Search

This research focuses on developing reinforcement learning approaches for the locomotion generation of small-size quadruped robots. The rat robot NeRmo is employed as the experimental platform. Due to the constrained volume, small-size quadruped robots typically possess fewer and weaker sensors, resulting in difficulty in accurately perceiving and responding to environmental changes. In this context, insufficient and imprecise feedback data from sensors makes it difficult to generate adaptive locomotion based on reinforcement learning. To overcome these challenges, this paper proposes a novel reinforcement learning approach that focuses on extracting effective perceptual information to enhance the environmental adaptability of small-size quadruped robots. According to the frequency of a robot's gait stride, key information of sensor data is analyzed utilizing sinusoidal functions derived from Fourier transform results. Additionally, a multifunctional reward mechanism is proposed to generate adaptive locomotion in different tasks. Extensive simulations are conducted to assess the effectiveness of the proposed reinforcement learning approach in generating rat robot locomotion in various environments. The experiment results illustrate the capability of the proposed approach to maintain stable locomotion of a rat robot across different terrains, including ramps, stairs, and spiral stairs.

Balancing oneself using the spine is a physiological alignment of the body posture in the most efficient manner by the muscular forces for mammals. For this reason, we can see many disabled quadruped animals can still stand or walk even with three limbs. This paper investigates the optimization of dynamic balance during trot gait based on the spatial relationship between the center of mass (CoM) and support area influenced by spinal flexion. During trotting, the robot balance is significantly influenced by the distance of the CoM to the support area formed by diagonal footholds. In this context, lateral spinal flexion, which is able to modify the position of footholds, holds promise for optimizing balance during trotting. This paper explores this phenomenon using a rat robot equipped with a soft actuated spine. Based on the lateral flexion of the spine, we establish a kinematic model to quantify the impact of spinal flexion on robot balance during trot gait. Subsequently, we develop an optimized controller for spinal flexion, designed to enhance balance without altering the leg locomotion. The effectiveness of our proposed controller is evaluated through extensive simulations and physical experiments conducted on a rat robot. Compared to both a non-spine based trot gait controller and a trot gait controller with lateral spinal flexion, our proposed optimized controller effectively improves the dynamic balance of the robot and retains the desired locomotion during trotting.

To handle graphs in which features or connectivities are evolving over time, a series of temporal graph neural networks (TGNNs) have been proposed. Despite the success of these TGNNs, the previous TGNN evaluations reveal several limitations regarding four critical issues: 1) inconsistent datasets, 2) inconsistent evaluation pipelines, 3) lacking workload diversity, and 4) lacking efficient comparison. Overall, there lacks an empirical study that puts TGNN models onto the same ground and compares them comprehensively. To this end, we propose BenchTemp, a general benchmark for evaluating TGNN models on various workloads. BenchTemp provides a set of benchmark datasets so that different TGNN models can be fairly compared. Further, BenchTemp engineers a standard pipeline that unifies the TGNN evaluation. With BenchTemp, we extensively compare the representative TGNN models on different tasks (e.g., link prediction and node classification) and settings (transductive and inductive), w.r.t. both effectiveness and efficiency metrics. We have made BenchTemp publicly available at https://github.com/qianghuangwhu/benchtemp., Comment: 28 pages, 23 figures, 27 tables. Submitted to the Conference on Neural Information Processing Systems 2023 Track on Datasets and Benchmarks

In e-commerce industry, graph neural network methods are the new trends for transaction risk modeling.The power of graph algorithms lie in the capability to catch transaction linking network information, which is very hard to be captured by other algorithms.However, in most existing approaches, transaction or user connections are defined by hard link strategies on shared properties, such as same credit card, same device, same ip address, same shipping address, etc. Those types of strategies will result in sparse linkages by entities with strong identification characteristics (ie. device) and over-linkages by entities that could be widely shared (ie. ip address), making it more difficult to learn useful information from graph. To address aforementioned problems, we present a novel behavioral biometric based method to establish transaction linkings based on user behavioral similarities, then train an unsupervised GNN to extract embedding features for downstream fraud prediction tasks. To our knowledge, this is the first time similarity based soft link has been used in graph embedding applications. To speed up similarity calculation, we apply an in-house GPU based HDBSCAN clustering method to remove highly concentrated and isolated nodes before graph construction. Our experiments show that embedding features learned from similarity based behavioral graph have achieved significant performance increase to the baseline fraud detection model in various business scenarios. In new guest buyer transaction scenario, this segment is a challenge for traditional method, we can make precision increase from 0.82 to 0.86 at the same recall of 0.27, which means we can decrease false positive rate using this method., Comment: 8 pages, 8 figures, ICDM

