Development of a Device for Diagnosing Unbalanced Short Circuits in 10 kV Overhead Power Lines

The article addresses the development, based on the Arduino platform, of an intelligent microprocessor device intended to determine the type and place of an unbalanced short circuit fault in 10 kV overhead power lines. The device analyzes phase currents and voltages, as well as their zero and negative sequence components. An algorithm based on a pre-trained neural network of the multilayer perceptron type (MLP) has been implemented, which is capable of classifying, in real time, 12 different types of short circuit faults, including single-phase, two-phase and three-phase short circuits, as well as ground faults. The neural network was trained, based on the data obtained, in the MATLAB/Simulink environment, followed by integrating the algorithm into the Arduino environment. Laboratory tests of the mock-up with simulation of emergency modes have confirmed high accuracy (more than 95%), high response speed (the response time is less than 80 ms), and stability of the device performance. The hardware solution features compactness, low power consumption, simple configuration, and the ability to visualize the result. The proposed device can be effectively integrated into local relay protection and monitoring systems. Its application is especially relevant in 6-10 kV distribution networks under the conditions of limited financing, remoteness of facilities and the need to improve the power supply reliability.