Non-technical loss (NTL) is the major source of energy loss in distribution networks. Since the direct calculation of NTL is impossible, it is mostly being obtained by calculating the difference between total loss and technical loss (TL). This is rather not possible without adequate equipment for measuring the TL. However, time-worn distribution networks with high limitations and lack of equipment could identify the regions with a high share of NTL to reduce/eradicate the sources of NTL. This strategy can increase efficiency and reduce costs compared to the one-by-one inspection of customers. This paper presents a novel NTL detection procedure based on load estimation in highly limited distribution networks. First, the low-voltage (LV) distribution network will be divided into a few smaller hypothetical sub-networks through a fuzzy c-means (FCM) clustering technique. Then, a meter placement based on the maximum likelihood criterion is deployed to find the best places for installing the existing meters in each cluster. In the next stage, a linear system of equations is utilized to extract the pattern load profiles (PLPs) related to each class of customers for each hypothetical sub-network separately. These PLPs will be used to estimate the transformers’ load in all LV feeders based on their share in the total consumption of the entire network. Finally, NTL possibility per capita will be achieved for each LV feeder by defining an index that estimates the degree of NTL in those areas.

Keywords