Application of machine learning methods to improve the efficiency of a gas turbine unit of a compressor station

The article discusses machine learning methods to improve the efficiency of a gas turbine unit of a compressor station. It is noted that when preparing and analyzing data to improve the efficiency of a gas turbine unit of a compressor station when forecasting the amount of used and generated electricity using machine learning methods, as well as assessing the importance and impact of the period of day, month, year, temperature, air humidity, atmospheric pressure and other features on forecasting. The data set used in this work contains information on the use and generation of electricity, as well as weather indicators for 11 months with a data recording period of 1 minute. It was revealed that data processing within the framework of statistical methods of information processing is most effective determination coefficient in assessing the accuracy of forecasting. It is concluded that the machine learning method greatly increases the efficiency of the gas turbine unit of the compressor station for solving the problem of forecasting the volumes of generation and consumption of electric energy in the MicroGrid based on the analysis of a large number of different parameters. At the same time, the use of preliminary data processing allows to increase the accuracy of forecasting by 2 to 25% for the considered dataset.