Journal: Computer-Aided Civil and Infrastructure Engineering
Title: A Vortex-Induced Vibration Warning Method Based on Ensemble-Learning-Embedded Neural Network
An ensemble-learning embedded neural network for vortex-induced vibration (VIV) warning has been proposed. The new model consists of a core module based on ensemble learning and peripheral modules. The core module identifies the abstract features of VIV, while the peripheral modules are responsible for feature extraction and weight control. The core module of the new model is trained entirely using an augmented dataset. Therefore, compared to other models with a similar number of parameters, the new model can be trained with significantly fewer datasets. The model was trained and tested using drainage records from three VIV activities of a cable-stayed bridge under construction. The new model demonstrated superior performance in testing. During the construction phase, data from a single VIV event of another cable-stayed bridge was used for retraining, and the new model successfully achieved VIV warning. Due to its lower data requirements, the new model shows great potential for providing early warnings of VIV.
