No abstract

Internal erosion is the most important failure mechanism of earth and rockfill dams. Since this type of erosion develops internally and silently, methodologies of data acquisition and processing for dam monitoring are crucial to guarantee a safe operation during the lifespan of these structures. In this context, artificial intelligence techniques show up as tools that can simplify the analysis and verification process not of the internal erosion itself, but of the effects that this pathology causes in the response of the dam to external stimuli. Therefore, within the scope of this paper, a methodological framework for monitoring internal erosion in the body of earth and rockfill dams will be proposed. For that, artificial intelligence methods, especially deep neural autoencoders, will be used to treat the acoustic data collected by geophones installed on a dam. The sensor data is processed to identify patterns and anomalies as well as to classify the dam’s structural health status. In short, the acoustic dataset is preprocessed to reduce its dimensionality. In this process, for each second of acquired data, three parameters are calculated (Hjorth parameters). For each parameter, the data from all the available sensors are used to calibrate an autoencoder. Then, the reconstruction error of each autoencoder is used to monitor how far from the original (normal) state the acoustic signature of the dam is. The time series of reconstruction errors are combined with a cumulative sum (CUSUM) algorithm, which indicates changes in the sequential data collected. Additionally, the outputs of the CUSUM algorithms are treated by a fuzzy logic framework to predict the status of the structure. A scale model is built and monitored to check the effectiveness of the methodology hereby developed, showing that the existence of anomalies is promptly detected by the algorithm. The framework introduced in the present paper aims to detect internal erosion inside dams by combining different techniques in a novel context and methodological workflow. Therefore, this paper seeks to close gaps in prior studies, which mostly treated just parts of the data acquisition–processing workflow.

This paper explores the potential of machine learning techniques to predict the soil water retention curve based on physical characterization parameters. Results from 794 water retention and suction points obtained from 51 different soils were used in the algorithm. The soil properties used are the percentages of gravel, sand, silt, and clay, the plasticity index, the porosity, and the relation between the volumetric water content and total suction. The data were used as input for machine learning estimators to predict the volumetric water content of a soil with specified physical characterization parameters and suction, the techniques of artificial intelligence were developed in python. Results show that an extremely randomized trees’ estimator can reach a coefficient of determination of 0.99 in the training dataset, with a coefficient of 0.90 in the cross-validation and testing dataset, which measures the generalization capacity. Furthermore, a continuous function can be obtained by fitting a model such as Cavalcante & Zornberg, or van Genuchten, or Costa & Cavalcante (bimodal) to the predictions of the machine learning for use in numerical methods. These results indicate that the proposed machine learning estimator can become an interesting alternative to estimate the soil water retention curve in engineering practice. This work is in progress and the predictions can be improved with the addition of new data. Know how to participate at the end of the paper.

No abstract

Universality has always played a major role in every branch of science. Since the advent of cellular automata (CAs), this type of model has been widely applicable to the modeling of physical phenomena. On the other hand, the way the evolution rules were described lacked a unified formulation in terms of mathematical functions. In the present paper, a general formulation that is able to describe every elementary CA is derived. The new representation is given in terms of a new function hereby defined: the iota-delta function.