Spatial-Temporal Estimation for Nontechnical Losses
MetadataShow full item record
This paper presents a novel method for estimating the spatial distribution in geographical space of the nontechnical losses over time. The method progresses in two stages: in the first stage, a generalized additive model is used to generate a map of current loss probabilities. The second stage employs the Markov chain to generate a map that indicates possible future changes in loss probabilities. The method yields an assessment of the location of the nontechnical losses now and in the future at the city subarea level, even indicating the variables that have greater statistical correlation with the nontechnical losses. We apply the method to a city with approximately 81 000 consumers, and the results are compared with those obtained through inspections carried out by a Brazilian power utility. The detection rate surpasses 78% in inspected subareas. The method we propose offers improved estimation of distribution of the nontechnical losses in urban regions.