One of the main challenges that deep mining faces is the occurrence of rockburst phenomena. Rockburst prediction with the use of machine learning (ML) is currently gaining attention, as its prognosis capability in many cases outperforms widely used empirical approaches. However, the required data for conducting any analysis are limited, while also having imbalances in their recorded instances associated with rockburst intensities. These, combined with the multiparametric nature of the phenomenon, can deteriorate the performance of the ML algorithms. This study focuses on the enhancement of the prediction performance of ML algorithms by utilizing the oversampling technique Synthetic Minority Oversampling TEchnique (SMOTE). Five ML algorithms, namely Decision Trees, Naïve Bayes, K-Nearest Neighbor, Random Forest and Logistic Regression, were used in a series of parametric analyses considering different combinations of input parameters, such as the maximum tangential stress, the uniaxial compressive and tensile strength, the stress coefficient, two brittleness coefficients and the elastic energy index. All models kept their hyperparameters fixed, and were trained with the initial dataset, in which synthetic instances were added gradually aiming in the attenuation of a balanced dataset and its further expansion, until the number of synthetic instances reached the number of real data. The assessment of the SMOTE technique is given and its performance is evaluated though the different strategies adopted. The results indicate that SMOTE has a considerable positive effect in the accuracy of the overall classification and especially in the improvement of the within-class classification accuracy, even after the balancing of the dataset.
The dominance of isotropic hummocky cross‐stratification, recording deposition solely by oscillatory flows, in many ancient storm‐dominated shoreface–shelf successions is enigmatic. Based on conventional sedimentological investigations, this study shows that storm deposits in three different and stratigraphically separated siliciclastic sediment wedges within the Lower Cretaceous succession in Svalbard record various depositional processes and principally contrasting sequence stratigraphic architectures. The lower wedge is characterized by low, but comparatively steeper, depositional dips than the middle and upper wedges, and records a change from storm‐dominated offshore transition – lower shoreface to storm‐dominated prodelta – distal delta front deposits. The occurrence of anisotropic hummocky cross‐stratification sandstone beds, scour‐and‐fill features of possible hyperpycnal‐flow origin, and wave‐modified turbidites within this part of the wedge suggests that the proximity to a fluvio‐deltaic system influenced the observed storm‐bed variability. The mudstone‐dominated part of the lower wedge records offshore shelf deposition below storm‐wave base. In the middle wedge, scours, gutter casts and anisotropic hummocky cross‐stratified storm beds occur in inferred distal settings in association with bathymetric steps situated across the platform break of retrogradationally stacked parasequences. These steps gave rise to localized, steeper‐gradient depositional dips which promoted the generation of basinward‐directed flows that occasionally scoured into the underlying seafloor. Storm‐wave and tidal current interaction promoted the development and migration of large‐scale, compound bedforms and smaller‐scale hummocky bedforms preserved as anisotropic hummocky cross‐stratification. The upper wedge consists of thick, seaward‐stepping successions of isotropic hummocky cross‐stratification‐bearing sandstone beds attributed to progradation across a shallow, gently dipping ramp‐type shelf. The associated distal facies are characterized by abundant lenticular, wave ripple cross‐laminated sandstone, suggesting that the basin floor was predominantly positioned above, but near, storm‐wave base. Consequently, shelf morphology and physiography, and the nature of the feeder system (for example, proximity to deltaic systems) are inferred to exert some control on storm‐bed variability and the resulting stratigraphic architecture. 相似文献
Dune palaeosurface sequences on the Eastern Canary Islands were investigated for stable isotope records in gastropod shells of the genus Theba. Due to the ecology of the taxon and the special oceanic insularity of the study site, we assume that δ18Oshell signals in our case mainly reflect shifts in δ18O signals of sea surface water. We found that a rapid decrease in δ18Oshell signals is associated with significant changes in gastropod associations. We suggest that these faunal changes were caused by strong (hot) winds at the end of glacial phases, that were described previously by Moreno et al. In addition, we assume that rapid declines in δ18Oshell signals due to marine transgressions were followed by geomorphologically stable phases, dominated by dust enrichment. Such palaeosurfaces correlate with maxima of gastropod biodiversity and with more negative δ13Cshell signals indicating a higher proportion of C3 plants. Based on our results, we also assume that these silty palaeosurfaces were associated with increased soil moisture conditions due to a higher water storage capacity of the finer substrate, independent of climatic moisture conditions. 相似文献
Landslides - We propose a novel fully automated deformation analysis pipeline capable of estimating real 3D displacement vectors from point cloud data. Different from the traditional methods that... 相似文献
Sea ice is an important climate variable and is also an obstacle for marine operations in polar regions. We have developed a small and lightweight, digitally operated frequency-domain electromagnetic-induction (EM) system, a so-called EM bird, dedicated for measurements of sea ice thickness. It is 3.5 m long and weighs only 105 kg, and can therefore easily be shipped to remote places and operated from icebreakers and small helicopters. Here, we describe the technical design of the bird operating at two frequencies of f1 = 3.68 kHz and f2 = 112 kHz, and study its technical performance. On average, noise amounts to ± 8.5 ppm and ± 17.5 ppm for f1 and f2, respectively. Electrical drift amounts to 200 ppm/h and 2000 ppm/h for f1 and f2, during the first 0.5 h of operation. It is reduced by 75% after 2 h. Calibration of the Inphase and Quadrature ppm signals varies by 2 to 3%. A sensitivity study shows that all these signal variations do affect the accuracy of the ice thickness retrieval, but that it remains better than ± 0.1 m over level ice in most cases. This accuracy is also confirmed by means of comparisons of the helicopter EM data with other thickness measurements. The paper also presents the ice thickness retrieval from single-component Inphase data of f1. 相似文献
