Application of Neural Networks for Prediction of Insulation Condition in Networks with Isolated Neutral

Abstract

Electrical insulation is an important parameter of the structural element of electrical equipment, provides reliability and durability of electrical networks with an isolated neutral and safe use of electricity in underground conditions. This article solves the problem of predicting changes in insulation resistance of electrical equipment, taking into account changes in external factors such as humidity and air temperature. An example of using a conventional neural network such as MLP did not give an acceptable result. To solve this problem, a recurrent neural network with long-term memory type LSTM was proposed. Its structure was developed. The simulation results showed that the neural network LSTM successfully coped with the task of predicting changes in insulation resistance, taking into account changing environmental factors. The prediction error for the one-hour interval was 0.052. The proposed idea of continuous monitoring of the insulation resistance with the prediction of its changes for a certain period will avoid sudden failures of electrical components, electric shock and fire caused by a decrease in insulation resistance below the critical value. The forecast of insulation resistance will allow, in cases of delay of actuation of the device leakage at high rate of change of the insulation resistance, increasing the efficiency of mining equipment through the reduction of damage from sudden shutdowns of the equipment underground mining machines in the course of their normal operation.