Use of ANN for identification of consumers with irregular electrical installations

This paper presents the results obtained with artificial intelligence techniques applied to identify consumers with irregular installations in the electrical distribution system. A multilayer perceptron artificial neural network was used, with 10 neurons in the hidden layer, sigmoid activation function and backpropagation algorithm for training. Information typically available for electric power distribution companies were used as input data. The results obtained showed a success rate of up to 67.5%, which can be considered high when compared to analytical techniques commonly used by power distribution companies, whose success rate is between 25% and 30%. By the end it is possible to conclude that the use of the presented technic may be very useful for power distribution companies while reducing nontechnical losses of electric energy.