基于辛几何模态分解与深度学习融合的推力轴承瓦温预测OA

To address the difficulty in accurately predicting the temperature of the thrust bearing guide pad of hydroelectric generator units due to the coupling and nonlinear effects of multiple factors,a combined prediction model based on SGMD-SE-AVOA-BiLSTM-Informer was proposed by using in-depth mechanism analysis to screen out key influencing factors.First,the temperature signal was de-composed into multiple scales and reconstructed into single components using symplectic geometric mode decomposition(SGMD)combined with sample entropy(SE).Then,for components with diffe-rent complexities,bidirectional long short-term memory network(BiLSTM)and Informer model were used for collaborative modeling.The key hyperparameters were optimized using the African vultures op-timization algorithm(AVOA).Finally,the prediction results of each component were integrated to achieve the accurate prediction of temperature.The experimental results show that the prediction indices RMSE,MAE,MAPE and R2 of the proposed model are 0.516 1℃,0.418 1℃,1.031 9%and 0.997 5,respectively,which show significant advantages over the two single models and the five coupled ablation models.This effectively verifies the applicability and superiority of the method in pro-cessing the multi-scale and nonlinear characteristics of the temperature signal of the thrust bearing guide pad.