Perdas não técnicas em irrigantes da região rizicultora do Rio Grande do Sul
Keywords: artificial intelligence, irrigating consumers, irregular consumption, non-technical losses
AbstractNon-technical losses (NTL) are a component of energy losses associated with energy theft or irregular billed energy measurements, including in rural areas. We developed a methodology that address the specificities of consumption with irrigation for rice crops, and use information from historical data of consumer units and associating with the size of the crop, energy demand, local topography and other pertinent information. Such information influenced and helped in the grouping (clustering) of consumer units composing their consumption profiles and cultivation areas. Clustering facilitated the association of rice growing areas with the corresponding consumer unit, and this also facilitated the definition of some rules for classifying consumer units with NTL potential. This study employs a large mass of input data, not only the energy consumption of the irrigation consumer units, but also phenological characteristics of the plant, type of irrigation, meteorological information, cultivated area and soil permeability. Based on the analysis of consumption historic from 2018 to 2021, the proposed artificial intelligence system generated as a result a list for planning inspection during the RGE-Sul’s 2021/2022 harvest. The result of inspections may provide an important return as to the assertiveness of the choice. This result may also suggest a possible restructuring of the methodology in the use of the concepts of RNA, KNN and Random Forest. The methodology for automated analysis reached results with high adherence with the manual process currently applied.