Muhammad Saeed - School of Electrical Engineering, University Technology Malaysia, Skudai, Johor Bahru ۸۱۳۱۰, Malaysia
Mohd. Wazir Mustafa - School of Electrical Engineering, University Technology Malaysia, Skudai, Johor Bahru ۸۱۳۱۰, Malaysia
Usman Sheikh - School of Electrical Engineering, University Technology Malaysia, Skudai, Johor Bahru ۸۱۳۱۰, Malaysia
Attaullah Khidrani - School of Electrical Engineering, University Technology Malaysia, Skudai, Johor Bahru ۸۱۳۱۰, Malaysia
Mohd Norzali Haji Mohd - School of Electrical Engineering, University Technology Malaysia, Skudai, Johor Bahru ۸۱۳۱۰, Malaysia

Electricity theft and fraud in billing are the primary concerns for Distribution System Operators (DSO). It is estimated that billions of dollars are lost each year due to these illegal activities. DSOs around the world, especially in underdeveloped countries, are still utilizing conventional time consuming and inefficient methods for Non-Technical Loss (NTL) detection. This research work attempts to solve the mentioned problems by developing an efficient energy theft detection model to identify the fraudster customers in a power distribution system. The key motivation for the current study is to assist the DSOs for their campaign against energy theft. The proposed method initially utilizes the monthly consumption data of energy customers, obtained from Multan Electric Power Company (MEPCO) Pakistan, to segregate the honest and the fraudulent customers. The Bagged Chi-square Automatic Interaction Detection (CHAID) Decision Tree (DT) algorithm is used to classify the honest and fraudster consumers. Furthermore, based on the mentioned metrics, the performance superiority of the Bagged CHAID-based NTL detection method is validated by comparing its efficacy with that of few well-known state-of-the-art machine learning algorithms such as Support Vector Machine (SVM), Artificial Neural Network (ANN) Logistic Regression (LR), Bayesian Network (BN) and Discriminant Analysis. The proposed NTL detection method provides an Accuracy of ۸۶.۳۵% and Area Under Curve (AUC) of ۰.۹۲۷, respectively, which are significantly higher than that of the same for the mentioned algorithms.

Electricity theft, fraud billing, Non-Technical Loss, Chi-square Automatic Interaction Detection