Neural Network Analysis for Classification of Defected Ceramic Insulators

doi:10.21203/rs.3.rs-3218605/v1

Download PDF

Research Article

Neural Network Analysis for Classification of Defected Ceramic Insulators

https://doi.org/10.21203/rs.3.rs-3218605/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this older preprint version

Read the latest preprint version →

Partial discharge (PD) is a phenomenon that could occur in insulating materials when there is a localized breakdown of the electrical insulation. PD can lead to the formation of small arcs or sparks within the insulating material, which can cause damage and degradation to the insulator over time. In ceramic insulators, there are several factors that can cause PD including manufacturing defects, aging, and exposure to environmental conditions such as moisture and temperature extremes. As a result, detecting and monitoring PD in ceramic insulators is important for ensuring the reliability and safety of electrical systems that rely on these insulators. This can be done using various techniques which can provide information about the location, severity, and frequency of PD events. In this study, acoustic emission technique is introduced for PD detection and condition monitoring of defective ceramic insulators. A sequence of data processing techniques are performed on the captured signals to extract and select the most significant signatures for classification of defects in insulator strings. Moreover, Fourier transform analysis is adopted to be compared to the wavelet transform analysis. Artificial neural network (ANN) has been used to build an intelligent classifier for easily and accurately classification of defective insulators. The overall recognition rate of the classifier was obtained as 96.03% from discrete wavelet transform analysis and 88.65% from fast Fourier transform analysis. This obtained result indicates high accuracy and performance classification.

partial discharge

acoustic emission

data processing

Artificial Neural Network

At both transmission and distribution voltage ratings, utilities have employed overhead lines that are supported by outdoor insulators. These insulators serve a crucial function of insulating the high voltage transmission line from the steel tower (Ahmed S. Haiba et al. 2019). Although due to their nature operation in overhead transmission and distribution lines, outdoor ceramic insulators are subjected to various defects under operating and environmental conditions. One of these defects is the expansion of cement which can form cracks in the insulator and this expansion can be accelerated under wet conditions leading to insulator failure (Anjum 2014). As a result, Cherney and Hoton (Cherney and Hooton 1987) introduced water expansion tests on the used cement to ensure its quality to be assembled with the insulator. Also, erosion and corrosion of pin and cap hardware in the insulator is another defect that can be occurred due to the salt solution formation on the insulator surface. At wet atmospheric conditions, the formed salt solution may lead to loss in cross-section area and mechanical strength in the insulator pin leading to dropping the overhead conductor (Cherney 1982). Furthermore, contamination flashover is the main problem of overhead insulators (Zhao et al. 2015; Zhang et al. 2013; Krystian Leonard Chrzan 2010; Krystian L. Chrzan, Vosloo, and Holtzhausen 2011). Contaminants, such as salts, chemical particles, dust, sand, and etc, can be accumulated on the surface of the insulator. At wet conditions, the pollution surface layer is converted into a conductive layer and the leakage current passes through these layers on the insulator surface causing heating of that layer. Dry bands are then formed and discharges can continue until occurring of insulator flashover and outage of the power line (He and Gorur 2016). Ceramic insulators may be broken and punctured due to the occurrence of over-voltages from the lightning on the transmission line (Cherney 1982). Also, during manufacturing process, very small internal voids or cracks may be developed within the insulator (Zener and Wills 1934; Polisetty 2019). These voids may cause continuous discharges under operating conditions leading to the insulator failure. As a result, PD growth measurement is the most significant indicator of all the above mentioned defects that occurred for the overhead insulators (Ghosh, Chatterjee, and Dalai 2017). Consequently, many researches have introduced various methods to evaluate the insulators’ condition based on PD signatures (Ahmed S. Haiba et al. 2019). Many detection techniques, based on both electrical and non-electrical methods, have been introduced for the detection, analysis and evaluation of PD activities (IEC Standards 60270 2000; Albano et al. 2016; FRĄCZ 2016; Si et al. 2010; Yaacob et al. 2014; Sahoo, Salama, and Bartnikas 2005). Each technique has benefits and drawbacks and one may be better than the other depending on the application. A brief overview of the widely techniques that used for monitoring the condition of the overhead line insulators will be provided in figure (1) (Hussain, Refaat, and Abu-Rub 2021).

Over time, various parameters have been manually extracted from recorded patterns or signals with the goal of using these parameters to build a classifier capable of differentiating and characterizing faults, partial discharges, defects, or degradation (Mantach et al. 2022). While the process has become somewhat automated, the need for experts to manually select the features presents a challenge, as different features may lead to different outcomes. This relying on selected features manually can adversely affect the classifier's performance.

