
energy signature can be used to distinguish different loads
with the same real power and reactive power.
V = Vm e jθV , I = I m e jθ I
= P + jQ
The current and voltage consumed for a periodically
nonlinear load can be represented by a Fourier series
expansion. The appropriate coefficients corresponding to the
current and voltage in each harmonic are extracted from the
results. The number of terms represented by the expansion
determines the dimension of the feature vector. The real power
and reactive power can be respectively computed by
3φ Current Measurements
Common Bus
P=
N
∑P
n
n=0
Local PC
2
Load 1
Load 2
N
1
= V 0 I 0 + ∑ V n I n cos( θ V n − θ I n )
n =1 2
(3)
and
N
Q =
Load N
N
∑
n =1
Qn =
N
∑
n =1
1
V n I n sin( θ V n − θ I n )
2
(4)
where n is the harmonic number; V 0 and I 0 are the average
Fig. 1. Data collection and load identification system for a NILM system.
II.
(2)
where the variables Vm and I m are respectively the maximum
value of voltage and current, and the variables θ V and θ I are
respectively the phase angles of voltage and current.
3φ Voltage Measurements
1
1
V m I m e j (θ V − θ I )
2
=
Distribution
Transformer
Non-intrusive
Load-monitoring
System
1
VI
2
Pcomplex =
Substation
Host PC
(1)
voltage and average current, respectively; Vn and I n are the
effective nth harmonic components of the voltage and current;
θVn and θ I n represent the nth harmonic components of the
DATA PREPARATION
Figure 1 schematically illustrates the overall scheme in the
NILM system. Three-phase or one-phase electricity powers
the loads, which are representative of important load classes in
an industrial or commercial building. A dedicated computer
connected to the circuit breaker panel controls the operation of
each load. The local computer can also be programmed to
stimulate various end-use scenarios. The computer supported
cooperative work presented in this paper is load recognition
using neural networks and the employment of those features to
estimate the energy consumption of major loads.
voltage and current phase angles, respectively.
B. Data Preprocessing
Neural network training can be made more efficient if
certain preprocessing steps are performed on the network
inputs. Before training, it is often useful to scale the inputs and
targets so that they always fall within a specified range. The
approach for scaling network inputs and targets is to normalize
the mean and standard deviation of the training set,
normalizing the inputs and targets so that they will have zero
mean and unity standard deviation. These can be computed by
A. Data Acquisition
The main parameters to be acquired are the voltage and
current of aggregated loads. To compile data for training
purposes, either every load of interest or a representative
sample of the loads should be monitored. Taking 256 samples
P n = ( P − meanp ) / stdp
and
28
(5)
t n = ( t − meant
) / stdt
(6)
K
where the matrices P and t are respectively the original
network inputs and targets, the matrices Pn and tn represent
respectively the normalized inputs and targets. The vectors
meanp and stdp contain the mean and standard deviations of
the original inputs, and the vectors meant and stdt contain the
means and standard deviations of the original targets.
k =0
where V(k) is derivative of transient voltage for sample k; I(k)
is average transient current for sample k; v( k ) is voltage
sampled for sample k; v( k − 1 ) is voltage sampled for sample
k-1; i( k ) is current sampled for sample k; i( k −1) is current
sampled for sample k-1; K is number of samples, k=1, 2, …K.
C. Experimental Data Sets
Experimental datasets were generated by preprocessing the
data on the voltage and current waveform of the total load.
Each final sample consists of 4,608 data samples obtained
over a period of 0.3s. Each example of the power feature
includes a voltage variation from − 5% to +5% at 1%
intervals, yielding eleven examples of power feature for each
scenario and ( 2 N − 1) × 11 raw data for 2 N − 1 scenarios
given N loads in a power system network. To confirm the
inferential power of the neural networks, the raw data
examples are categorized into (( 2 N − 1) × 11 ) / 2 learning and
test datasets, respectively. The full input dataset comprises a
(( 2 N − 1) × 11 ) × 4608 matrix as both the training dataset
and the test dataset. Notably, the learning data and test data are
selected randomly from all data. A neural network simulation
program was designed using MATLAB. The program was run
to identify load on an IBM PC with an Intel 1.5GHz Pentium
M CPU.