At online retail platforms, detecting fraudulent accounts and transactions is crucial to improve customer experience, minimize loss, and avoid unauthorized transactions. Despite the variety of different models for deep learning on graphs, few approaches have been proposed for dealing with graphs that are both heterogeneous and dynamic. In this paper, we propose DyHGN (Dynamic Heterogeneous Graph Neural Network) and its variants to capture both temporal and heterogeneous information. We first construct dynamic heterogeneous graphs from registration and transaction data from eBay. Then, we build models with diachronic entity embedding and heterogeneous graph transformer. We also use model explainability techniques to understand the behaviors of DyHGN-* models. Our findings reveal that modelling graph dynamics with heterogeneous inputs need to be conducted with "attention" depending on the data structure, distribution, and computation cost., Comment: Manuscript under review. arXiv admin note: text overlap with arXiv:2012.10831

As one of the most popular machine learning models today, graph neural networks (GNNs) have attracted intense interest recently, and so does their explainability. Users are increasingly interested in a better understanding of GNN models and their outcomes. Unfortunately, today's evaluation frameworks for GNN explainability often rely on few inadequate synthetic datasets, leading to conclusions of limited scope due to a lack of complexity in the problem instances. As GNN models are deployed to more mission-critical applications, we are in dire need for a common evaluation protocol of explainability methods of GNNs. In this paper, we propose, to our best knowledge, the first systematic evaluation framework for GNN explainability, considering explainability on three different "user needs". We propose a unique metric that combines the fidelity measures and classifies explanations based on their quality of being sufficient or necessary. We scope ourselves to node classification tasks and compare the most representative techniques in the field of input-level explainability for GNNs. For the inadequate but widely used synthetic benchmarks, surprisingly shallow techniques such as personalized PageRank have the best performance for a minimum computation time. But when the graph structure is more complex and nodes have meaningful features, gradient-based methods are the best according to our evaluation criteria. However, none dominates the others on all evaluation dimensions and there is always a trade-off. We further apply our evaluation protocol in a case study for frauds explanation on eBay transaction graphs to reflect the production environment., Comment: Published at the Learning on Graphs 2022 (LoG2022) and New Frontiers in Graph Learning Workshop (Neurips 2022)

Detecting fraudulent transactions is an essential component to control risk in e-commerce marketplaces. Apart from rule-based and machine learning filters that are already deployed in production, we want to enable efficient real-time inference with graph neural networks (GNNs), which is useful to catch multihop risk propagation in a transaction graph. However, two challenges arise in the implementation of GNNs in production. First, future information in a dynamic graph should not be considered in message passing to predict the past. Second, the latency of graph query and GNN model inference is usually up to hundreds of milliseconds, which is costly for some critical online services. To tackle these challenges, we propose a Batch and Real-time Inception GrapH Topology (BRIGHT) framework to conduct an end-to-end GNN learning that allows efficient online real-time inference. BRIGHT framework consists of a graph transformation module (Two-Stage Directed Graph) and a corresponding GNN architecture (Lambda Neural Network). The Two-Stage Directed Graph guarantees that the information passed through neighbors is only from the historical payment transactions. It consists of two subgraphs representing historical relationships and real-time links, respectively. The Lambda Neural Network decouples inference into two stages: batch inference of entity embeddings and real-time inference of transaction prediction. Our experiments show that BRIGHT outperforms the baseline models by >2\% in average w.r.t.~precision. Furthermore, BRIGHT is computationally efficient for real-time fraud detection. Regarding end-to-end performance (including neighbor query and inference), BRIGHT can reduce the P99 latency by >75\%. For the inference stage, our speedup is on average 7.8$\times$ compared to the traditional GNN., Comment: CIKM Acceptance