Deep learning enables the integration of feature selection with the learning process, thereby automating the entire process. In the realm of applications of high voltage, the primary objective has been to localize or classify faults, defects, or partial discharges that could occur in high voltage equipment, or to detect the deterioration of insulating materials. Classification involves differentiating between various sources of faults, defects, or partial discharges, or levels of degradation. Artificial intelligence techniques that applied to classify the condition of insulators holds great promise, as defects could be automatically fixed out of pattern recognition (Stefenon et al. 2022). Deep learning techniques can be utilized to detect the presence of a fluctuating number of missing disks in transmission lines consisting of chain insulators (Sampedro et al. 2019). The main challenge in using computer vision to detect insulator failures is the infrequency of such failures. As a result, training a network to recognize specific conditions becomes challenging due to the limited dataset available. (Sampedro et al. 2019). However, applying engineering constraints for datasets captured through inspections could enhance the model's accuracy, resulting in an accuracy rate of up to 92.86%, as demonstrated in (Shi and Huang 2021).

Many techniques of artificial intelligence have been developed for predicting the stability of smart grids (Arzamasov, Bohm, and Jochem 2018; Aghamohammadi and Abedi 2018; Gupta, Gurrala, and Sastry 2019; Zare, Alinejad-Beromi, and Yaghobi 2019; Chen et al. 2019). Furthermore, researchers have shown the widespread utilization of these techniques such as Artificial Neural Network (ANN) (Dadashizadeh Samakosh and Mirzaie 2019; Belhouchet, Bayadi, and Bendib 2019; A. A. Salem et al. 2020), Support Vector Machine (SVM) (Mahdjoubi et al. 2019), fuzzy logic (FL) (Asimakopoulou et al. 2011), K-means clustering (Farshad 2019), and Hidden Markov Model (HMM) (Lu et al. 2019) in addressing electrical power system and high voltage engineering issues (A. Salem et al. 2021; Tahir Khan Niazi et al. 2020). Intelligent systems can improve the reliability of the transmission and distribution power system, cost savings, and reduce human effort by facilitating effective assessing the state and performance of outdoor insulators during voltage operation (Surya Prasad and Prabhakara Rao 2016). In their study, Salem et al. (A. A. A. Salem et al. 2020) utilized insulator diameter, height, creepage distance, form factor, and Equivalent Salt Deposit Density (ESDD) as input parameters to train a model that combined the Adaptive Neuro Fuzzy Inference System (ANFIS) with ANN. Also, by establishing correlations between leakage current and weather conditions, A. Din et al. (Din et al. 2021) used SVM technique to evaluate the leakage current for outdoor insulators. In another work, Saranya et al. (Saranya and Muniraj 2016) put forward a novel approach for assessment the status and performance of outdoor insulators, which involves recognizing the arc faults of insulators through measurements of phasor angle.

In this current work, detection technique of acoustic emission is used for detection PD signatures resulting in artificial defects in ceramic outdoor insulators. A series of advanced signal processing techniques are performed on the captured signals from the experimental section to extract and select the most significant features to be as input data for the proposed classifier. Finally, artificial neural network (ANN) is proposed in order to assess easily, cost-effectively and accurately the classification of defective insulators.

2.1 Experimental test set up

The common defects that could be occurred in the ceramic outdoor insulators are breaking and cracking of the ceramic shell dielectric material. Occurrence of these defects depends on several factors such as the places of these insulators, contamination degree, environmental conditions and operating stresses.

Three samples of ceramic insulators with 11 kV rating voltage were tested in this work. Variable artificial defects were introduced in two of them. One sample was completely broken and the other was cracked. The third one is considered the reference healthy sample without any apparent defects. Figures (1) shows the three samples tested in this study.

This passage describes the experimental test setup used in a particular study. The setup is shown in Figure (2) and involves a 100 kV high voltage transformer that applies a test voltage to the samples being evaluated. To generate PD activities, the individual disc samples are exposed to an AC voltage of 40 kV, 50 Hz, which causes PD activities due to various defects. The signals emitted by the defective insulators, which are indicative of PD activities, are captured using an 80 MHz acoustic sensor that senses acoustic waves. These signals are then transmitted via a BNC coaxial cable to a digital oscilloscope with a 500 MHz bandwidth and a 500 MS/s sampling rate for further analysis. The oscilloscope is linked to a personal computer (PC) through a GPIB cable for recording the radiated signals. 100 PD pulses are captured and recorded for each tested insulator, with each pulse comprising 4000 raw data points indicating the amplitude in mV versus time in µs.

2.2 feature extraction and selection

After collecting the acoustic emission signals from various PD activities, these signals shall be analyzed to identify the correlation between the obtained signal and the corresponding defect. As a result, a series of procedures are applied to the collected signals in order to extract and get useful information. In this study, the acoustic collected signals are processed in four steps: wavelet transform, feature extraction, feature selection, and classification. Furthermore, Fourier transform technique is also used to be compared to the wavelet transform analysis tool.

