Load Identification in Nonintrusive Load Monitoring Using Steady ...

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



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


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




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.