Development of fault diagnosis platform for liquid rocket engine based on experimental data

Abstract Aiming at the defects of difficult to establish fault diagnosis models caused by small scale liquid rocket engine test data, imbalanced categories and high coupling, platform with 10 are developed for diagnosis, covering K-nearest neighbor (KNN) model, optimized KNN logistic regression (LR) LR support vector machine K-means decision tree neural network random forest model light-GBM model. The prediction accuracies these validated based on experimental data. Among models, provide best accuracies, 5 have accuracy larger than 98%.