Discrete wavelet transform (DWT) is used as a first stage to decompose the original signal into two groups namely: approximations which reflect the signal's high-scale and include low-frequency components and details which indicate the signal's low-scale and include high-frequency components.

DWT is used as a feature extraction technique for extracting characteristics or features from the decomposed high and low frequency signals. After decomposing the original signal by DWT into 5-levels of approximation and detail signals, the features can be then extracted from each decomposed level. Each of the wavelet components that have been decomposed yields seven descriptive waveform features: mean, variance, maximum amplitude, minimum amplitude, mean of energy, skewness, and kurtosis, which are characterized as follows:

Mean (m) is defined as the average value of the number of samples (n).

\(\text{m}=\frac{1}{\text{n}} {?}_{\text{i}=1}^{\text{n}}{\text{x}}_{\text{i}}\)

(1)

Where: n is samples’ number and \(\text{x}\) is the sequence of the samples.

Variance (V) is a statistical measure that quantifies the amount of variability or spread in a set of data values. It is calculated by taking the average of the squared differences of each data point from the mean.

\(\text{V}=\frac{1}{\text{n}-1}\underset{\text{i}=1}{\overset{\text{n}}{?}}{({\text{x}}_{\text{i}}-\text{m})}^{2}\)

(2)

Maximum amplitude of a signal refers to the highest value of the amplitude that the signal reaches during its cycle.

Minimum amplitude of a signal refers to the lowest value of the amplitude that the signal reaches during its cycle.

Mean of energy (m_e) is the energy mean value (Ahmed S. Haiba et al. 2019) and can be calculated by taking the sum of the squares of the values in the signal, divided by the total number of values.

\({\text{m}}_{\text{e}}=\frac{1}{\text{n}} {?}_{\text{i}=1}^{\text{n}}{{\text{x}}_{\text{i}}}^{2}\)

(3)

Skewness (S) is a measure of the extent of asymmetry in a distribution with reference to the sample mean (Ahmed S. Haiba et al. 2019).

\(\text{S}=\frac{{?}_{\text{i}=1}^{\text{n}}{{(\text{x}}_{\text{i}}-\text{m})}^{3}}{(\text{n}-1){\text{s}}^{3}}\)

(4)

Where: \({\sigma }\) is the standard deviation.

Kurtosis (K) is a measure of how peaked or flat a distribution is compared to a normal distribution (Ahmed S. Haiba et al. 2019).

\(\text{K}=\frac{{?}_{\text{i}=1}^{\text{n}}{{(\text{x}}_{\text{i}}-\text{m})}^{4}}{(\text{n}-1){\text{s}}^{4}}-3\)

(5)

Following the extraction of the above seven features, ANOVA test (the analysis of variance) was utilized as a feature selection technique to identify significant features at each level for the purpose of detecting different types of defects. Specifically, a one-way ANOVA test was employed to measure the statistical significant of differences between the means of three distinct groups: a healthy sample and two different types of defects. The variance ratio of mean squares between and within groups is the ANOVA F-distribution test statistics. The decision rule for this test is based on P-value. If the estimated P-value is more than a value of 0.05, at α-level = 0.05 with a 95% confidence level, significant differences exist between the tested groups.

2.3 Artificial neural network

The last stage of data processing is the pattern recognition or classification process which identifies the types of PD activities. The present study utilizes an artificial neural network (ANN), commonly referred to as a neural network (NN), as an advanced classification tool for identifying various types of defective insulators. Generally, pattern recognition or classification algorithm includes two stages: training which is known as learning and testing which is known as classification (Ma et al. 2013). The role in the training stage is to model a mathematical relation between the data sets and their respective PD pulses (Ma et al. 2013). While in phase of the testing, new input data points which were not part of the data of training are defined as a single category of PD sources (Ma et al. 2013). ANNs are a type of computational model that takes inspiration from the way biological neural networks in the human brain process input data. Figures (4) displays the structure of an ANN which comprises of three layers: an input layer, one or more hidden layers, and an output layer. In ANNs, a sequence of interconnected units or nodes are used, which are commonly referred to as artificial neurons. Each interlink, like the neurons in a living brain, has the capacity to transmit data or signals to other connections. These connections and neurons typically have a weight that varies as learning progresses. The weight can rise or decrease the signal intensity at a connection link. To get the neuron output, first the weighted sum of all points are taken, then a bias term is added to this sum (Ma et al. 2013). Such weighted sum, which usually called the activation, is then passed via an activation function that generates the output results (Ma et al. 2013).

3.1 Acoustic emission signals

In a laboratory environment, three samples were tested at 40 kV, sinusoidal waveform at 50 Hz. Two of the samples were defective insulators while the other one was healthy. Using a 500 MHz oscilloscope, the shape of the discharge pulse for each test sample was captured based on the amplitude of the discharge pulse in mV vs time or samples. This is referred to as the time domain characteristics of acoustic signals obtained from PD activities. Using the established data acquisition system (DAS), 100 discharge sound pulses were randomly obtained and recorded for each test sample.