This paper reports the innovative development of a flexible, multiple-layer, economical but robust, and automated high-speed shear wave (Vs) tomographic control and data acquisition system for process monitoring in the laboratory and the associated validation for monitoring the dynamic process of model pile installation. This tailor-made tomographic system consists of four major parts: (i) the control unit built with a microcontroller board, (ii) a signal generating unit, (iii) three layers of bender element sensing arrays (including 14 source and receiver bender elements per layer), and (iv) a signal receiving unit. The system allows for sampling rates up to 1 GS/s (20 MS/s were used in this study); the required total measurement time is 2.94 seconds (i.e., for 196 ray path measurements) to allow for dynamic process monitoring. The obtained time-lapse Vs tomographic images during pile installation show that Vs first gradually increases when the pile toe is approaching the sensing layer and gradually decreases afterward as the pile toe penetrates through the layer. The validity of the Vs tomographic images is supported by similar responses in variations of the circumferential stress σc0 and radial stress σr0, measured by the tactile pressure sensors.

Transaction checkout fraud detection is an essential risk control components for E-commerce marketplaces. In order to leverage graph networks to decrease fraud rate efficiently and guarantee the information flow passed through neighbors only from the past of the checkouts, we first present a novel Directed Dynamic Snapshot (DDS) linkage design for graph construction and a Lambda Neural Networks (LNN) architecture for effective inference with Graph Neural Networks embeddings. Experiments show that our LNN on DDS graph, outperforms baseline models significantly and is computational efficient for real-time fraud detection.

Massive account registration has raised concerns on risk management in e-commerce companies, especially when registration increases rapidly within a short time frame. To monitor these registrations constantly and minimize the potential loss they might incur, detecting massive registration and predicting their riskiness are necessary. In this paper, we propose a Dynamic Heterogeneous Graph Neural Network framework to capture suspicious massive registrations (DHGReg). We first construct a dynamic heterogeneous graph from the registration data, which is composed of a structural subgraph and a temporal subgraph. Then, we design an efficient architecture to predict suspicious/benign accounts. Our proposed model outperforms the baseline models and is computationally efficient in processing a dynamic heterogeneous graph constructed from a real-world dataset. In practice, the DHGReg framework would benefit the detection of suspicious registration behaviors at an early stage., Comment: 8 pages, 1 figure, accepted in the AAAI Workshop on Deep Learning on Graphs 2021

At online retail platforms, it is crucial to actively detect the risks of transactions to improve customer experience and minimize financial loss. In this work, we propose xFraud, an explainable fraud transaction prediction framework which is mainly composed of a detector and an explainer. The xFraud detector can effectively and efficiently predict the legitimacy of incoming transactions. Specifically, it utilizes a heterogeneous graph neural network to learn expressive representations from the informative heterogeneously typed entities in the transaction logs. The explainer in xFraud can generate meaningful and human-understandable explanations from graphs to facilitate further processes in the business unit. In our experiments with xFraud on real transaction networks with up to 1.1 billion nodes and 3.7 billion edges, xFraud is able to outperform various baseline models in many evaluation metrics while remaining scalable in distributed settings. In addition, we show that xFraud explainer can generate reasonable explanations to significantly assist the business analysis via both quantitative and qualitative evaluations., Comment: This is the extended version of a full paper to appear in PVLDB 15 (3) (VLDB 2022)

Underwater navigation has been a big challenge for autonomous underwater vehicles (AUVs) for a long time. It is highly dependent on acoustic methods called SONAR. There are two kinds of sonar sensors which are commonly used, the multibeam sonar and the sidescan sonar. Both of them have some advantages and limitations. Substantial improvements can be made if a machine interpretation method can be developed for the translation between these two sonar data.The objective of this thesis project is to find an effective way to do translation from seabed bathymetry (underwater depth) data (from multibeam sonar) to sidescan sonar images. In the project, we explored the feasibility of machine learning based translation methods. Some different generative models based on the idea of generative adversarial nets were tried. This project is an experimental trial, and it still needs more improvement before production. But the result shows a strong potential for the ability of machine learning based methods to handle this kind of translation tasks., Navigeringen har varit en stor utmaning för autonoma undervattensfordon (AUV) under lång tid. Typiskt används akustiska metoder, så kallad SONAR. Det finns två typer av sonarsensorer, multibeam sonar och sidescan sonar. Båda har styrkor och svagheter. Genom att översätta mellan dessa två sensordata kan betydande förbättringar uppnås.Syftet med detta avhandlingsprojekt är att hitta ett effektivt sätt för att översätta data från batymetri (undervattensdjup, från multibeam sonar) till sidescan sonarbilder. I projektet undersökte vi genomförbarheten för översättningsmetoder baserad på maskininlärning. Olika generativa modeller baserade på generative adversarial nets (GANs) hade undersöktes. Detta projekt kan ses som en förstudie. Ytterligare förbättringar krävs fortfarande, men resultatet visar en stark potential för maskininlärningsmetoder att hantera denna typ av översättningsuppgifter.