The three-phase turn-on transient energy is computed as
follows:
UT =U3φ,transient= ∑(Va(k)⋅ Ia(k) +Vb(k)⋅ Ib(k) +Vc(k)⋅ Ic(k)) (10)
where V a ( k ), V b ( k ), V c ( k ) are derivatives of transient
voltage in phases a, b, and c for sample k;
I a ( k ), I b ( k ), I c ( k ) are the average value of transient
current in phases a, b, and c for sample k.
900
Pow
er
, kW
800
700
600
500
400
300
200
100
0
0
0 .1
Time, s
0.2
0.3
(a)
TURN-ON TRANSIENT ENERGY ALGORITHMS
4 00
P
ow
er
, kW
III.
(9)
UT = U1φ ,transient = ∑V (k )I (k )
The transient properties of a typical electrical load are
mainly determined by the physical task that the load performs
[22], [23]. Transient energy may assume different forms in
consumer appliances, depending on the generating mechanism
[1]. Estimating current waveform envelopes at the utility
service entry of a building, for example, allows accurate
transient event detection in the NILM [22]. Load classes
performing physically different tasks are therefore
distinguishable by their transient behavior [22], [23]. Since the
envelopes of turn-on transient instantaneous power are closely
linked to unique physical quantities, they can serve as reliable
metrics for load identification. However, the transient is the
dominant state directly after load inception. Figure 2 plots the
turn-on real-power transient of each load for an NIEM system
at the entry of an electrical service. In Figs. 2(a) and 2(b),
these loads are respectively a 160hp induction motor and a
123hp induction motor, driven by variable-voltage drives. The
turn-on real-power transients differ from each other because
the induction motor is started using different methods. In Fig.
2(c), this load is a bank of loads that is supplied by a six-pulse
thyristor rectifier that delivers A.C. power. The real-power
transient is slowly increased to the normal rated power
because of the control method of the thyristor rectifier. The
one-phase turn-on transient energy is determined as follows.
V ( k ) = v ( k ) − v ( k − 1)
(7)
I ( k ) = ( i ( k ) + i ( k − 1 )) / 2
(8)
3 00
2 00
1 00
0
0
0. 1
T ime, s
0.2
0.3
(b)
Pow
er
, kW
400
300
200
100
0
0
0 .1
0.2
0.3
Time, s
(c)
Fig. 2. Turn-on real-power transient for a NIEM system, (a) a 160-hp
induction motor; (b) a 123-hp induction motor driven by line frequency
variable-voltage drives; (c) a bank of loads supplied by a six-pulse thyristor
rectifier for A.C. power.
IV.
MULTI-LAYER FEEDFORWARD NEURAL NETWORK
Most back-propagation (BP) neural network applications
employ single- or multi-layer perceptron networks using
gradient-descent training techniques, with learning by back
propagation. These multi-layer perceptrons can be trained with
supervision using analytical functions to activate network
nodes (“neurons”) and by applying a backward errorpropagation algorithm to update interconnecting weights and
thresholds until proper recognition capability is attained. In the
present study, the back-propagation classifier is generally used
29
as a trainable classifier for a multi-layer feedforward neural
network (MFNN). “Classification” in this context denotes a
mapping from a feature space to the set of class labels – the
names of commercial or industrial load combinations.
transient energy feature can also be used to recognize different
loads with the same real power and reactive power in a NILM
system.
TABLE I
VARIATION COEFFICIENT DURING PERIODS OF NEARLY STEADY ENERGY FOR
EACH LOAD
A supervised MFNN is generally divided into three layers:
input, hidden, and output, including neurons. The neurons are
connected by links with weights that are selected to meet the
desired associations between the input and output neurons.
These weights should be trained with existing input-output
pairs using an appropriate algorithm. An appropriate
momentum and learning rate should be given during the
training phase. The purpose of the MFNN in this paper is to
identify loads of the NILM system. The MFNN based on the
back-propagation method is adopted in this paper and this
ANN can identify the similarity between given data and know
data [24]. The input, output and hidden layers of the ANN are
described as follows:
Loads
C.V. (%)
Load 2
0.16077
Load 3
0.10885
B. Case Study Environment and Results
Each entry in the table represents 10 different trials, where
different random initial weights are used in each trial. In each
case, the network is trained until the mean square error is less
than 0.0001 or the maximum of epoch is 3000.
(1)Case Study 1, EMTP Simulation: In case study 1, a
simulated NILM system monitors the voltage and current
waveforms in a three-phase electrical service entry powering
representative loads in an industrial building. The neural
network algorithm in the NILM system identifies three loads
with transient and steady-state signatures observed during
operation of the 480-V common bus. These loads include a
160-hp induction motor, a 123-hp induction motor driven by
line frequency variable-voltage drives, and a bank of loads
supplied by a six-pulse thyristor rectifier for A.C. power.
