Estimation of Cable Lines Insulating Materials Resource Using Multistage Neural Network Forecasting Method

Abstract

The use of artificial neural networks (ANN) for predicting the temperature regime of a power cable line (PCL) core is considered. The relevance of the task of creating neural networks for assessing the throughput, calculating and forecasting the PCL core temperature in real time based on the data from a temperature monitoring system, taking into account changes in the current load of the line and external conditions of heat removal, is substantiated. When analyzing the data, it was determined that the maximum deviation of the neural network (NN) data from the data of the training sample was less than 3% and 5.6% for experimental data, which is an acceptable result. It was established that ANN can be used to make a forecast of the cable core temperature regime with a given accuracy of the core temperature. The comparison of the predicted values with the actual ones allows us to talk about the adequacy of the selected network model and its applicability in practice for reliable operation of the cable power supply system of consumers. The model allows evaluating the insulation current state and predicting the PCL residual life. The results analysis showed that the more aged PCL insulation material, the greater the temperature difference between the original and aged sample. This is due to the loss of electrical insulation properties of the material due to the accumulation of the destroyed structure fragments, containing, in an increasing amount, inclusions of pure carbon and other conductive inclusions.