High Voltage Insulators Condition Analysis using Convolutional Neural Network

Serikbay, Arailym; Bagheri, Mehdi; Zollanvari, Amin

doi:10.1109/eeeic/icpseurope51590.2021.9584624

Cited by 3 publications

(5 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Another limitation in the present work, which also extends to some of our previous studies [1,[55][56], is the utility of artificial contamination. The high-voltage laboratory condition was utilized to emulate the insulators' contamination (in this study specifically raindrops), and all data in this study were collected in a laboratory environment.…”

Section: Discussionmentioning

confidence: 84%

“…Therefore, a new insulator dataset (see Fig. 2) composed of 239 images of ceramic and rubber-post insulators was collected (see [1] for details) in HV and Power System Laboratory. The findings of this research are relevant to both rubber-post and ceramic insulators and can be applied to both materials.…”

Section: A Test Objects Main Goals and Data Collectionmentioning

confidence: 99%

“…For training classifiers, the dataset was split into three sets (70% training, 15% test and 15% validation). Data augmentation techniques such as rotation (100), rescaling (1/255), width and height shift range (10%), fill mode (nearest), shear (0.1 rad) and zoom mode (10%) were applied to training data to create 761 additional images [1]. Some samples of augmented insulator images are shown in Fig.…”

Section: A Test Objects Main Goals and Data Collectionmentioning

confidence: 99%

“…Therefore, it is necessary to take alleviating actions to increase power distribution and transmission work efficiency. Researchers have conducted encouraging investigations regarding contaminated insulator condition analysis using different approaches such as coating [1][2][3][4]39,40], creepage distance extension [5], and washing [6]. The most used preventive action is covering the insulator surface with hydrophobic material, the so-called insulation protective coating.…”

Section: Introductionmentioning

confidence: 99%

“…With the advent of deep learning techniques and their numerous applications in engineering, previously a convolutional neural network (CNN) was proposed as the backbone data analysis used to classify contaminated insulator surfaces [1]. Specifically, a brute-force model selection method within a limited hyperparameter space enabled the classification of contaminated insulator surfaces.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Stacked Ensemble Deep Learning for Outdoor Insulator Surface Condition Classification: A Profound Study on Water Droplets

Serikbay,

Bagheri,

Zollanvari

2023

IEEE Access

Self Cite

View full text Add to dashboard Cite

Insulators are vital protection and isolation barriers used in power transmission systems. To prevent unexpected failures caused by severe weather conditions, it is important to develop intelligent insulator fault prognosis systems. To this end, this study has focused to examine and analyze normal (clean) and contaminated (wet) insulators through a new technique that can act as a catalyst for possible future solutions to the data-driven classification of insulator surface conditions. In particular, a novel stacked ensemble learning based on six pretrained deep convolutional neural networks (CNNs), which are used as level-0 generalizers in the stacked structure, is proposed. As the level-1 generalizer that is used in the stacked structure, a majority vote is considered among selected subsets of level-0 generalizers. In other words, training the level-1 generalizer is equivalent to selecting the best subset of pretrained CNNs for classifying a clean insulator surface from those sprayed with water droplets. Since a wetted insulator surface can enhance the electric field intensity and may lead to flashover when combined with other contaminants (dust, soil, cement, etc.), water droplets are considered a type of contamination in our study. Considering a tradeoff between performance and model complexity in training the level-1 generalizer points to two combinations of pretrained CNNs, namely, EfficientNetB2-ResNet50-Xception, and MobileNet-DenseNet121-Xception. Our empirical results show that (i) both combinations lead to better performance when compared with individual pretrained models, and (ii) the latter combination leads to a considerably lower complexity (~39% less parameters) at the expense of ~9% reduction in accuracy.INDEX TERMS deep learning, high-voltage insulators, pretrained CNN architectures, stacked ensemble learning.

show abstract

Section: Discussionmentioning

confidence: 84%

Section: A Test Objects Main Goals and Data Collectionmentioning

confidence: 99%

Section: A Test Objects Main Goals and Data Collectionmentioning

confidence: 99%