The recorded acoustic emission signals are subjected to a Fast Fourier Transform (FFT) tool, which converts them into their frequency domain. This process enables the frequency bands for each test sample to be shown.

Figures (5–7) display one of the collected waves for each of the three test samples in both domains the time and frequency. Visual inspection of the figures reveals differences in waveform shapes between the three test samples. Additionally, the magnitude of the PD sound pulse of the defective samples is higher than that of the healthy sample.

Figure (8) shows PD signatures of all samples in frequency domain. It has been observed from this figure that peaks of the spectrum amplitude have occurred at variable frequencies for each sample. That means there are significant differences between the tested samples.

3.2 Feature extraction and selection

The PD acoustic emission signals collected by the established DAS were recorded and then imported into the MATLAB software program for analysis using the wavelet analyzer tool. Each captured signal is decomposed into 5-levels of approximations, which contain low frequency components, and other 5-levels of details, which contain high frequency components. This process is very difficult to be done manually for each signal of 100 PD pulses and repeated for all cases. So, a MATLAB code for DWT was designed to be used in any time for all sample cases. Figures (9, 10) show the wavelet transform decomposition, in MATLAB environment, of one PD acoustic wave captured with broken insulator in separate and tree decomposition modes respectively. Finally, each PD test signal in the wavelet analyzer is decomposed into five distinct decomposition signals using the below equation:

\({S={a}_{5}+d}_{1}+ {d}_{2}+ {d}_{3}+ {d}_{4}+ {d}_{5}\)

(6)

Where: S is the main signal, a₅ is the 5th level in approximation components, and d₁, d₂, d₃, d₄, d₅ are the 5-levels in details respectively.

The DWT m-file code was developed to intelligently extract the seven features being studied from the decomposed details and approximation signal pulses. For an example, table (1) provides the features that extracted from the 6-decomposed components (a₅, d₁, d₂, d₃, d₄, d₅) of figure (6) in the time domain. These decomposed signals related to a single PD pulse collected from the broken insulator. This study focuses on the relative change in the estimated feature values as an indicator of the type of defect present, rather than on the absolute values of these features. Likewise, an algorithm has been devised to extract and compute the features of the decomposed signals for the 100 PD signal waves acquired from each of the three samples that studied in this work.

Table 1

Extracted features for a single wave of the broken sample
Features	Levels
Features	a₅	d₁	d₂	d₃	d₄	d₅
M	6.63E-04	1.59E-05	3.09E-05	-7.03E-08	-1.76E-05	1.19E-05
V	5.82E-08	1.34E-08	2.15E-08	2.97E-07	9.32E-07	3.37E-08
Max	0.0015	5.50E-04	5.76E-04	0.0055	0.0064	9.60E-04
Min	1.80E-04	-5.07E-04	-6.77E-04	-0.0059	-0.0108	-6.09E-04
m_e	4.97E-07	1.37E-08	2.25E-08	2.96E-07	9.28E-07	3.36E-08
S	0.7710	-0.1474	0.0248	0.5437	-4.866	0.5395
K	3.691	3.531	3.767	78.16	74.07	7.874

ANOVA analysis test is included in the m-file MATLAB code for minimizing the extracted features/levels and selecting the significant features/levels from the seven original features. Table (2) shows the estimated P-value for each feature/level. Each calculated P-value in that table shall be compared to 0.05 and the decision rule here is; if the computed P-values are less than 0.05, the corresponding features/levels are considered significant and could be ignored otherwise. The lowest two values are chosen for each corresponding feature. Consequently, 2-levels are selected from each extracted feature to be the most significant levels. Finally, 14-levels/features, as shown in table (3), are selected and ready to be used as input data to the neural network for identification the type of the defects in the insulators.

The outcomes of the ANOVA test are highly consistent with the findings obtained through the F-value, J-value and B-value criteria, which were discussed in (Ahmed S. Haiba et al. 2019), (Sinaga, Phung, and Blackburn 2014) and (Anjum et al. 2017), respectively.

Table 2

P-values of ANOVA test for each feature/level at 95% confidence
Features	Levels
Features	a₅	d₁	d₂	d₃	d₄	d₅
M	6.88E-52	0.410	0.278	0.518	0.020	1.41E-06
V	0.217	9.77E-47	0.059	0.161	0.0004	4.10E-11
Max	4.39E-25	0.022	1.28E-10	0.0003	2.79E-09	1.25E-12
Min	2.01E-21	0.280	8.82E-10	0.013	0.025	1.78E-12
m_e	5.89E-45	4.55E-45	0.001	0.160	0.0004	1.13E-10
S	0.686	0.084	0.005	0.152	0.060	3.35E-07
K	0.204	3.74E-13	1.59E-11	5.61E-21	0.003	2.28E-10

