Developing a Digital Model of an Electricity Consumer using Deep Learning

The paper deals with the challenge of creating a digital model of an electricity consumer using Deep Learning technologies. The key benefits of the digital model include the forecasts of energy consumption and market electricity prices, as well as the prediction of peak power hours. To implement the digital model of the energy consumer, a Python application was developed, with the popular Keras library for Deep Learning. The Yaroslavl State Technical University was taken as a sample consumer of electricity. The paper also presents the results of numerical experiments on forecasting energy consumption and electricity prices, analysis of the forecast quality in relation to the parameters of the neural network and the applied training algorithms. The tested potential for predicting peak power hours so far provides low accuracy of results, but could have a great practical impact on reducing energy costs.