Publication: Evaluación de la capacidad de predicción de la evolución de árboles eléctricos en resina epóxica mediante el uso de técnicas de inteligencia artificial
Loading...
Date
2023-09-08
Authors
Salas Henríquez, Franco Eduardo
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Failures in electrical equipment due to insulation deterioration lead to interruption and restoration times in the operation of the electrical system, resulting in associated economic losses. One of the aging mechanisms in insulation are electrical trees, cavities in the insulation that grow in shapes similar to botanical trees. Monitoring electrical variables associated with electrical trees, such as partial discharges, would enable preventive actions against potential failures in electrical equipment. In this study, the predictive capability of the evolution of electrical trees was evaluated using artificial intelligence techniques. Data from partial discharges and lengths of electrical trees were collected through tests at 15 kV AC, 15 kV AC +
15 kV DC, and 15 kV AC-15 kV DC in epoxy resin samples. Prediction was performed using random forest and neural network models. The data were labeled with the deterioration state of each electrical tree based on its advancement, determined by inspecting its lengths and partial discharge characteristics. The training data were based on parameters from the IEC 60270 standard and phase-resolved partial discharge patterns. Prediction capabilities for the prediction of the states of electrical trees were achieved with approximately 90% accuracy for both random
forest and neural network models. Additionally, it was observed that the phase-resolved partial discharge analysis within the feature set is of greater importance than the parameters of IEC 60270
Description
Keywords
Sistemas de energía eléctrica , Fallas eléctricas , Inteligencia Artificial