|
This article reveals the features of the practical application of energy consumption forecasting algorithms at mining enterprises. To assess the productivity, reliability and efficiency of the selected forecasting algorithms (support vector machine, artificial neural network, decision tree, random forest), four parameters were identified: correlation coefficient, mean absolute error, mean square error, relative square error. The results of this study showed that the forecast model based on the random forest method has the best productivity, reliability and efficiency, through which it is possible to evaluate the behavior of experimental energy loads and simulate scenarios in order to identify the best values of key performance indicators of the ongoing production business process.
Keywords:forecasting algorithm, energy consumption, machine learning, mining, energy management, energy efficiency, productivity and reliability
|