Trend prediction of civil aircraft engine vibration signal using ensemble process neural networks

An ensemble prediction model based on boosting process neural networks (PNN) is proposed to predict vibration signal trends of civil aircraft engines. First, the error functions of the AdaBoost. RT algorithm are improved, and an adaptive adjustment strategy is adopted to adjust the classification threshold during the training process. Then, the improved AdaBoost. RT is utilized as the ensemble framework so as to build the ensemble PNN prediction model. The performance of the proposed model is evaluated through the prediction of two actual civil aircraft engine vibration signal series. The results show that the ensemble model performs better than the single PNN model while with simpler structures. The proposed modification version of AdaBoost. RT is superior to the original AdaBoost. RT and a contrast modification version with only improved threshold adjustment strategy. Therefore, the proposed model is suitable for the prediction of civil aircraft engine vibration signal trends.