An Evaluation of Alternative Techniques for Monitoring Insulator Pollution (original) (raw)
Related papers
Classification of Contaminated Insulators Using k-Nearest Neighbors Based on Computer Vision
The contamination on the insulators may increase its surface conductivity and, as a consequence, electrical discharges occur more frequently, which can lead to interruptions in the power supply. To maintain reliability in the electrical distribution power system, components that have lost their insulating properties must be replaced. Identifying the components that need maintenance, is a difficult task as there are several levels of contamination that are hardly noticed during inspections. To improve the quality of inspections, this paper proposes to use the k-nearest neighbours (k-NN) to classify the levels of insulator contamination, based on the image of insulators at various levels of contamination simulated in the laboratory. Using computer vision features such as mean, variance, asymmetry, kurtosis, energy, and entropy are used for training the k-NN. To assess the robustness of the proposed approach, statistical analysis and a comparative assessment with well-consolidated algo...
2011
This work presents improvements in an insulation pollution estimation system. Such system proved functional under laboratory conditions, and extracts spectral information from the ultrasonic noise emitted by the corona discharges that occur in electric equipment. Then, it correlates the compacted spectral information with degrees of pollution previously defined. To achieve the classification, Auto-Associative Artificial Neural Networks are employed. Eight different insulation structures, from eight equipments were chosen for inspection. An inspection routine was defined, so all equipments had their ultrasonic sounds recorded in the same order, and from about the same point. The results show that the method suffers influence from the humidity, since it changes the ultrasound spectrum. On the other hand, this effect can be minimized if the database that is used to train the Auto-Associative Artificial Neural Networks is sufficiently diversified.
2018
Damaged insulators may affect the mechanical and electrical performance of an electric power grid, which can lead to the flow of leakage currents through the line supports. This increases electrical losses and voltage drop in the power grid. It also poses a risk to human safety. Thus, it is crucial to monitor and inspect the condition of insulators to detect degradation or damage. However, the traditional method of inspection is inadequate in meeting the growth and development of the present electric power grid. Hence an automated system such as the computer vision method is presently being explored as a means to resolve this crisis safely, speedily and accurately. This paper presents a method that distinguishes between defectuous and non-defectuous power-line insulators. Active Contour model is applied for insulator segmentation in order to determine insulator region of interest. Local binary pattern is used for feature extraction from the insulator region of interest which is then...
In the present work, tracking phenomenon in high density polyethylene material has been studied under AC voltage, with ammonium chloride as a contaminant. It is observed that the tracking time depends on both conductivity and flow rate of the contaminants. The surface conditions of the insulation structure such as normal conditions and surface discharges or tracking were characterized by using the leakage current measurement, utilizing the multi resolution signal decomposition technique. The process of identification of surface condition of the insulation structure was automated using Artificial Neural Network. #
Contaminated insulators in polluted areas may lead to flashovers if they are not cleaned periodically. Flashover in most cases leads to lengthy service outages; hence it has a considerable impact on power system reliability. In this paper, a combined image processing artificial neural networks algorithm has been developed for the estimation of contamination level in high voltage insulators. Image processing has been used to extract needed features form images captured by digital cameras. The type of features which is considered is the “histogram based statistical feature” such as mean, variance, skewness, kurtosis, energy and normalized histogram error. On the other hand, using these features, a neural network has been successfully designed to correlate the insulator captured image and the contamination level. Testing of the developed algorithm showed a high successful rate in estimating the contamination levels of unseen insulators.
Machine learning techniques for power transformer insulation diagnosis
2011
Power transformers are one of the most critical equipments in electricity network. A number of techniques such as dissolved gas analysis (DGA), polarization and depolarization currents (PDC) measurement and frequency domain spectroscopy (FDS) have been adopted across utilities for transformer insulation diagnosis. However, there are still considerable challenges remaining in interpreting measured data of these techniques. This paper develops machine learning algorithms, which utilise archived data for making insulation diagnosis on the transformer of interest. Analysis and interpretation of field test data are presented in the paper.
Automated Statistical Recognition of Partial Discharges in Insulation Systems
Perspectives in Neural Computing, 1998
We present the development of the successive stages of a statistical system for a classical diagnosis problem in the domain of power systems. It is shown how the different steps may be designed in a sound statistical way in order to develop a complete and efficient diagnosis system which provides the end user with a maximum of information about the system behaviour.
Feature extraction and classification for detecting the thermal faults in electrical installations
Measurement, 2014
This paper proposed an effort to investigate the suitability of input features and classifier for identifying thermal faults within electrical installations. The features are extracted from the thermal images of electrical equipment and classified using a multilayered perceptron (MLP) artificial neural network and support vector machine (SVM). In the experiments, the classification performances from various input features are evaluated. The commonly used classification performance indices, including sensitivity, specificity, accuracy, area under curve (AUC), receiver operating characteristic (ROC) and F-score are employed to identify the most suitable input feature as well as the best configuration of classifiers. The experimental results demonstrate that the combination of features set T max , T delta and DT bg produce the best input feature for thermal fault detection. In addition, the implementation of SVM using radial basis kernel function (RBF) produces slightly better performance than the MLP artificial neural network.
2014 IEEE Conference on Electrical Insulation and Dielectric Phenomena (CEIDP), 2014
This paper expose a novel algorithm to monitor and classify the pollution severity level of insulators based on Leakage Current (LC) waveforms investigation. For this purpose, LC waveforms acquisition is firstly carried out on a plane insulator model under various saline pollution conductivities. Then, LC is investigated and decomposed in five levels using the Wavelet Packet Transform (WPT). Two, four, eight, sixteen and thirty-two coefficients are obtained from the first level to the fifth one respectively. Next, Standard Deviation-Multi Resolution Analysis (STD-MRA) is used to extract features from WPT coefficients. It is noted that the higher the pollution severity, the higher the STD value. Finally, STD values are used as inputs to three well known classification methods (K-Nearest Neighbors, Naïve Bayes and Support Vector Machines), while the sole output is the pollution conductivity value. Results announce that the higher the decomposition level, the better the classification performance. WPT methodology is presented as a highly efficient technique for LC investigation and classification.
Effective insulator maintenance scheduling using artificial neural networks
Iet Generation Transmission & Distribution, 2010
One of the most frequent causes of failure of overhead high-and medium-voltage transmission and distribution lines is contamination of the insulators with diverse substances such as saline and industrial substances. The contamination mechanically degrades the insulators and affects the electrical characteristics of the insulating material, leading to flashovers. Periodic maintenance of insulators can reduce or even prevent the outages caused by contamination. The maintenance scheduling is planned based either on measurements, which are quite expensive and time consuming processes or on experience, a definitely inaccurate process. The current work presents a new approach for the assessment of contamination of insulators on the basis of artificial intelligence and, more specifically, artificial neural networks (ANNs). An ANN model is defined and when applied on operating voltage insulators it presented results similar to experimental results. The proposed approach can be useful in the work of electrical maintenance engineers, reducing the time and cost of insulator maintenance. † 1 -5 hidden layers † 2 -100 neurons in each hidden layer † gradient descent † conjugate gradient † quasi-Newton † Levenberg-Marquardt † random order incremental † hyperbolic tangent sigmoid † logarithmic sigmoid † hard-limit † competitive † linear