Search

The increase in the concentration of geological gas emissions in the atmosphere and particularly the increase of methane is considered by the majority of the scientific community as the main cause of global climate change. The main reasons that place methane at the center of interest, lie in its high global warming potential (GWP) and its lifetime in the atmosphere. Anthropogenic processes, like engineering geology ones, highly affect the daily profile of gasses in the atmosphere. Should direct measures be taken to reduce emissions of methane, immediate global warming mitigation could be achieved. Due to its significance, methane has been monitored by many space missions over the years and as of 2017 by the Sentinel-5P mission. Considering the above, we conclude that monitoring and predicting future methane concentration based on past data is of vital importance for the course of climate change over the next decades. To that end, we introduce a method exploiting state-of-the-art recurrent neural networks (RNNs), which have been proven particularly effective in regression problems, such as time-series forecasting. Aligned with the green artificial intelligence (AI) initiative, the paper at hand investigates the ability of different RNN architectures to predict future methane concentration in the most active regions of Texas, Pennsylvania and West Virginia, by using Sentinel-5P methane data and focusing on computational and complexity efficiency. We conduct several empirical studies and utilize the obtained results to conclude the most effective architecture for the specific use case, establishing a competitive prediction performance that reaches up to a 0.7578 mean squared error on the evaluation set. Yet, taking into consideration the overall efficiency of the investigated models, we conclude that the exploitation of RNN architectures with less number of layers and a restricted number of units, i.e., one recurrent layer with 8 neurons, is able to better compensate for competitive prediction performance, meanwhile sustaining lower computational complexity and execution time. Finally, we compare RNN models against deep neural networks along with the well-established support vector regression, clearly highlighting the supremacy of the recurrent ones, as well as discuss future extensions of the introduced work. [ABSTRACT FROM AUTHOR]

Although the principal aim of the rockfall management is to prevent rock boulders from reaching the buildings instead of the buildings resisting the boulder impacts, there usually exists a residual risk that has to be assessed, even when structural protection measurements are taken. The evaluation of the expected damage of buildings due to rockfalls using empirical data from past events is not always possible, as transferring and applying damage observations from one area to another can be unrealistic. In order to simulate potential rockfall scenarios and their damage on buildings, numerical methods can be an alternative. However due to their increased requirements in expertise and computational costs, their integration into the risk analysis is limited, and simpler tools to assess the rockfall vulnerability of buildings are needed. This paper focuses on the application of artificial intelligence AI methods for providing the expected damage of masonry walls which are subjected to rockfall impacts. First, a damage database with 672 datasets was created numerically using the particle finite element method and the finite element method. The input variables are the rock volume (VR), the rock velocity (RV), the masonry wall (t) and the masonry tensile strength f m . The output variable is a damage index (DI) equal to the percentage of the damaged wall area. Different AI algorithms were investigated and the ANN LM 4-21-1 model was selected to optimally assess the expected wall damage. The optimum model is provided here (a) as an analytical equation and (b) in the form of contour graphs, mapping the DI value. Known the VR and the RV, the DI can be directly used as an input for the vulnerability of masonry walls into the quantitative rockfall risk assessment equation. [ABSTRACT FROM AUTHOR]

This study is an efficiency comparison between four methods for the production of landslide susceptibility maps (LSMs), which include random forest (RF), artificial neural network (ANN), and logistic regression (LR) as the machine learning (ML) techniques and frequency ratio (FR) as a statistical method. The study area is located in the Southern Hiroshima Prefecture in western Japan, a locality known to suffer from rainfall-induced landslide disasters, the most recent one in July 2018. The landslide conditioning factors (LCFs) considered in this study are lithology, land use, altitude, slope angle, slope aspect, distance to drainage, distance to lineament, soil class, and mean annual precipitation. The rainfall LCF data comprise XRAIN (eXtended RAdar Information Network) radar records, which are novel in the task of LSM production. The accuracy of the produced LSMs was calculated with the area under the receiver operating characteristic curve (AUROC), and an automatic hyperparameter tuning and result comparison system based on AUROC scores was utilized. The calculated AUROC scores of the resulting LSMs were 0.952 for the RF method, 0.9247 for the ANN method, 0.9016 for the LR method, and 0.8424 for the FR. It is also noteworthy that the ML methods are substantially swifter and more practical than the FR method and allow for multiple and automatic experimentations with different hyperparameter settings, providing fine and accurate outcomes with the given data. The results evidence that ML techniques are more efficient when dealing with hazard assessment problems such as the one exemplified in this study. Although the conclusion that the RF method is the most accurate for LSM production as found by other authors in the literature, ML method efficiency may vary depending on the specific study area, and thus the use of an automatic multi-method LSM production system with hyperparameter tuning such as the one utilized in this study is advised. It was also found that XRAIN radar-acquired mean annual precipitation data are effective when used as an LCF in LSM production. [ABSTRACT FROM AUTHOR]

