Research on transformer-based recovery method for structural health monitoring data
Aiming at the data loss problem of structural health monitoring system,a Transformer-based data reconvery method for structural health monitoring data is proposed,which adopts an asymmetric encoder-decoder structure.In the encoder,a masking mechanism is used to mask the data to be repaired,and the multi-channel structural health monitoring data are combined with the multi-head self-attention mechanism for feature extraction and feature fusion.In the decoder,the high-dimensional features in the encoder are again decoded using the multi-head self-attention mechanism,and the repaired data is output.The proposed method is validated using publicly available Dowling Hall footbridge health monitoring data,and the results show that the proposed method is able to repair the data with high accuracy under different data failure rates,which is of high application value.
structural health monitoringdata recoverydeep learning
万方数据
维普
