Metodologia para identificação de perdas não técnicas em unidades consumidoras localizadas em áreas rurais com cultivo de arroz irrigado

Abstract

Non-technical losses are one of the biggest problems faced by companies in the energy sector, since they directly affect the financial revenues of concessionaires, the security of electrical systems and the quality of electricity supply to consumers. They are associated with errors in readings and measurements, clandestine connections to the secondary distribution network and fraud in meters. In rural feeders, the problem of non-technical losses becomes more complex, as the large extensions of the distribution networks and the presence of consumers located in remote or difficult-to-access regions make it difficult for inspection teams to investigation. Furthermore, rural consumers are extremely relevant for concessionaires, as they represent a significant portion of electricity consumption, due to irrigation systems installed with pumps and motors with high active power and used practically continuously during the crop period. This work presents a proposal for the identification of non-technical losses in consumers located in rural areas with irrigated rice cultivation. The methodology developed in this study based on the correlation of electricity consumption patterns and the characteristics of planting areas and irrigation systems. The methodology employs in the structure of its algorithm, the k-Means and Random Forest methods, for the development of the data clustering and classification stages, respectively, being used as inputs, monthly records of billed energy from a set of rural consumers areas. Furthermore, the algorithm has the capacity to estimate the consumption of electricity, based on referential information of some variables that make up the irrigation systems and the cultivated planted areas for a certain crop. The validation of the proposed methodology was accomplished in a group of seven rural consumers located in the West Frontier Region of the state of Rio Grande do Sul. In order to verify for possible irregularities, indicators referring to the class and the comparison between actual and estimated consumptions for each consumer were evaluated. The results demonstrated the proper operation of the model, indicating five suspected cases of occurrences of non-technical losses.