Early detection of anomalies in dam performance: A methodology based on boosted regression trees

Abstract

Summary The advances in information and communication technologies led to a general trend towards the availability of more detailed information on dam behaviour. This allows applying advanced data‐based algorithms in its analysis, which has been reflected in an increasing interest in the field. However, most of the related literature is limited to the evaluation of model prediction accuracy, whereas the ulterior objective of data analysis is dam safety assessment. In this work, a machine‐learning algorithm (boosted regression trees) is the core of a methodology for early detection of anomalies. It also includes a criterion to determine whether certain discrepancy between predictions and observations is normal, a procedure to compute a realistic estimate of the model accuracy, and an original approach to identify extraordinary load combinations. The performance of causal and noncausal models is assessed in terms of their ability to detect different types of anomalies, which were artificially introduced on reference time series generated with a numerical model of a 100‐m‐high arch dam. The final approach was implemented in an online application to visualise the results in an intuitive way to support decision making.

A Performance Comparison of Machine Learning Algorithms for Arced Labyrinth Spillways

“…Thus, neural networks (NNs) were used for years in hydrology to model rainfall-runoff processes and predict flooding (e.g., References [26][27][28]). In dam safety, there is also a growing community of researchers that employ various algorithms to estimate the structural response of a dam based on the acting loads and to detect anomalies [29][30][31][32][33]. Other fields of application in hydraulic engineering include the management of water supply networks [34] and velocity estimation in air-water flows [35].…”

Anomaly detection with the Switching Kalman Filter for structural health monitoring

“…The seasonal pattern reaches its maximum during winter and minimum during summer and is nonharmonic because of the lack of symmetry with respect to the horizontal axis. The nonharmonic behaviour can be explained by the dependence of the displacement data on the water temperature, where its variation is not harmonic due to the unbalanced duration between the reservoir warming and cooling periods. Figure presents the time‐step length for the entire data set.…”

“…In the field of dam engineering, the most common regression method is the hydrostatic, seasonal, time method and analogue derivations that have been applied to interpret dam behaviour through displacement, pressure, and flow‐rate data. In addition to the hydrostatic, seasonal, time, neural network, support vector machines, boosted regression trees, and others are used for the same purpose. The first drawback of these RMs is that once the model is built using a training set, it stops evolving as new data is collected.…”

A deformation separation method for gravity dam body and foundation based on the observed displacements

“…Dam safety assessment has received much attention since the end of the last century . The safety assessment of a gravity dam requires a wide range of information that is acquired from monitoring systems .…”

“…Dam safety assessment has received much attention since the end of the last century. 1,2 The safety assessment of a gravity dam requires a wide range of information that is acquired from monitoring systems. 3 Usually, there are many instruments equipped in the dam and its surroundings for monitoring the water level, temperature, deformation, and other aspects.…”