Abstract:

In order to improve the damage prediction ability of the composite structure of the civil aircraft flap in a complex service environment, a damage prediction method of the flap structure based on RBF-ANN considering the complex service environment was proposed. Firstly, a multi-dimensional operation and maintenance database covering structural maintenance information and corresponding environmental information was constructed. The key environmental factors were extracted by grey correlation analysis to realize the quantitative characterization of the service environment, and the damage factors were identified by cluster analysis and variance analysis. Secondly, the RBF-ANN model was constructed, and the hyperparameters of the RBF-ANN model were optimized by Bayesian optimization to realize the prediction of feature damage under the condition of small sample data. Finally, the effectiveness of the proposed model was verified by ablation experiments and comparative experiments. The results show that Bayesian optimization can significantly improve the performance of the model. Mean square error(MSE) and R2 of the model are 0.031 and 0.962, respectively. The optimized model is superior to support vector machine (SVM), long short-term memory network (LSTM) and Bayesian neural network (BNN) in prediction accuracy, error control and fitting ability.

Key words: service environment, radial basis function(RBF)-artificial neural network(ANN), aircraft flap, structure damage, operation and maintenance data, Bayesian optimization, small sample