Predicting the range of achievable strength and stiffness from stabilized soil mixtures is critical for engineering design and construction, especially for organic soils, which are often considered "unsuitable" due to their high compressibility and the lack of knowledge about their mechanical behavior after stabilization. This study investigates the mechanical behavior of stabilized organic soils using machine learning (ML) methods. ML algorithms were developed and trained using a database from a comprehensive experimental study (see Part I), including more than one thousand unconfined compression tests on organic clay samples stabilized by wet soil mixing (WSM) technique. Three different ML methods were adopted and compared, including two artificial neural networks (ANN) and a linear regression method. ANN models proved reliable in the prediction of the stiffness and strength of stabilized organic soils, significantly outperforming linear regression models. Binder type, mixing ratio, soil organic and water content, sample size, aging, temperature, relative humidity, and carbonation were the control variables (input parameters) incorporated into the ML models. The impacts of these factors were evaluated through rigorous ANN-based parametric analyses. Additionally, the nonlinear relations of stiffness and strength with these parameters were developed, and their optimum ranges were identified through the ANN models. Overall, the robust ML approach presented in this paper can significantly improve the mixture design for organic soil stabilization and minimize the experimental cost for implementing WSM in engineering projects. [ABSTRACT FROM AUTHOR]

Landslides pose a significant risk to human life. The Twisting Theory (TWT) and Crown Clustering Algorithm (CCA) are innovative adaptive algorithms that can determine the shape of a landslide and predict its future evolution based on the movement of position sensors located in the affected area. In the first part of this study, the TWT and CCA will be thoroughly explained from a mathematical and theoretical perspective. In the second part, these algorithms will be applied to real-life cases, the Assisi landslide (1995–2008) and the Corvara landslide (2000–2008). A correlation of 0.9997 was attained between the model estimates and the expert's posterior measurements at both examined sites. The results of these applications reveal that the TWT can accurately identify the overall shape of the landslides and predict their progression, while the CCA identifies complex cause-and-effect relationships among the sensors and represents them in a clear, weighted graph. To apply this model to a wider area and secure regions at risk of landslides, it is important to emphasize its operational feasibility as it only requires the installation of GNSS sensors in a predetermined grid in the target area. [ABSTRACT FROM AUTHOR]

This paper presents the results from a study on the application of an artificial neural network (ANN) model for regional flood frequency analysis (RFFA). The study was conducted using stream flow data from 88 gauging stations across New South Wales (NSW) in Australia. Five different models consisting of three to eight predictor variables (i.e., annual rainfall, drainage area, fraction forested area, potential evapotranspiration, rainfall intensity, river slope, shape factor and stream density) were tested. The results show that an ANN model with a higher number of predictor variables does not always improve the performance of RFFA models. For example, the model with three predictor variables performs considerably better than the models using a higher number of predictor variables, except for the one which contains all the eight predictor variables. The model with three predictor variables exhibits smaller median relative error values for 2- and 20-year return periods compared to the model containing eight predictor variables. However, for 5-, 10-, 50- and 100-year return periods, the model with eight predictor variables shows smaller median relative error values. The proposed ANN modelling framework can be adapted to other regions in Australia and abroad. [ABSTRACT FROM AUTHOR]