Table 3

Selected features to be inputs to the classifier
Features	Levels
m	a₅	d₅
V	d₁	d₅
Max	a₅	d₅
Min	a₅	d₅
m_e	a₅	d₁
S	d₂	d₅
K	d₁	d₃

3.3 Artificial Neural Network

According to feature selection analysis, two-levels were selected for each feature of extracted seven features to be as input data to the classifier which known as ANN as shown in table (3). So, input data is a matrix of 14×300 representing 300 samples of 14 elements and the target (output) data is proposed to be 3×300 matrix representing 300 samples of 3 elements as shown in Figure (11). In this classifier, the 300 samples are randomly divided into 70% (210 samples) for training, 15% (45 samples) for validation and 15% (45 samples) for testing. It has been observed from the results that the best performance of validation process is achieved at error 0.0816. This error is calculated as the deviation between the target (proposed output to the ANN) and the actual output of the ANN itself. Further that, the obtained overall recognition rate of these data is 68.119%. This obtained result is considered poor in classification process indicating low accuracy of the designed classifier. In order to overcome this problem, authors have decided to increase the dataset to the ANN by collecting more acoustic emission signals to improve the performance of the classifier. So, experimental process was repeated to obtain 200 acoustic signals for each test sample and data processing techniques were performed on 200 signals for each sample instead of 100 signals.

Now, input data becomes a matrix of 14×600 and the target data is proposed to be a matrix of 3×600. Figures (12) shows that the best validation performance is obtained at 0.00068856 mean squared error (MSE). This small error indicates high performance of the developed classifier. Moreover, Figure (13) shows regression plot of training, validation, testing and the overall recognition rate of the data. It has been seen that the overall recognition rate of the classifier is obtained as 96.034% indicating high accuracy and performance classifier.

On the other hand, from FFT analysis, the first ten peak values of each pulse are assumed to be input data to the ANN. So, the input data is a matrix of 10×600 and the output data is proposed to be the same of the previous classifier 3×600 matrix as shown in Figure (14). By this tool, the best validation performance is achieved at error 0.031838. In addition to that, the overall accuracy recognition rate is obtained as 88.647% as illustrated in figure (15).

In conclusion, it has been observed as cleared in table (4) that the recognition rate obtained from DWT is higher than that obtained from FFT analysis. Moreover, the best validation performance is achieved at lower error in case of wavelet transform than that of FFT. Consequently, DWT analysis is more accurate and recommended to be used in classifying defects in ceramic outdoor insulators.

Table 4

Comparison between DWT and FFT
Technique	Validation performance error	Recognition rate
DWT	0.000688	0.96034
FFT	0.031838	0.88647

Detection and monitoring of PD is a significant diagnosis of defects in outdoor ceramic insulators for ensuring the reliability and preventing any interruption in the electrical power network. As a result, variable defects in ceramic insulators have been studied in this current paper, which are known to be sources of PD activities. A DAS has been established for capturing and recording the acoustic emission signals resulting in PD activities in defective ceramic insulators. DWT has been introduced for extracting features from the captured signals and decomposed the original signals into multi-level signals to get more information about the acoustic signatures for each test sample. ANOVA test has been adopted and used as a feature selection tool. Two levels for each extracted feature have been selected to be the most significant signatures for classification of defective insulators and ready to be the input dataset to the neural network which is known as a classifier. ANN is used in this current work for classifying defects in ceramic insulators. It has been observed from ANN results that the overall recognition rate depends on the number of collected signals. That means higher number of capturing signals, higher obtained recognition rate. Overall recognition rate is obtained as 96.03% from DWT and 88.65% from FFT indicating high accuracy and performance classifier. It is concluded that a successful neural network analysis for the classification of defected ceramic insulators could have important practical applications for safety and reliability of the electrical power transmission and distribution system.

Author contributions

Both authors contributed in this work. A.S.H. suggested the idea of the current work and implemented the experimental section and contributed in analysis of signal processing section. A.E. contributed in designing the ANN for both DWT and FFT. A.S.H. wrote the original draft preparation and A.E. shared completing the paper writing. Both A.S.H. and A.E. reviewed and edited the paper. The tables and figures were prepared by both A.S.H. and A.E. Both authors read and approved the final manuscript. The first author A.S.H. is the corresponding author.

Data Availability

The data sets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Funding: None

Conflict of interest: The authors declare no competing interests.

Ethical approval: The research participants have been informed of the aims of this research and consented to the use of interview data for publication. This article does not contain any studies with human participants or animals performed by any of the authors.

