1,721,159 research outputs found
Inferring semi-parametric Gaussian process model parameters for missing geotechnical data prediction
No full textData points in geotechnical site investigation data (i.e., CPT data) may be missing sometimes due to various reasons. This study proposed to use a semi-parametric Gaussian process regression (GPR) method for predicting missing data in geotechnical testing results. Semi-parametric GPR divides the spatial data into the trend function, spatial residual, and measurement errors. Compared with conventional GPR method, semi-parametric GPR enhances model interpretability and accuracy. However, this involves challenges in estimating the parameters in the model. Conventional GPR applications infer the model parameters based on maximum a posteriori (MAP) estimation. However, this method can only provide a point estimation of the model parameters. Point estimation may be trapped by a local optimum result. This study utilizes the Hamiltonian Monte Carlo (HMC) method to get the full posterior distribution of the model parameters. MAP and HMC methods are both applied to infer the model parameters based on a synthetic CPT data set. The performances of both methods are compared with the true model values. The results show that the model parameters estimated from the HMC are more reliable.</p
Deep Bayesian survival analysis of rail useful lifetime
No full textReliable estimation of rail useful lifetime can provide valuable information for predictive maintenance in railway systems. However, in most cases, lifetime data is incomplete because not all pieces of rail experience failure by the end of the study horizon, a problem known as censoring. Ignoring or otherwise mistreating the censored cases might lead to false conclusions. Survival approach is particularly designed to handle censored data for analysing the expected duration of time until one event occurs, which is rail failure in this paper. This paper proposes a deep Bayesian survival approach named BNN-Surv to properly handle censored data for rail useful lifetime modelling. The proposed BNN-Surv model applies the deep neural network in the survival approach to capture the non-linear relationship between covariates and rail useful lifetime. To consider and quantify uncertainty in the model, Monte Carlo dropout, regarded as the approximate Bayesian inference, is incorporated into the deep neural network to provide the confidence interval of the estimated lifetime. The proposed approach is implemented on a four-year dataset including track geometry monitoring data, track characteristics data, various types of defect data, and maintenance and replacement (M&R) data collected from a section of railway tracks in Australia. Through extensive evaluation, including Concordance index (C-index) and root mean square error (RMSE) for evaluating model performance, as well as a proposed CW-index for evaluating uncertainty estimations, the effectiveness of the proposed approach is confirmed. The results show that, compared with other commonly used models, the proposed approach can achieve the best concordance index (C-index) of 0.80, and the estimated rail useful lifetimes are closer to real lifetimes. In addition, the proposed approach can provide the confidence interval of the estimated lifetime, with a correct coverage of 81% of the actual lifetime when the confidence interval is 1.38, which is more useful than point estimates in decision-making and maintenance planning of railroad systems.Railway Engineerin
Advancing spatial-temporal rock fracture prediction with virtual camera-based data augmentation
No full textPredicting rock fractures in unexcavated areas is a critical yet challenging aspect of geotechnical projects. This task involves forecasting the fracture mapping sequences for unexcavated rock faces using the sequences from excavated ones, which is well-suited for spatial–temporal deep learning techniques. Fracture mapping sequences for deep learning model training can be achieved based on field photography. However, the main obstacle lies in the insufficient availability of high-quality photos. Existing data augmentation techniques rely on slices taken from Discrete Fracture Network (DFN) models. However, slices differ significantly from actual photos taken from the field. To overcome this limitation, this study introduces a new framework that uses Virtual Camera Technology (VCT) to generate “virtual photos” from DFN models. The external (e.g., camera location, direction) and internal parameters (e.g., focal length, resolution, sensor size) of cameras can be considered in this method. The “virtual photos” generated from the VCT and conventional slicing method have been extensively compared. The framework is designed to adapt to any distribution of field fractures and camera settings, serving as a universal tool for practical applications. The whole framework has been packaged as an open-source tool for rock “photos” generation. An open-source benchmark database has also been established based on this tool. To validate the framework's feasibility, the Predictive Recurrent Neural Network (PredRNN) method is applied to the generated database. A high degree of similarity is observed between the predicted mapping sequences and the ground truth. The model successfully captured the dynamic changes in fracture patterns across different sections, thereby confirming the framework's practical utility. The source code and dataset can be freely downloaded from GitHub repository (https://github.com/GEO-ATLAS/Rock-Camera).</p
Updating reliability of pile groups with load tests considering spatially variable soils