Zhen Wang,1,* Chong Zheng,2,* Siyuan Wen,1 Junfei Wang,1 Zitao Zhang,1 Xusheng Qiu,1 Yixin Chen11Department of Orthopedics, Drum Tower Hospital Clinical College of Nanjing University, Nanjing, People’s Republic of China; 2Department of Orthopedics, Changzhou Traditional Chinese Medicine Hospital Affiliated to Nanjing University of Chinese Medicine, Changzhou, People’s Republic of China*These authors contributed equally to this workPurpose: Infected nonunion after open reduction internal fixation (ORIF) is a serious complication. The aim of this study was to evaluate the usefulness of serum D-dimer for preoperative diagnosis of infected nonunion.Patients and methods: Patients undergoing debridement and external fixation for infected nonunion (n=32) and replacement of internal fixation due to aseptic failure (n=34) were enrolled and compared in this retrospective study. The optimum cutoff value of D-dimer for identification of infected nonunion was determined by calculating the Youden J statistic. The sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of four preoperative laboratory parameters—serum D-dimer level, white blood cell (WBC) count, erythrocyte sedimentation rate (ESR), and C-reactive protein (CRP)—for diagnosis of infected nonunion were compared.Results: Serum D-dimer level was significantly higher in patients with infected nonunion than in patients with aseptic nonunion: 2.62 mg/mL (range, 0.13–11.90 mg/mL) vs 0.35 mg/mL (range, 0.07–6.46 mg/mL; p<0.001). WBC count, CRP, and ESR demonstrated sensitivity of 12.5% (95% CI: 4.08–29.93), 40.6% (95% CI: 24.22–59.21), and 56.3% (95% CI: 37.88–73.16), respectively, and specificity of 94.1% (95% CI: 78.94–98.97), 88.2% (95% CI: 71.61–96.16), and 85.3% (95% CI: 68.17–94.46), respectively. Using the Youden index, 1.70 mg/mL was determined as the o

Underwater navigation has been a big challenge for autonomous underwater vehicles (AUVs) for a long time. It is highly dependent on acoustic methods called SONAR. There are two kinds of sonar sensors which are commonly used, the multibeam sonar and the sidescan sonar. Both of them have some advantages and limitations. Substantial improvements can be made if a machine interpretation method can be developed for the translation between these two sonar data.The objective of this thesis project is to find an effective way to do translation from seabed bathymetry (underwater depth) data (from multibeam sonar) to sidescan sonar images. In the project, we explored the feasibility of machine learning based translation methods. Some different generative models based on the idea of generative adversarial nets were tried. This project is an experimental trial, and it still needs more improvement before production. But the result shows a strong potential for the ability of machine learning based methods to handle this kind of translation tasks., Navigeringen har varit en stor utmaning för autonoma undervattensfordon (AUV) under lång tid. Typiskt används akustiska metoder, så kallad SONAR. Det finns två typer av sonarsensorer, multibeam sonar och sidescan sonar. Båda har styrkor och svagheter. Genom att översätta mellan dessa två sensordata kan betydande förbättringar uppnås.Syftet med detta avhandlingsprojekt är att hitta ett effektivt sätt för att översätta data från batymetri (undervattensdjup, från multibeam sonar) till sidescan sonarbilder. I projektet undersökte vi genomförbarheten för översättningsmetoder baserad på maskininlärning. Olika generativa modeller baserade på generative adversarial nets (GANs) hade undersöktes. Detta projekt kan ses som en förstudie. Ytterligare förbättringar krävs fortfarande, men resultatet visar en stark potential för maskininlärningsmetoder att hantera denna typ av översättningsuppgifter.