Aghamohammadi, Mohammad Reza, and Morteza Abedi. 2018. “DT Based Intelligent Predictor for out of Step Condition of Generator by Using PMU Data.” International Journal of Electrical Power and Energy Systems 99 (July): 95–106. https://doi.org/10.1016/j.ijepes.2018.01.001.
Ahmed S. Haiba, Ahmed Gad, Soliman M. El-Debeikey, and Mamdouh M. Halawa. 2019. “Statistical Significance of Wavelet Extracted Features in the Condition Monitor- Ing of Ceramic Outdoor Insulators.” In IEEE Electrical Insulation Conference (EIC), 335–432. https://doi.org/10.1109/EIC43217.2019.9046590.
Albano, M., R. T. Waters, P. Charalampidis, H. Griffiths, and A. Haddad. 2016. “Infrared Analysis of Dry-Band Flashover of Silicone Rubber Insulators.” IEEE Transactions on Dielectrics and Electrical Insulation 23 (1): 304–10. https://doi.org/10.1109/TDEI.2015.005026.
Anjum, Shaharyar. 2014. “A Study of the Detection of Defects in Ceramic Insulators Based on Radio Frequency.” University of Waterloo. http://hdl.handle.net/10012/9006.
Anjum, Shaharyar, Shesha Jayaram, Ayman El-Hag, and Ali Naderian Jahromi. 2017. “Detection and Classification of Defects in Ceramic Insulators Using RF Antenna.” IEEE Transactions on Dielectrics and Electrical Insulation 24 (1). https://doi.org/10.1109/TDEI.2016.005867.
Arzamasov, Vadim, Klemens Bohm, and Patrick Jochem. 2018. “Towards Concise Models of Grid Stability.” In 2018 IEEE International Conference on Communications, Control, and Computing Technologies for Smart Grids, SmartGridComm 2018. https://doi.org/10.1109/SmartGridComm.2018.8587498.
Asimakopoulou, G. E., V. T. Kontargyri, G. J. Tsekouras, Ch N. Elias, F. E. Asimakopoulou, and I. A. Stathopulos. 2011. “A Fuzzy Logic Optimization Methodology for the Estimation of the Critical Flashover Voltage on Insulators.” Electric Power Systems Research 81 (2): 580–88. https://doi.org/10.1016/j.epsr.2010.10.024.
Belhouchet, Khaled, Abdelhafid Bayadi, and M Elhadi Bendib. 2019. “Artificial Neural Networks and Genetic Algorithm Modelling and Identification of Arc Parameter in Insulators Flashover Voltage and Leakage Current.” Int. J. Computer Aided Engineering and Technology 11 (1): 1–13. https://doi.org/10.1504/IJCAET.2019.096708.
Chen, Minghua, Qunying Liu, Shuheng Chen, Yicen Liu, Chang Hua Zhang, and Ruihua Liu. 2019. “XGBoost-Based Algorithm Interpretation and Application on Post-Fault Transient Stability Status Prediction of Power System.” IEEE Access 7. https://doi.org/10.1109/ACCESS.2019.2893448.
Cherney, E. A. 1982. “ELECTROMECHANICAL INTEGRITY OF SUSPENSION INSULATORS.” Research Review - Ontario Hydro, no. 5: 19–23.
Cherney, E A, and R D Hooton. 1987. “CEMENT GROWTH FAILURE MECHANISM IN PORCELAIN SUSPENSION INSULATORS.” IEEE Transactions on Power Delivery, no. 1. https://doi.org/10.1109/MPER.1987.5527324.
Chrzan, Krystian L., Wallace L. Vosloo, and Jacobus P. Holtzhausen. 2011. “Leakage Current on Porcelain and Silicone Insulators under Sea or Light Industrial Pollution.” IEEE Transactions on Power Delivery 26 (3). https://doi.org/10.1109/TPWRD.2011.2123470.
Chrzan, Krystian Leonard. 2010. “Leakage Currents on Naturally Contaminated Porcelain and Silicone Insulators.” IEEE Transactions on Power Delivery 25 (2): 904–10. https://doi.org/10.1109/TPWRD.2009.2034665.
Dadashizadeh Samakosh, J., and Mohammad Mirzaie. 2019. “Flash-over Voltage Prediction of Silicone Rubber Insulators under Longitudinal and Fan-Shaped Non-Uniform Pollution Conditions.” Computers and Electrical Engineering 78 (September): 50–62. https://doi.org/10.1016/j.compeleceng.2019.07.004.
Din, A., M. A.M. Piah, A. R. Abdullah, and F. S. Abdullah. 2021. “Classification of Degraded Polymer Insulator Using Support Vector Machine.” In Proceedings of the IEEE International Conference on Properties and Applications of Dielectric Materials. Vol. 2021-July. https://doi.org/10.1109/ICPADM49635.2021.9493974.
Farshad, Mohammad. 2019. “Detection and Classification of Internal Faults in Bipolar HVDC Transmission Lines Based on K-Means Data Description Method.” International Journal of Electrical Power and Energy Systems 104 (January): 615–25. https://doi.org/10.1016/j.ijepes.2018.07.044.
