A Novel Intelligent Method for Fault Diagnosis of Steam Turbines Based on T-SNE and XGBoost

Since failure of steam turbines occurs frequently and can causes huge losses for thermal plants, it is important to identify a fault in advance. A novel clustering diagnosis method based on t-distribution stochastic neighborhood embedding (t-SNE) extreme gradient boosting (XGBoost) proposed this paper. First, the t-SNE algorithm was used map high-dimensional data low-dimensional space; K-means performed space distinguish from normal data. Then, imbalance problem processed by synthetic minority over-sampling technique (SMOTE) obtain turbine characteristic set with labels. Finally, XGBoost solve multi-classification problem. The paper derived time series power plant. In processing analysis, achieved best performance an overall accuracy 97% early warning at least two hours experimental results show that effectively evaluate condition provide plant equipment.