Machine learning methods for data processing are gaining momentum in many geoscience industries. This includes the mining industry, where machine learning is primarily being applied to autonomously driven vehicles such as haul trucks, and ore body and resource delineation. However, the development of machine learning applications in rock engineering literature is relatively recent, despite being widely used and generally accepted for decades in other risk assessment-type design areas, such as flood forecasting. Operating mines and underground infrastructure projects collect more instrumentation data than ever before, however, only a small fraction of the useful information is typically extracted for rock engineering design, and there is often insufficient time to investigate complex rock mass phenomena in detail. This paper presents a summary of current practice in rock engineering design, as well as a review of literature and methods at the intersection of machine learning and rock engineering. It identifies gaps, such as standards for architecture, input selection and performance metrics, and areas for future work. These gaps present an opportunity to define a framework for integrating machine learning into conventional rock engineering design methodologies to make them more rigorous and reliable in predicting probable underlying physical mechanics and phenomenon. [ABSTRACT FROM AUTHOR]

The study of soil–structure interface behavior contributes to the fundamental understanding of engineering performance and foundation design optimization. Previous research studies the effect of soil characteristics and surface roughness property on the soil–material interface mechanism via interface shear test. The reviews utilizing past established laboratory studies and more recent tests based on state-of-the-art technologies reveal that surface roughness significantly affects interface shear performances in the studies of soil–structure interactions, especially in peak shear strength development. A preliminary but original investigative study by the authors was also carried out using a sophisticated portable surface roughness gauge to define the material surface roughness properties in order to study the interface behavior parametrically. Additionally, using the authors' own original research findings as a proof-of-concept innovation, particle image velocimetry (PIV) technology is applied using a digital single-lens reflex (DSLR) camera to capture sequential images of particle interactions in a custom-built transparent shear box, which validate the well-established four-stage soil shearing model. The authors also envisaged that machine learning, e.g., artificial neural network (ANN) and Bayesian inference method, amongst others, as well as numerical modeling, e.g., discrete element method (DEM), have the potential to also promote research advances on interface shear mechanisms, which will assist in developing a greater understanding in the complex study of soil–structure interactions. [ABSTRACT FROM AUTHOR]

Since its inception in the mid-1950s [[1]], artificial intelligence (AI) has become a disruptive and pervasive technology. In recent years, with the rapid development of ML and its spread over many engineering fields, many researchers have started looking into disciplinary or thematic applications of ML methods. Machine learning (ML) is a subset of AI and is the study of computer algorithms that automatically improve through experience. [Extracted from the article]

What is the optimal level of chaos in a computational system? If a system is too chaotic, it cannot reliably store information. If it is too ordered, it cannot transmit information. A variety of computational systems exhibit dynamics at the "edge of chaos", the transition between the ordered and chaotic regimes. In this work, we examine the evolved neural networks of Polyworld, an artificial life model consisting of a simulated ecology populated with biologically inspired agents. As these agents adapt to their environment, their initially simple neural networks become increasingly capable of exhibiting rich dynamics. Dynamical systems analysis reveals that natural selection drives these networks toward the edge of chaos until the agent population is able to sustain itself. After this point, the evolutionary trend stabilizes, with neural dynamics remaining on average significantly far from the transition to chaos. [ABSTRACT FROM AUTHOR]

Soil moisture monitoring and characterization of the spatial and temporal variability of this hydrologic parameter at scales from small catchments to large river basins continues to receive much attention, reflecting its critical role in subsurface-land surface-atmospheric interactions and its importance to drought analysis, irrigation planning, crop yield forecasting, flood protection, and forest fire prevention. Synthetic Aperture Radar (SAR) data acquired at different spatial resolutions have been successfully used to estimate soil moisture in different semi-arid areas of the world for many years. This research investigated the potential of linear multiple regressions and Artificial Neural Networks (ANN) based models that incorporate different geophysical variables with Radarsat 1 SAR fine imagery and concurrently measured soil moisture measurements to estimate surface soil moisture in Nash Draw, NM. An artificial neural network based model with vegetation density, soil type, and elevation data as input in addition to radar backscatter values was found suitable to estimate surface soil moisture in this area with reasonable accuracy. This model was applied to a time series of SAR data acquired in 2006 to produce soil moisture data covering a normal wet season in the study site. [ABSTRACT FROM AUTHOR]

