Classification of Gas Turbine Fault Group Using Machine Learning Methods

The article examines the problem of diagnosing and forecasting faults in gas turbines (GT). The novelty of the study lies in the application of machine learning methods for fault diagnosis-a method that can surpass traditional statistical approaches, enabling more efficient and accurate predictions. The researchers’ contribution also includes the creation of datasets characterizing the operation of GT under the influence of defect groups. The research methodology involved the creation and study of these datasets using GT models, assessing technical conditions, and accounting for defects. This approach allowed for the application of machine learning models in the absence of sufficient real-world data. The main goal of the research was to evaluate the effectiveness of machine learning in assessing the technical condition, classifying defect groups, and estimating the development stages of defects based on thermogasdynamic parameters measured during operation. Key research results demonstrate the potential of models to successfully classify fault groups and predict the stage of defect development. The study showed high accuracy in determining the technical condition of GT, confirming the potential of machine learning in improving diagnostic and maintenance systems. However, the research also underscores the need for further studies using real data to confirm these findings.