An empirical comparison of machine learning techniques for dam behaviour modelling. (September 2015)

• Record Type: Journal Article
• Authors: Salazar, F.; Toledo, M.A.; Oñate, E.; Morán, R.
• Is Part Of: Structural safety. Volume 56(2015)
• Page Start: 9
• Page End: 17
• Publication Date: 2015-09
• Subjects: Dam monitoring -- Dam safety -- Machine learning -- Boosted regression trees -- Neural networks -- Random forests -- MARS -- Support vector machines -- Leakage flow; Structural stability -- Periodicals; Safety factor in engineering -- Periodicals; Reliability (Engineering) -- Periodicals; Constructions -- Stabilité -- Périodiques; Coefficient de sécurité en ingénierie -- Périodiques; Fiabilité -- Périodiques; 620.86
• DOI: 10.1016/j.strusafe.2015.05.001
• Languages: English
• ISSNs: 0167-4730

Abstract

Highlights: Predictive models for displacements and leakage in an arch dam were built. The prediction accuracy of five machine learning tools was compared with HST method. A sensitivity analysis to the training set size was performed. Machine learning tools mostly outperform HST, especially boosted regression trees. Abstract: Predictive models are essential in dam safety assessment. Both deterministic and statistical models applied in the day-to-day practice have demonstrated to be useful, although they show relevant limitations at the same time. On another note, powerful learning algorithms have been developed in the field of machine learning (ML), which have been applied to solve practical problems. The work aims at testing the prediction capability of some state-of-the-art algorithms to model dam behaviour, in terms of displacements and leakage. Models based on random forests (RF), boosted regression trees (BRT), neural networks (NN), support vector machines (SVM) and multivariate adaptive regression splines (MARS) are fitted to predict 14 target variables. Prediction accuracy is compared with the conventional statistical model, which shows poorer performance on average. BRT models stand out as the most accurate overall, followed by NN and RF. It was also verified that the model fit can be improved by removing the records of the first years of dam functioning from the training set.