This analysis demonstrates improved prediction accuracy using BP neural network in structural health monitoring, indicating its effectiveness over LSTM.
Key Points
The improved BP neural network method shows significantly better prediction accuracy than LSTM.
Mean square error for strain data predictions was 20.9719 for the improved BP model, outperforming LSTM's 23.5895.
Analysis reveals the monitoring data of bridges as time series data amenable to BP neural network methods.
The improvements are crucial for effective bridge maintenance and provide valuable insights for structural performance evaluations.