Search

Browse

• Communities
• Authors
• Year
• Titles
• Subjects

My Space

• Login

Development and application of advanced artificial intelligence methods in the detection of power theft in smart electric power distribution networks with strong renewable energy penetration

Blazakis Konstantinos

Full record

---

Summary

Modern power systems, particularly distribution networks, are currently dealing with a number of challenging issues brought on by advancements in technology and variations in consumer demands. Nevertheless, as power grid components become more interconnected, so does their susceptibility to fraud, cyberattacks, and software bugs. Non-technical losses (NTLs), which include indicative electricity theft, broken or malfunctioning meters, and intentionally arranged misleading meter readings, pose a threat to modern power systems, which are essential infrastructural assets. NTLs are a major concern in emerging nations, as they can account for as much as 40% of all electricity distributed. It is anticipated that NTLs cause yearly global utility expenses of about 100 billion USD. It is consequently imperative for utilities and authorities to reduce NTLs in order to boost income, profit, and grid reliability. Electricity theft is a widespread problem with significant negative economic, social, and financial impacts. This illegal practice can have serious consequences for both utilities and society as a whole. Benefits of power theft detection include reduced financial losses for utility companies, fairer electricity pricing for consumers, and enhanced grid reliability. Additionally, it can help reduce the environmental impact of illegal power consumption. The goal of this PhD thesis is to propose solutions in modern distribution networks by creating innovative concepts and algorithms for identifying power theft using smart meter data. Moreover, this thesis examines the crucial role of power output forecasting for renewable energy sources (RES) in addressing power theft in modern electricity distribution systems. An extensive analysis of NTLs detection techniques is presented, classifying techniques for detecting electricity theft based on the different types of algorithms employed. Additionally, in order to comprehend the various problems with NTLs detection systems, analysis of parameters is carried out, including data requirements, extracted features, performance metrics, response time, etc. Most NTLs detection systems make use of data analytics and machine learning technologies as their main operation algorithms. This thesis primarily analyzes the proposed innovative techniques and the issues that arise during their development and application.

Available Files

Services