Identification of unauthorized electricity based on the consumption data of subscribers with the help of artificial intelligence

One of the most important challenges in electricity distribution companies is the use of unauthorized electricity or electricity theft, which is an important factor in reducing the income of electricity distribution companies. The number of subscribers as well as the large amount of data related to their consumption is increasing rapidly, and the traditional methods to detect suspicious subscribers are difficult, expensive and in some cases almost impossible. In this article, we will deal with five classification models including decision tree, support vector machine, Bayesian rule, neural network and k-nearest neighbor classification on 300 real data of electricity affairs in order to detect subscribers with unauthorized electricity, the results of which according to the real input data in the diagnosis The best smart model will be very reliable to identify subscribers with unauthorized electricity.