Analysis of the Condition of a Gas Turbine System. 3. Using Machine Learning

A method based on machine learning is proposed for assessing the condition of a gas turbine system for transporting natural gas. The initial data are archived gas-dynamic parameters recorded by the automatic control system. The initial data set is created by determining the power from the change in enthalpy of the natural gas before and beyond the pump. The software is written in Python. The Sci-kit library is used for the machine learning models. The quality of prediction is assessed in terms of the mean absolute error (%). The quality of prediction of the machine learning models is assessed with different parameter sets and sample sizes. Recommendations are made regarding the use of the models.