1) Input layer: the power signature information including
the real power, reactive power, and/or the turn-on
transient energy for an electrical service entry severs as
inputs.
2) Output layer: the number of output neurons is the same
that of the identified individual appliances. Each binary
bit serves as a load indicator for the ON/OFF status.
Table 2 shows that values for training and test recognition
accuracy of load identification in multiple operations are
100% for features with real power and reactive power (PQ),
and/or with turn-on transient energy (UT) for one of the
features.
3) Hidden layer: Only one hidden layer is used in this
paper. Some heuristics have been proposed to
determine the number of neurons in hidden layer [25].
The common number of neurons for the hidden layer is
(number of input neurons + number of output neurons)/
2 or (number of input neurons + number of output
neurons) 0.5. The simulation results show no significant
difference between these two alternatives.
V.
Load 1
0.36648
TABLE II
THE RESULTS OF LOAD IDENTIFICATION IN CASE STUDY 1
PQ
Training
Test
39
38
PQUT
Training
Test
39
38
Number of
features
Recognizable
39
38
39
38
number
Recognition
100
100
100
100
accuracy (%)
Time (s)
6.8561
0.4724
2.1343
0.4857
Number of
626.9
143.5
epochs
(2)Case Study 2, Experiment: The NILM system in case study
2 monitors the voltage and current waveforms in a three-phase
electrical service entry powering representative loads in the
laboratory. The neural network algorithm in the NILM system
identifies three actual loads with transient and steady-state
signatures on a 220-V common bus. These loads include a
three-phase R-L linear load, a one-phase 0.2-hp induction
motor, and a three-phase 1-hp induction motor.
EXPERIMENTAL RESULTS
A. Turn-on Transient Energy Repeatability
To determine whether turn-on transient energy exhibits
repeatability, a NILM system with three important loads was
examined in an industrial plant. These loads include a 95-hp
induction motor, a 140-hp induction motor, and a bank of
loads supplied by a six-pulse thyristor rectifier for A.C. power.
The turn-on transient energy for each load was computed from
the measured voltage and current waveform at the service
entry according to Eq. 10. Because of the varying transients
(which often depend on the exact point in the voltage cycle at
which the switch opens or closes), it is essential that data sets
for load identification have highly repeatable transient energy
signatures. Therefore, the instantaneous power profile for each
turn-on transient load is sampled when the system utility
voltage is switched from 0° to 350° in 10° intervals, i.e., the
number of load samples (n) is 36.
Table 3 shows that values for the training accuracy of load
identification in multiple operations are 100% for features
with real power and reactive power (PQ), and/or with turn-on
transient energy (UT) for one of the features. Furthermore, the
test accuracy of load identification in multiple operations is at
least 94%.
Table 1 shows C.V. values during periods of nearly steady
energy for each load, all of which are less than 1% [8]. The
simulation results indicate that the turn-on transient energy
should have good repeatability. Therefore, the turn-on
transient energy can be used as a power signature to recognize
commercial or industrial loads. Moreover, the turn-on
30
with transient and steady-state signatures on a 110-V common
bus. These loads include a 119-W dehumidifier, a 590-W
vacuum cleaner, and an R-L linear load with real power and
reactive power equivalent to that of a 590-W vacuum cleaner.
TABLE III
THE RESULTS OF LOAD IDENTIFICATION IN CASE STUDY 2
PQ
Training
Test
39
38
PQUT
Training
Test
39
38
Number of
features
Recognizable
39
36
39
38
number
Recognition
100
94.73
100
100
accuracy (%)
Time (s)
23.7187
0.4874
1.4029
0.4736
Number of
2412.1
67.4
epochs
(3)Case study 3, EMTP Simulation for Different Loads with
The Same Real Power and Reactive Power: In case study 3,
the NILM system monitors voltage and current waveforms in
a three-phase electrical service entry powering a collection of
loads representative of the major load classes in a commercial
building. The neural network algorithm in the NILM system
identifies three loads with transient and steady-state signatures
operating on a 220-V common bus. These loads include a 2.6hp induction motor, a 4.7-hp induction motor, and an R-L
linear load with real power and reactive power equivalent to
that of a 4.7-hp induction motor.