Despite the wide application of Graph Convolutional Network (GCN), one major limitation is that it does not benefit from the increasing depth and suffers from the oversmoothing problem. In this work, we first characterize this phenomenon from the information-theoretic perspective and show that under certain conditions, the mutual information between the output after $l$ layers and the input of GCN converges to 0 exponentially with respect to $l$. We also show that, on the other hand, graph decomposition can potentially weaken the condition of such convergence rate, which enabled our analysis for GraphCNN. While different graph structures can only benefit from the corresponding decomposition, in practice, we propose an automatic connectivity-aware graph decomposition algorithm, DeGNN, to improve the performance of general graph neural networks. Extensive experiments on widely adopted benchmark datasets demonstrate that DeGNN can not only significantly boost the performance of corresponding GNNs, but also achieves the state-of-the-art performances., Comment: 20 pages, 5 figures, 5 tables

In this study, an innovative experimental setup for model pile tests was developed to characterize friction fatigue in sand during cyclic jacking. In the experimental setup, a high-performance precision electric linear actuator was used to control jacking of the model piles. P-wave tomographic imaging was used to characterize the associated changes in the distribution of the P-wave velocities in the soil, and tactile pressure sensors were utilized to monitor related variations in stresses in the soil surrounding the pile. Measurements obtained from the experiments allow us to explore the behavior of fiction fatigue from a different perspective, especially from the viewpoint of stress changes in the soil surrounding the pile. It was found that for a soil element surrounding the pile in a jacking cycle, the maximum values of the stationary radial and hoop stresses and the ultimate radial and hoop stresses in that soil element are reached when the pile penetrates to nearly the same depth, which is therefore defined as the transition depth. For the soil elements at the same depth, such a transition depth increases with increasing distance from the pile centerline. When the pile base gradually penetrates through the transition depth, these stresses in the soil elements surrounding the pile will exceed the peak value, which in turn leads to a reduction in the ultimate unit shear resistance (τf), giving rise to friction fatigue. Based on these experimental findings, a simple model is also proposed to explain friction fatigue.

Zitao Zhang,1 Yuanhui Ji,2 Zhen Wang,1 Xusheng Qiu,1 Yixin Chen1 1Department of Orthopedics, The Affiliated Drum Tower Hospital of Nanjing University Medical School, Nanjing 210008, China; 2Department of Clinical Laboratory, The Affiliated Drum Tower Hospital of Nanjing University Medical School, Nanjing 210008, China Objectives: Deep surgical site infection (DSSI) is one of the most serious complications after open induction internal fixation (ORIF) for traumatic limb fractures. In this study, we aimed to investigate the diagnostic role of platelet indices (platelet count [PLT], mean platelet volume [MPV], and platelet distribution width [PDW]) in DSSI.Patients and methods: Data obtained between January 2011 and December 2017 in The Affiliated Drum Tower Hospital of Nanjing University Medical School from cases (n=29) with DSSI and fracture control subjects (n=29) matched for age, gender, and fracture type were analyzed. The white blood cell (WBC) count, neutrophil count, neutrophil percentage, and platelet indices from blood samples were compared between case and control groups. In addition, the cutoff value, sensitivity, and specificity were calculated by receiver-operating characteristic (ROC) curves.Results: No significant differences were detected in demographic features, the WBC count, neutrophil count, neutrophil percentage, and MPV values between two groups (P>0.05). The PLT values were significantly higher in the case group than in the control group (303.00±139.27 vs 196.10±59.61 [109/μL], P=0.001). The PDW values of the case and control groups were 11.77±2.71 and 13.19±2.39%, respectively, and were significantly lower in the case group (P=0.001). ROC curve analysis suggested a cutoff point for PLT as 215.50 (109/μL, larger values indicate pathology) for the diagnosis of DSSI with the sensitivity and specificity of 79.3 and 72.4%, respectively. For PDW, the cutoff point was 10.35% (smaller val

