Machine Learning Approach for Classifying Power Outage in Secondary Electric Distribution Network

Abstract

Power outage is the problem that hinders social and economic development especially for developing countries like Tanzania. Frequent power outages damage electric equipment, and negatively affect the industrial production process. Power outages cannot be completely eradicated due to uncontrolled cause like natural calamities but technical challenges can be managed and hence reducing power outages. The existing manual methods used to locate power outage like customer calls is inefficient and time consuming. On the other hand, modern method like the Advanced Metering Infrastructure (AMI) still faces a challenge in effectively classifying power line outage due to the nature of imbalanced datasets. Therefore, there is a need to develop a Machine Learning (ML) model to accurately classify power line outage. In this study, machine learning models are constructed from ensemble algorithms and tested using outage AMI data from 2012 to 2019 with 2 hours interval records. We propose the following ensemble-based machine learning approach to enhance classification; data sampling, algorithm weighting and finally ensembling. Results show that the Hybrid Stacking Ensemble Classifier (HSEC) model outperforms the others by accuracy of 0.981 G-mean, followed by Extra tree with accuracy of 0.964 G-mean. This model can be used in power line outage classification in any Secondary Electrical Distribution Network (SEDN). This study can be extended to locate power outage to household. 

Keywords: Imbalanced dataset, Power outage location, Stacking ensemble classifier