Table 5 shows that values for the training and test
recognition accuracy of load identification in multiple
operations are also all 100% for features with real power and
reactive power, as well as the turn-on transient energy (PQUT).
However, the accuracy of training and test recognition of load
identification in multiple operations are only 51.28% and
39.47%, respectively, for features with real power and reactive
power (PQ). The test recognition for those loads in multiple
operations is also quite low when using only real power and
reactive power features. The reason is the same as that for the
previous section. In other words, the presence of different
loads with the same real power and reactive power can be
confirmed in two ways. First, test recognition in multiple
operations is quite low when only using features of real power
and reactive power. Second, the turn-on transient energy for
one of the features can improve load identification, especially
for different loads with the same real power and reactive
power.
TABLE V
THE RESULTS OF LOAD IDENTIFICATION IN CASE STUDY 4
Table 4 shows that values for the training and test
recognition accuracy of load identification in multiple
operations are all 100% for features with real power and
reactive power, as well as the turn-on transient energy (PQUT).
However, the training and test recognition accuracy of load
identification in multiple operations are only 58.97% and
39.47%, respectively, for features with real power and reactive
power (PQ). Those loads cannot be identified by real power
and reactive power features because the second load and the
third load are different loads with the same real power and
reactive power, as are combinations of the first and second
loads and combinations of the first and third loads. In other
words, test recognition for those loads in multiple operations
is quite low when using only real power and reactive power
features.
Number of
features
Recognizable
number
Recognition
accuracy (%)
Time (s)
Number of
epochs
PQUT
Training
Test
39
38
20
15
39
38
51.28
39.47
100
100
30.1998
0.483
3000
VI.
1.6281
0.4782
79.9
CONCLUSIONS
The problem of power signatures can limit the use of nonintrusive load-monitoring system results for load identification
unless power signature is integrated into the evaluation
process. Therefore, transient power signature analysis is
incorporated into NILM, so that the associated recognition
accuracy can be improved, and NILM results are not
misapplied. Based on experimental results and EMTP
simulation of NILM, the transient power signature for load
identification in NILM can be applied extensively to any case.
ANN and turn-on transient energy analysis are useful tools for
improving load recognition accuracy from 94% to 100% and
reducing computation time from 23.7 seconds to 1.4 seconds
from table 3 in a NILM system.
TABLE IV
THE RESULTS OF LOAD IDENTIFICATION IN CASE STUDY 3
PQ
Training
Test
39
38
PQ
Training
Test
39
38
PQUT
Training
Test
39
38
Number of
features
Recognizable
23
15
39
38
number
Recognition
58.97
39.47
100
100
accuracy (%)
Time (s)
29.3734
0.4938
5.6843
0.4937
Number of
3000
498.8
epochs
(4)Case study 4, Experiment for Different Loads with The
Same Real Power and Reactive Power: In case study 4, the
NILM system is used to monitor voltage and current
waveforms in a one-phase electrical service entry powering
representative loads in the laboratory. The neural network
algorithm in the NILM system identifies three actual loads
To improve recognition accuracy within multiple
operations, especially for different loads with the same real
power and reactive power but no harmonic components,
features cannot be adequately measured only from steady-state
parameters, i.e., real power and reactive power. In contrast to
steady-state properties, transient properties such as turn-on
transient energy can play an important role. Combining
transient and steady-state signatures is necessary to improve
31
recognition accuracy and computational speed. Although the
number of weights and biases with the PQUT network is more
than the PQ network (24 versus. 21), recognition accuracy for
these features PQUT is 100%.
[21]
[22]
REFERENCES
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[23]
G. W. Hart, "Nonintrusive appliance load monitoring," in Proc. 1992
IEEE Conf., pp. 1870–1891.
J. G. Roos, I. E. Lane, E. C. Lane, and G. P. Hanche, "Using Neural
networks for non-intrusive monitoring of industrial electrical loads," in
Proc. 1994 IEEE Instrumentation and Measurement Technology Conf.,
pp. 1115-1118.
S. B. Leeb, "A conjoint pattern recognition approach to nonintrusive
load monitoring," Ph.D. dissertation, MIT, 1993.
A. I. Cole and A. Albicki, "Nonintrusive identification of electrical loads
in a three-phase environment based on harmonic content," in Proc. 2000
IEEE Instrumentation and Measurement Technology Conf., pp. 24-29.