Zhen Wang, Ji Xiao, Zitao Zhang, Xusheng Qiu, Yixin Chen Department of Orthopedics, Drum Tower Hospital Clinical College of Nanjing University, Nanjing, China Background: Preoperative deep vein thrombosis (DVT) is a common complication in patients with hip fractures. Chronic kidney disease (CKD) as a frequent comorbidity in middle-aged and elderly patients with hip fractures is known to promote a proinflammatory and prothrombotic state. We aimed to identify whether CKD can increase the risk of DVT in middle-aged and elderly patients with hip fractures, as well as identify other risk factors.Patients and methods: We retrospectively studied 248 middle-aged and elderly patients with hip fractures who were admitted to our hospital from January 2016 to June 2017, meeting all the inclusion criteria. Doppler ultrasonography was used to diagnose DVT. Patients with CKD were classified into five stages according to the Kidney Diseases Outcomes Quality Initiative. We identified whether CKD could increase the occurrence of preoperative DVT in middle-aged and elderly patients with hip fractures and further investigated other independent risk factors for preoperative DVT by using univariate and multivariate analyses.Results: The mean estimated glomerular filtration rate was 125.7±41.4 mL/min/1.73 m2. Briefly, 82.3% (n=204) had been diagnosed with normal kidney function, 11.7% (n=29) with mildly decreased kidney function, and 6.0% (n=15) with stage 3 and 4 CKD. In addition, of the 248 patients, 11.7% (n=29) developed DVT before surgery. Independent predictors of preoperative DVT identified by multivariate analyses were age, prolonged bedridden time, increased levels of fibrinogen, and lower estimated glomerular filtration rate.Conclusion: CKD can increase the risk rate of preoperative DVT in middle-aged and elderly patients with hip fractures. Additionally, age, prolonged bedridden time, and increased levels of fibrinogen were also independent risk factors for preoperat

薄膜压力传感器因其厚度小、柔软、测压单元面积小等特点,可准确测量土体应力及土颗粒间作用力,因而在土工试验中有着广阔的应用前景。初步探究了薄膜压力传感器在静力和动力土工试验中的适用性。首先,提供了结合3D打印技术,将薄膜压力传感器安装于土体内部设定位置上测量相应土体应力及颗粒间作用力的方法。其次,提出了测量静力时直接建立初始DO值与施加压强值间函数关系的标定方法,并建议选用拟合优度判定系数（R2）判定测压单元表现的优劣。最后,结合模型桩试验,验证了在动力试验中通过薄膜压力传感器、利用初始DO值可大体了解土体应力在动力荷载下的变化趋势。 Due to their small thickness, high flexibility and tiny sensing element, the tactile pressure sensors can be used to measure the soil stresses and contact forces between soil particles. Hence, the tactile pressure sensors have a vast application prospect in laboratory geotechnical tests. This study aims to examine the feasibility of applying the tactile pressure sensors in static and dynamic tests. First, an installation approach for the tactile pressure sensors, which relies on 3D printed sensor holders, is proposed to place the sensors in the designed locations. Second, a modified calibration method is proposed for the static tests. In this calibration method, the "equilibration" step is abandoned. A relationship between the original digital output and the applied pressure is used in the calibration. In addition, it is suggested that only the measurements from the sensors with higher values of R2 are used in the analysis. Finally, for the dynamic tests, the results of the model pile tests validate the feasibility of tactile pressure sensors and suggest that the change in the digital output might imply the trend of the soils stresses. © 2017, Editorial Office of Chinese Journal of Geotechnical Engineering. All right reserved.

Driven by the confluence of technology trends in the past half-decade - frictionless manufacturing, big data technologies, hardware-software convergence, ever-falling cost along with increasing performance of sensors and the maturing of mechanical and electromagnetic wave-based characterization techniques in geophysical testing, we believe we have reached the watershed moment of a disruptive change for geotechnical testing and monitoring. In this paper, using 3D printer as customizable manufacturing agent that integrates the new sensing technologies; as for the acquisition system, our Smart Soil Particle (SSP) system adopts the powerful credit-card sized single board computer which is capable of running a general purpose operating system as the brain that manages the underlying hardware and sensors. For the data center, we adopt the open source big data technologies VerneMQ, Apache Cassandra, FiloDB and Apache Spark to achieve linear scalability in both performance and cost on distributed cheap commodity hardware. Through the technology stack of 3D printer, SSP sensor node instrumentation and data center, we hope that the tightly integrated software intelligence and physical world of characterization measurement will steer geotechnical testing and monitoring to a truly smart and data-driven enterprise.

