Pattern Recognition using Support Vector Machines as a Solution for Non-Technical Losses in Electricity Distribution Industry
Azubuike N. Aniedu1, Hyacinth C. Inyiama2, Augustine C. O. Azubogu3, Sandra C. Nwokoye4
1Azubuike N. Aniedu*, Department of Electronic and Computer Engineering, Nnamdi Azikiwe University, Awka, Nigeria.
2Hyacinth C. Inyiama, Department of Electronic and Computer Engineering, Nnamdi Azikiwe University, Awka, Nigeria.
3Augustine C. O. Azubogu, Department of Electronic and Computer Engineering, Nnamdi Azikiwe University, Awka, Nigeria.
4Sandra C. Nwokoye, Department of Electronic and Computer Engineering, Nnamdi Azikiwe University, Awka, Nigeria.
Manuscript received on March 09, 2021. | Revised Manuscript received on March 15, 2021. | Manuscript published on March 30, 2021. | PP: 1-8 | Volume-7 Issue-2, March 2021. | Retrieval Number: 100.1/ijisme.B1280037221 | DOI: 10.35940/ijisme.B1280.037221
Open Access | Ethics and Policies | Cite | Mendeley
© The Authors. Published By: Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Contending with Non-Technical Losses (NTL) is a major problem for electricity utility companies. Hence providing a lasting solution to this menace motivates this and many more research work in the electricity sector in recent times. Non-technical losses are classed under losses incurred by the electricity utility companies in terms of energy used but not billed due to activities of users or malfunction of metering equipment. This paper therefore is aimed at proffering a solution to this problem by first detecting such loopholes via the analysis of consumers’ consumption pattern leveraging Machine learning (ML) techniques. Support vector machine classifier was chosen and used for classifying the customers’ energy consumption data, training the system and also for performing predictive analysis for the given dataset after a careful survey of a number of machine learning classifiers. A classification accuracy (and subsequently, class prediction) of 79.46% % was achieved using this technique. It has been shown, through this research work, that fraud detection in Electricity monitoring, and hence a solution to non-technical losses can be achieved using the right combinations of Machine Learning techniques in conjunction with AMI technology.
Keywords: Clustering, classification and association rules, Correlation and regression analysis, Machine learning