Modelling of shallow porous aquifers in scenarios where boundary conditions change over time can be a difficult task. In particular, this is true when data modelling is pursued, i.e., models are directly constructed by measured data. In fact, data contain not only the information related to the physical phenomenon under investigation, but also the effects of time-varying boundary conditions, which work as a disturbance. This undesired component conditions the training of data-driven models, as they are fitted by models, which can produce predictions diverging from measured data. Here, a very shallow porous aquifer is modelled in terms of its response to water table to precipitation. The aquifer is characterized by the presence of a low permeability silty top layer covering the lower sandy strata, where the aquifer normally flows. Therefore, when the piezometric level increases up to the low permeability layer, the aquifer changes its behavior from phreatic to confined. This determines the changing boundary condition, which makes the response of the aquifer to rain precipitations complex, as it is related to a two-fold condition: confined or phreatic. The aquifer here is investigated by two machine learning approaches, the earlier based on an evolutionary modeling, and the latter based on artificial neural networks. Evolutionary modeling returned explicit equations with a fitness efficiency up to 0.8 for 1 month for predictions and 0.48 for simulations, while neural networks arrived at 0.85 and 0.28, respectively. The aim of this study is to get an explicit model of the response of the piezometric heights of the aquifer to the precipitations, which is useful for planning the use of groundwater resources. [ABSTRACT FROM AUTHOR]

The classification of soils into categories with a similar range of properties is a fundamental geotechnical engineering procedure. At present, this classification is based on various types of cost- and time-intensive laboratory and/or in situ tests. These soil investigations are essential for each individual construction site and have to be performed prior to the design of a project. Since Machine Learning could play a key role in reducing the costs and time needed for a suitable site investigation program, the basic ability of Machine Learning models to classify soils from Cone Penetration Tests (CPT) is evaluated. To find an appropriate classification model, 24 different Machine Learning models, based on three different algorithms, are built and trained on a dataset consisting of 1339 CPT. The applied algorithms are a Support Vector Machine, an Artificial Neural Network and a Random Forest. As input features, different combinations of direct cone penetration test data (tip resistance qc, sleeve friction fs, friction ratio Rf, depth d), combined with "defined", thus, not directly measured data (total vertical stresses σv, effective vertical stresses σ'v and hydrostatic pore pressure u0), are used. Standard soil classes based on grain size distributions and soil classes based on soil behavior types according to Robertson are applied as targets. The different models are compared with respect to their prediction performance and the required learning time. The best results for all targets were obtained with models using a Random Forest classifier. For the soil classes based on grain size distribution, an accuracy of about 75%, and for soil classes according to Robertson, an accuracy of about 97–99%, was reached. [ABSTRACT FROM AUTHOR]

Pipelines are exposed to the severe threat of natural disasters, where the damage caused by landslides are particularly bad. Hence, in the route arrangement and maintenance management of pipeline projects, it is particularly important to evaluate the regional landslide hazards in advance. However, most models are based on the subjective determination of evaluation factors and index weights; this study establishes a quantitative hazard assessment model based on the location of historical landslides and the Levenberg–Marquardt Back Propagation (LM-BP) Neural Network model was applied to the pipeline area. We established an evaluation index system by analyzing the spatial patterns of single assessment factors and the mechanism of landslides. Then, different from previous studies, we built the standard sample matrix of the LM-BP neural network by using interpolation theory to avoid the serious influence of human factors on the hazard assessment. Finally, we used the standard sample matrix and the historical data to learn, train, test, and simulate future results. Our results showed 33 slopes with low hazard (accounting for 10.48% of the total number of slopes and corresponding to approximately 32.63 km2), 62 slopes with moderate hazard (accounting for 19.68% of the total number of slopes and corresponding to approximately 65.53 km2), 112 slopes with high hazard (accounting for 35.56% of the total number of slopes and corresponding to approximately 123.55 km2), and 108 slopes with extremely high hazard (accounting for 34.29% of the total number of slopes and corresponding to approximately 150.65 km2). Local spatial autocorrelation analysis indicated that there are significant "high–high" and "low–low" aggregation of landslide hazards in the pipeline area. By comparing the model results with the past landslides, new landslides and landslide potential points, its prediction capability and accuracy were confirmed. On the basis of the results, our study has developed effective risk prevention and mitigation strategies in mountain areas to promote pipeline safety. [ABSTRACT FROM AUTHOR]