FRĄCZ, Paweł. 2016. “Comparison of Two Methods for Detection of UV Signals Emitted by PD on HV Insulators Made of Porcelain.” PRZEGLĄD ELEKTROTECHNICZNY 1 (8): 104–6. https://doi.org/10.15199/48.2016.08.28.
Ghosh, R., B. Chatterjee, and S. Dalai. 2017. “A Method for the Localization of Partial Discharge Sources Using Partial Discharge Pulse Information from Acoustic Emissions.” IEEE Transactions on Dielectrics and Electrical Insulation 24 (1): 237–45. https://doi.org/10.1109/TDEI.2016.006080.
Gupta, Ankita, Gurunath Gurrala, and P. S. Sastry. 2019. “An Online Power System Stability Monitoring System Using Convolutional Neural Networks.” IEEE Transactions on Power Systems 34 (2). https://doi.org/10.1109/TPWRS.2018.2872505.
He, L., and R. S. Gorur. 2016. “Source Strength Impact Analysis on Insulator Flashover under Contaminated Conditions.” IEEE Transactions on Dielectrics and Electrical Insulation 23 (2): 1005–11. https://doi.org/10.1109/TDEI.2015.005264.
Hussain, Md Rashid, Shady S. Refaat, and Haitham Abu-Rub. 2021. “Overview and Partial Discharge Analysis of Power Transformers: A Literature Review.” IEEE Access 9: 64587–605. https://doi.org/10.1109/ACCESS.2021.3075288.
IEC Standards 60270. 2000. High-Voltage Test Techniques. Partial Discharge Measurements (IEC 60270:2000). Bsi. 3rd ed. https://webstore.iec.ch/publication/1247#additionalinfo.
Lu, Shixiang, Guoying Lin, Hanlin Liu, Chengjin Ye, Huakun Que, and Yi Ding. 2019. “A Weekly Load Data Mining Approach Based on Hidden Markov Model.” IEEE Access 7: 34609–19. https://doi.org/10.1109/ACCESS.2019.2901197.
Ma, Hui, Jeffery Chan, Tapan K. Saha, and Chandima Ekanayake. 2013. “Pattern Recognition Techniques and Their Applications for Automatic Classification of Artificial Partial Discharge Sources.” IEEE Transactions on Dielectrics and Electrical Insulation 20 (2). https://doi.org/10.1109/TDEI.2013.6508749.
Mahdjoubi, Abdelhalim, Boubakeur Zegnini, Mohammed Belkheiri, and Tahar Seghier. 2019. “Fixed Least Squares Support Vector Machines for Flashover Modelling of Outdoor Insulators.” Electric Power Systems Research 173 (August): 29–37. https://doi.org/10.1016/j.epsr.2019.03.010.
Mantach, Sara, Abdulla Lutfi, Hamed Moradi Tavasani, Ahmed Ashraf, Ayman El-Hag, and Behzad Kordi. 2022. “Deep Learning in High Voltage Engineering: A Literature Review.” Energies. MDPI. https://doi.org/10.3390/en15145005.
Polisetty, Satish Kumar. 2019. “Partial Discharge Classification Using Acoustic Signals and Artificial Neural Networks and Its Application in Detection of Defects in Ceramic Insulators.” University of Waterloo. http://hdl.handle.net/10012/14412.
Sahoo, N. C., M. M.A. Salama, and R. Bartnikas. 2005. “Trends in Partial Discharge Pattern Classification: A Survey.” IEEE Transactions on Dielectrics and Electrical Insulation 12 (2). https://doi.org/10.1109/TDEI.2005.1430395.
Salem, Ali A., R. Abd-Rahman, Samir A. Al-Gailani, M. S. Kamarudin, N. A. Othman, and N. A.M. Jamail. 2020. “Artificial Intelligence Techniques for Predicting the Flashover Voltage on Polluted Cup-Pin Insulators.” In Advances in Intelligent Systems and Computing, 1073:362–72. Springer. https://doi.org/10.1007/978-3-030-33582-3_35.
Salem, Ali, Rahisham Abd-Rahman, Waheed Ghanem, Samir Al-Gailani, and Salem Al-Ameri. 2021. “Prediction Flashover Voltage on Polluted Porcelain Insulator Using ANN.” Computers, Materials and Continua 68 (3): 3755–71. https://doi.org/10.32604/cmc.2021.016988.
Salem, Ali Ahmed Ali, Rahisman Abd Rahman, M. S. Kamarudin, Nordiana Azlin Othman, Nor Akmal Mohd Jamail, Haziah Abdul Hamid, and Mohd Taufiq Ishak. 2020. “An Alternative Approaches to Predict Flashover Voltage on Polluted Outdoor Insulators Using Artificial Intelligence Techniques.” Bulletin of Electrical Engineering and Informatics 9 (2). https://doi.org/10.11591/eei.v9i2.1864.