We investigate helicopter electromagnetic (HEM) inversion schemes applied to synthetic and measured HEM sea ice profiling data. Direct HEM data-to-ice-thickness inversion is compared to three different formal, least squares layered earth inversion algorithms.By making several approximations, it is possible to directly invert a single channel measurement (i.e., the in-phase or quadrature component of a single frequency measurement) to obtain an estimate of sea ice thickness. Measurements from multiple input channels, however, can be used in a layered earth inversion to simultaneously recover several model parameters such as sea ice thickness, sea ice conductivity and sub-ice bathymetry. Synthetic data sets for a particular two-frequency HEM system showed that simple least squares inversion algorithms produce reliable estimates of sea ice thickness in cases where the ice is thicker than 3 m. These methods could also recover acceptable estimates of sea ice thickness when a thin, conductive, partially melted sea ice layer was present, and could determine shallow, sub-ice bathymetry in brackish water. As expected, 1D transformations and inversions of synthetic data for a three-dimensional pressure ridge keel structure contained artifacts, notably broadening of the apparent width of the keel.Prior to inverting a field data set acquired over rather thin (~ 0.5 m) Antarctic sea ice, we found it necessary to recalibrate the phase angle of the measurements using a phasor diagram-based method. Direct transformation of a single channel from the recalibrated data set produced more accurate estimates of sea-ice thickness than formal inversion of multi-channel data. We suggest that the least squares inversion methods are inferior in this situation because of the particular characteristics of the two-frequency HEM system used in this evaluation; the extreme differences in sensitivity of high and low frequency data components, the overall low sensitivity to sea ice conductivity (especially for thin ice), and the partially low signal-to noise ratios of the measurements. The data sets used in this study will be made available to the public to allow alternate inversion approaches to be applied and evaluated. It is suggested that inclusion of parameter bounds and other forms of regularization could help to improve the inversion results. 相似文献
Based on recent findings of a formation mechanism of substructure in tidal tails by Küpper et al., we investigate a more comprehensive set of N -body models of star clusters on orbits about a Milky Way like potential. We find that the predicted epicyclic overdensities arise in any tidal tail no matter which orbit the cluster follows as long as the cluster lives long enough for the overdensities to build up. The distance of the overdensities along the tidal tail from the cluster centre depends for circular orbits only on the mass of the cluster and the strength of the tidal field, and therefore decreases monotonically with time, while for eccentric orbits the orbital motion influences the distance, causing a periodic compression and stretching of the tails and making the distance oscillate with time. We provide an approximation for estimating the distance of the overdensities in this case. We describe an additional type of overdensity which arises in extended tidal tails of clusters on eccentric orbits, when the acceleration of the tidal field on the stellar stream is no longer homogeneous. Moreover, we conclude that a pericentre passage or a disc shock is not the direct origin of an overdensity within a tidal tail. Escape due to such tidal perturbations does not take place immediately after the perturbation but is rather delayed and spread over the orbit of the cluster. All observable overdensities are therefore of the mentioned two types. In particular, we note that substructured tidal tails do not imply the existence of dark matter substructures in the haloes of galaxies. 相似文献
The combination of high spatial and spectral resolution in optical astronomy enables new observational approaches to many
open problems in stellar and circumstellar astrophysics. However, constructing a high-resolution spectrograph for an interferometer
is a costly and time-intensive undertaking. Our aim is to show that, by coupling existing high-resolution spectrographs to
existing interferometers, one could observe in the domain of high spectral and spatial resolution, and avoid the construction
of a new complex and expensive instrument. We investigate in this article the different challenges which arise from combining
an interferometer with a high-resolution spectrograph. The requirements for the different sub-systems are determined, with
special attention given to the problems of fringe tracking and dispersion. A concept study for the combination of the VLTI
(Very Large Telescope Interferometer) with UVES (UV-Visual Echelle Spectrograph) is carried out, and several other specific
instrument pairings are discussed. We show that the proposed combination of an interferometer with a high-resolution spectrograph
is indeed feasible with current technology, for a fraction of the cost of building a whole new spectrograph. The impact on
the existing instruments and their ongoing programs would be minimal. 相似文献
A modification is proposed of Torricelli’s (1608–1647) formula for the velocity of water discharging from a small hole at the bottom of a large tank filled with fractal solid material. The new formula takes proper account of the mechanical energy losses due to flow in the solid matrix, thus expanding the area of validity of the classical Torricelli’s formula. Moreover, it offers a convenient alternative to Darcy’s law for estimating the discharge rate from an aquifer. The new formula was derived from laboratory experiments, with a low-Reynolds number discharge flow (Darcian flow). It was tested in a natural karst aquifer where the flow is non-Darcian, at Almiros spring on the island of Crete (Greece). In both cases, the predictive capability of the modified formula is established. 相似文献