A new method for estimating the losses of low voltage feeders based on deep neural networks in the space of information deficit

Electricity losses as an integral part of power delivery systems have always been recognized by electricity industry actors as a measure of how the network managed. In a way, the high losses of the network show the weak managers and the inefficiency of the programs implemented in this network. The exact amount of losses is measured by load flow; But distribution networks, especially low voltage networks, have many information deficiencies due to the expansion and lack of development of information systems; This has left no measure of losses available in this section. Therefore, the electricity network regulator and distribution companies do not have a good view to assessing the management status of the network, the efficiency of the implemented projects, future planning, and budget allocation. In this paper, by using statistical analysis and high responsiveness of deep neural networks to nonlinear problems, a new method for estimating losses in the weak space of information systems is presented.