Classification of electricity customers based on real consumption values using data mining and machine learning techniques and its corresponding applications


Thesis Type: Postgraduate

Institution Of The Thesis: Orta Doğu Teknik Üniversitesi, Faculty of Engineering, Department of Computer Engineering, Turkey

Approval Date: 2013

Student: MUHAMMET TUĞBERK İŞYAPAR

Supervisor: FERDA NUR ALPASLAN

Abstract:

Classifying electricity customers based on real power consumptions has been particularly important in the last decade following the liberalization of the electricity markets in numerous countries and ubiquitous use of Automatic Meter Reading devices that collect consumption data at hourly intervals. Collection of vast amounts of consumption data has made it possible to identify customer classes by clustering. Classification of customer load profiles provides the basis of several applications offering solutions to encountered problems in the area. Dedicated tariff design, load forecasting and fraud detection are among deployable applications. Evaluated scalability of the methods reveals an implicit estimation about the size of local regions to which the framework can be applied within reasonable time. Forming further relations among local regions holding consumption data may be useful in developing national energy policies. This study covers a comprehensive scope of the recent work in the problem domain and puts forward foundations of corresponding applications by processing real consumption data of customers of an electricity distribution company in Turkey.