No full textThis paper proposes a rigorous framework to update the reliability of pile groups based on load tests. The proposed approach enables the consideration of the spatial variability of soils, which is disregarded in previous studies. To achieve this, the random finite difference method (RFDM) is utilised to assess the group efficiency, individual pile capacities, and the correlation between individual pile capacities in spatially variable soils. Subsequently, Bayes’ theorem is employed to update individual pile capacities based on load test results, taking into account the correlation between individual pile capacities. Finally, the reliability of pile groups is evaluated based on the group efficiency and updated individual pile capacities. An axially loaded pile group in undrained clays is utilised for demonstration. Results indicate that neglecting the spatial variability of soils may lead to unrealistic assessments of the reliability of pile groups. Specifically, in cases where all piles fail, the ignorance of spatial variability results in an overconservative design. Conversely, in cases where one or more piles pass, it leads to an unconservative design.</p
Calibrating resistance factors of pile groups based on individual pile proof load tests
No full textPile load tests have been utilized to reduce the uncertainty of pile resistance, thus leading to a higher resistance factor used in the Load and Resistance Factor Design (LRFD). Previous studies have primarily focused on calibrating resistance factors for single piles based on load tests. This calibration hinges upon the resistance bias factor of single piles, defined as the ratio of measured resistance to predicted resistance. Due to the redundancy in the pile group system, it is conventionally assumed that if the individual piles within the group achieve a lower reliability index (e.g., 2.0–2.5), the pile group as a whole attains the target reliability index of 3. However, the approach is empirical as it does not consider system redundancy directly. Moreover, this empirical approach disregards the correlation between resistance bias factors of individual piles, which is inherently influenced by the spatial variability of soils. In this study, the random finite difference method (RFDM) is employed to evaluate the correlation between resistance bias factors of individual piles in spatially variable soils. The resultant correlation matrix is subsequentially employed in Bayes’ theorem to update resistance bias factors using individual pile load test results and their corresponding test locations. The updated resistance bias factors are then used for the direct calibration of resistance factors for pile groups within the framework of LRFD. A pile group subject to vertical loading in undrained clays is adopted for illustration. Comparative analyses between the proposed approach and the empirical approach demonstrate that the latter tends to overestimate the resistance factor. Furthermore, the proposed approach enables the determination of optimal locations for conducting subsequent load tests based on previous test results.</p
Robust calibration of shaft and base resistance factors for piles based on multiobjective optimization
No full textResistance factors are used to account for the uncertainties associated with pile resistance in load and resistance factor design (LRFD). Current design codes and most previous studies recommend a single resistance factor applied to the total pile resistance (shaft and base resistances). However, the uncertainties associated with shaft and base resistances are significantly different. Moreover, resistance factors are generally calibrated based on the statistics of resistance bias factors derived using all data collected from different sites, whereas the variability of the statistics between various sites (i.e., cross-site variability) has been ignored in the traditional calibration approaches, which may result in the designs based on the calibrated resistance factors violating safety requirements. In this paper, a robust calibration approach is proposed to calibrate shaft and base resistance factors, explicitly considering the cross-site variability in the statistics of resistance bias factors in the calibration process. To achieve that, the feasible robustness concept is adopted to describe the probability that the design remains able to achieve the target reliability index when the statistics of resistance bias factor exhibit cross-site variability. The calibration process is implemented through a multiobjective optimization, which leads to a Pareto front that describes the trade-off relationship between shaft and base resistance factors and feasible robustness. The optimal shaft and base resistance factors are determined using the minimum distance approach. The proposed approach is demonstrated and applied to calibrate shaft and base resistance factors for three design methods, the Vesic, Meyerhof, and Nordlund methods. Results show that resistance factors are significantly affected by design methods and the ratio of shaft and base resistances.</p
Bayesian updating on resistance factors of H-Piles with axial load tests
No full textIn the Load and Resistance Factor Design (LRFD) of piles, several design codes recommend higher resistance factors if load tests are conducted. However, no information is provided on how these resistance factors are determined. In this paper, a probabilistic approach based on Bayes' theorem and the First Order Reliability Method (FORM) is proposed to calibrate resistance factors for different numbers of load tests and the corresponding test results. In addition, within-site variability, design methods, types of piles, and ground conditions can also be considered. The proposed approach applied to H-piles under axial load tests shows consistent results with current design codes. Results show that resistance factors are significantly increased even if only one positive test is observed among all the tests. For low variability sites, the differences of resistance factors between various design methods are significantly reduced if one or more tests are positive, while for high variability sites, the differences of resistance factors are only slightly decreased, indicating that design methods should be considered in the latter case. Most of the increase in resistance factors is achieved with a small number of tests. For β-Method used in clay sites, 80% of the increase in resistance factors is achieved with two, four and five consecutive positive tests are observed for low, medium and high variability sites, respectively.</p
- …
