Summary of Application Of Long-short Term Memory and Convolutional Neural Networks For Real-time Bridge Scour Prediction, by Tahrima Hashem et al.
Application of Long-Short Term Memory and Convolutional Neural Networks for Real-Time Bridge Scour Prediction
by Tahrima Hashem, Negin Yousefpour
First submitted to arxiv on: 25 Apr 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary In this paper, researchers leverage deep learning algorithms to forecast scour depth variations around bridge piers based on historical sensor monitoring data. They investigate the performance of LSTM and CNN models using data from Alaska and Oregon bridges from 2006-2021. The results show that LSTM models achieved a mean absolute error (MAE) ranging from 0.1m to 0.5m for predicting bed level variations up to one week in advance, demonstrating reasonable performance. The FCN variant of CNN outperformed other CNN configurations with comparable performance to LSTMs but significantly lower computational costs. The study explores innovative random-search heuristics for hyperparameter tuning and model optimization, reducing computational cost compared to the grid-search method. Additionally, it highlights the significance of historical time series scour data for predicting upcoming events. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper uses special computer algorithms called deep learning to predict how much erosion (scour) will happen around bridge piers in rivers. Erosion is important because it can damage bridges and make them unsafe. The researchers used old sensor data from Alaska and Oregon to train the computers, then tested how well they could predict what would happen next. They found that some algorithms were better than others at predicting erosion, but all of them did a pretty good job. This study shows that using computer algorithms like these can help us make predictions about erosion and maybe even prevent damage to bridges. |
Keywords
» Artificial intelligence » Cnn » Deep learning » Grid search » Hyperparameter » Lstm » Mae » Optimization » Time series