Sampedro, Carlos, Javier Rodriguez-Vazquez, Alejandro Rodriguez-Ramos, Adrian Carrio, and Pascual Campoy. 2019. “Deep Learning-Based System for Automatic Recognition and Diagnosis of Electrical Insulator Strings.” IEEE Access 7. https://doi.org/10.1109/ACCESS.2019.2931144.
Saranya, Kailasam, and Chinnusamy Muniraj. 2016. “A SVM Based Condition Monitoring of Transmission Line Insulators Using PMU for Smart Grid Environment.” Journal of Power and Energy Engineering 04 (03). https://doi.org/10.4236/jpee.2016.43005.
Shi, Chunxue, and Yaping Huang. 2021. “Cap-Count Guided Weakly Supervised Insulator Cap Missing Detection in Aerial Images.” IEEE Sensors Journal 21 (1). https://doi.org/10.1109/JSEN.2020.3012780.
Si, Wen Rong, Jun Hao Li, Da Jian Li, Jing Gang Yang, and Yan Ming Li. 2010. “Investigation of a Comprehensive Identification Method Used in Acoustic Detection System for GIS.” IEEE Transactions on Dielectrics and Electrical Insulation 17 (3). https://doi.org/10.1109/TDEI.2010.5492244.
Sinaga, Herman H., B. T. Phung, and Trevor R. Blackburn. 2014. “Recognition of Single and Multiple Partial Discharge Sources in Transformers Based on Ultra-High Frequency Signals.” IET Generation, Transmission and Distribution 8 (1). https://doi.org/10.1049/iet-gtd.2013.0131.
Stefenon, Stéfano Frizzo, Marcelo Picolotto Corso, Ademir Nied, Fabio Luis Perez, Kin Choong Yow, Gabriel Villarrubia Gonzalez, and Valderi Reis Quietinho Leithardt. 2022. “Classification of Insulators Using Neural Network Based on Computer Vision.” IET Generation, Transmission and Distribution 16 (6). https://doi.org/10.1049/gtd2.12353.
Surya Prasad, P., and B. Prabhakara Rao. 2016. “Review on Machine Vision Based Insulator Inspection Systems for Power Distribution System.” Journal of Engineering Science and Technology Review 9 (5). https://doi.org/10.25103/jestr.095.21.
Tahir Khan Niazi, M., Arshad, Jawad Ahmad, Fehaid Alqahtani, Fatmah A.B. Baotham, and Fadi Abu-Amara. 2020. “Prediction of Critical Flashover Voltage of High Voltage Insulators Leveraging Bootstrap Neural Network.” Electronics (Switzerland) 9 (10). https://doi.org/10.3390/electronics9101620.
Yaacob, M. M., M. A. Alsaedi, J. R. Rashed, A. M. Dakhil, and S. F. Atyah. 2014. “Review on Partial Discharge Detection Techniques Related to High Voltage Power Equipment Using Different Sensors.” Photonic Sensors. Springer Verlag. https://doi.org/10.1007/s13320-014-0146-7.
Zare, Hassan, Yousef Alinejad-Beromi, and Hamid Yaghobi. 2019. “Intelligent Prediction of Out-of-Step Condition on Synchronous Generators Because of Transient Instability Crisis.” International Transactions on Electrical Energy Systems 29 (1). https://doi.org/10.1002/etep.2686.
Zener, Clarence, and H H Wills. 1934. “A Theory of the Electrical Breakdown of Solid Dielectrics.” Z. Physik 145 (855): 523–29. https://doi.org/https://doi.org/10.1098/rspa.1934.0116.
Zhang, Zhijin, Xiaohuan Liu, Xingliang Jiang, Jianlin Hu, and David Wenzhong Gao. 2013. “Study on AC Flashover Performance for Different Types of Porcelain and Glass Insulators with Non-Uniform Pollution.” IEEE Transactions on Power Delivery 28 (3): 1691–98. https://doi.org/10.1109/TPWRD.2013.2245153.
Zhao, Chenlong, Hongwei Mei, Mingxi Zhu, Hanqi Dai, Liming Wang, and Zhicheng Zhou. 2015. “Development of Contamination Flashover Pre-Warning System and Analysis of Operating Experience.” IEEE Transactions on Dielectrics and Electrical Insulation 22 (4). https://doi.org/10.1109/TDEI.2015.004472.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this older preprint version

Read the latest preprint version →

Neural Network Analysis for Classification of Defected Ceramic Insulators

Status:

Version 1

Abstract

Figures

1. Introduction

2. Materials and methods

2.1 Experimental test set up

2.2 feature extraction and selection

2.3 Artificial neural network

3. Results and discussion

3.1 Acoustic emission signals

3.2 Feature extraction and selection

3.3 Artificial Neural Network

4. Conclusion

Declarations

References

Additional Declarations

Status:

Version 1