D. C. Robertson, O. I. Camps, J. S. Mayer, and W. B. Gish, Sr.,
"Wavelets and electromagnetic power system transients," IEEE Trans.
Power Delivery, vol. 11, (2), pp. 1050-1056, 1996.
A. I. Cole and A. Albicki, "Data extraction for effective non-intrusive
identification of residential power loads," in Proc. 1998 IEEE
Instrumentation and Measurement Technology Conf., pp. 812-815.
A. I. Cole and A. Albicki, "Algorithm for non-intrusive identification of
residential appliances," in Proc. 1998 IEEE International Symposium on
Circuits and Systems, pp. 338-341.
L. K. Norford and S. B. Leeb, "Non-intrusive electrical load monitoring
in commercial buildings based on steady-state and transient loaddetection algorithms," Energy and Buildings, vol. 24, pp. 51-64, 1996.
C. Laughman, K. Lee, R. Cox, S. Shaw, S. B. Leeb, L. Norford, and P.
Armstrong, "Power signature analysis," IEEE Power & Energy
Magazine, pp. 56-63, 2003.
K. D. Lee, S. B. Leeb, L. K. Norford, P. R. Armstrong, J. Holloway, and
S. R. Shaw, "Estimation of variable-speed-drive power consumption
from harmonic content," IEEE Trans. Energy Conversion, vol. 20, (3),
pp. 566-574, 2005.
S. R. Shaw and C. R. Laughman, "A Kalman-Filter spectral envelope
preprocessor," IEEE Trans. Instrumentation and Measurement, vol. 56,
(5), pp. 2010-2017, 2007.
S. R. Shaw, S. B. Leeb, L. K. Norford, and R. W. Cox, "Nonintrusive
load monitoring and diagnostics in power systems," IEEE Trans.
Instrumentation and Measurement, vol. 57, (7), pp. 1445-1454, 2008.
M. L. Marceau and R. Zmeureanu, "Nonintrusive load disaggregation
computer program to estimate the energy consumption of major end
users in residential buildings," Energy Conversion & Management, vol.
41, pp. 1389-1403, 2000.
L. Farinaccio and R. Zmeureanu, "Using a pattern recognition approach
to disaggregate the total electricity consumption in a house into the
major end-uses," Energy and Buildings, vol. 30, pp. 245-259, 1999.
M. Baranski and J. Voss, "Genetic algorithm for pattern detection in
NIALM systems," IEEE 2004 International conference on System, Man
and Cybernetics, pp. 3462-3468.
D. Srinivasan, W. S. Ng, and A. C. Liew, "Neural-network-based
signature recognition for harmonic source identification," IEEE Trans.
Power Delivery, vol. 21, (1), pp. 398-405, 2006.
C. N. Hsu, H. J. Huang, and D. Schuschel, "The ANNIGMA-Wrapper
approach to fast feature selection for neural nets," IEEE Trans. System,
Man and Cybernetics C., vol. 32, (2), pp. 207-212, 2002.
N. Kwak and C. H. Choi, "Input feature selection for classification
problems," IEEE Trans. Neural Networks, vol. 13, (1), pp. 143-159,
2002.
M. Kudo and J. Sklansky, "Comparison of algorithms that select features
for pattern classifiers," Pattern Recognition, vol. 33, pp. 25-41, 2000.
H. T. Yang, H. H. Chang, and C. L. Lin, "Design a neural network for
features selection in non-intrusive monitoring of industrial electrical
[24]
[25]
32
loads," in Proc. 2007 International Conference on Computer Supported
Cooperative Work in Design, pp. 1022-1027.
C. A. Desoer and E. S. Kuh, Basic Circuit Theory, New York: McGrawHill, 1969.
S. B. Leeb, S. R. Shaw, and J. L. Kirtly, Jr., "Transient event detection in
spectral envelop estimates for nonintrusive load monitoring," IEEE
Trans. Power Delivery, vol. 10, (3), pp. 1200-1210, 1995.
U. A. Khan, S. B. Leeb, and M. C. Lee, "A multiprocessor for transient
event detection," IEEE Trans. Power Delivery, vol. 12, (1), pp. 51-60,
1997.
B. Kosko, Neural networks and fuzzy systems, New Jersey: PrenticeHall, 1992.
Y. Y. Hong and B. Y. Chen, "Locating switched capacitor using wavelet
transform and hybrid principal component analysis network," IEEE
Trans. Power Delivery, vol. 22, (2), pp. 1145-1152, 2007.