A New Kind of Anti-bending and Anti-deformation Cathode Frame for Electrostatic Precipitators

Abstract

The aeroengine boasts a complex structure and operates in a demanding environment characterized by prolonged exposure to extreme conditions such as high temperature and pressure. This type of system is susceptible to nonlinear faults. To enhance the accuracy of aeroengine predictions, this paper proposes a fault diagnosis model based on stacking-based ensemble learning. Initially, the KNN algorithm (K-Nearest Neighbor) and the AdaBoost algorithm (Adaptive Boosting) are optimized by hyperparameters to further augment the accuracy of a single model. Subsequently, the XGBoost model is employed as a meta-learner to fuse the prediction outcomes of the optimized KNN and AdaBoost models. The stacking ensemble learning technique is then applied, followed by the output of the prediction results. Through confirmatory experiments, the accuracy rate following stacking integrated learning improved to 0.9591, while the standard deviation was further minimized to 0.0074. The findings demonstrate that the model is remarkably precise and robust, and can be implemented in aeroengine fault diagnosis.