Big data is the foundation of modern energy management systems. There are two energy consumption models where systems are one of the consumers with intelligent equipment: static and dynamic. The dynamic model uses a two-tariff closed-loop accounting scheme, which implies changes in tariffs based on the analysis of current consumption. The results of an experimental study of both models using energy consumption data are presented. The influence of the number of such devices on the possibility of achieving uniform consumption when using the second model is shown. Usage of machine learning for generation a consumption forecast based on time series is shown