Driven by the confluence of technology trends in the past half-decade - frictionless manufacturing, big data technologies, hardware-software convergence, ever-falling cost along with increasing performance of sensors and the maturing of mechanical and electromagnetic wave-based characterization techniques in geophysical testing, we believe we have reached the watershed moment of a disruptive change for geotechnical testing and monitoring. In this paper, using 3D printer as customizable manufacturing agent that integrates the new sensing technologies; as for the acquisition system, our Smart Soil Particle (SSP) system adopts the powerful credit-card sized single board computer which is capable of running a general purpose operating system as the brain that manages the underlying hardware and sensors. For the data center, we adopt the open source big data technologies VerneMQ, Apache Cassandra, FiloDB and Apache Spark to achieve linear scalability in both performance and cost on distributed cheap commodity hardware. Through the technology stack of 3D printer, SSP sensor node instrumentation and data center, we hope that the tightly integrated software intelligence and physical world of characterization measurement will steer geotechnical testing and monitoring to a truly smart and data-driven enterprise.

Aging- or creep-related phenomena in sand have been widely studied, and the discrete element method (DEM) has been frequently used to model the associated soil behavior and then to explore the associated underlying mechanisms. However, several difficulties involved in modeling still remain unsolved. To resolve these difficulties, a new approach based on the effect of the microfracturing of asperities is proposed in this study for the DEM modeling of the sand aging or creep process through several aging cycles of associated reduction in the mobilized friction resistance at particle contacts and subsequent particle rearrangement to reach a new equilibrium state. This approach can be easily incorporated into different contact models and DEM simulations of the loading, unloading, and/or reloading processes, in either drained or undrained conditions, before and/or after aging. This new approach is proven effective because the DEM simulations incorporated with this new approach can satisfactorily reproduce the experimental observations in the triaxial creep process, drained and undrained recompression after aging, and 1D secondary compression and rebound. The simulation results also indicate that, based on the stress–force–fabric relationship, the contribution from the contact normal anisotropy to the deviatoric stress q gradually increases, whereas the contribution from the tangential force anisotropy becomes less during triaxial creep under a constant q. Moreover, the contacts between particles are gradually away from the state where the frictional resistance is fully mobilized, and then become more stable. During the subsequent triaxial recompression after creep, the aged samples exhibit enhanced soil stiffness, which is also found to be associated with the evolution of the invariants of the anisotropy tensors. It is worthwhile noting that the aging or creep effects on the microstructural changes, e.g., the invariants of the anisotropy tensors, can be gradually erased

Aging- or creep-related phenomena in sand have been widely studied, and the discrete element method (DEM) has been frequently used to model the associated soil behavior and then to explore the associated underlying mechanisms. However, several difficulties involved in modeling still remain unsolved. To resolve these difficulties, a new approach based on the effect of the microfracturing of asperities is proposed in this study for the DEM modeling of the sand aging or creep process through several aging cycles of associated reduction in the mobilized friction resistance at particle contacts and subsequent particle rearrangement to reach a new equilibrium state. This approach can be easily incorporated into different contact models and DEM simulations of the loading, unloading, and/or reloading processes, in either drained or undrained conditions, before and/or after aging. This new approach is proven effective because the DEM simulations incorporated with this new approach can satisfactorily reproduce the experimental observations in the triaxial creep process, drained and undrained recompression after aging, and 1D secondary compression and rebound. The simulation results also indicate that, based on the stress–force–fabric relationship, the contribution from the contact normal anisotropy to the deviatoric stress q gradually increases, whereas the contribution from the tangential force anisotropy becomes less during triaxial creep under a constant q. Moreover, the contacts between particles are gradually away from the state where the frictional resistance is fully mobilized, and then become more stable. During the subsequent triaxial recompression after creep, the aged samples exhibit enhanced soil stiffness, which is also found to be associated with the evolution of the invariants of the anisotropy tensors. It is worthwhile noting that the aging or creep effects on the microstructural changes, e.g., the invariants of the anisotropy tensors, can be gradually erased

Aging- or creep-related phenomena in sand have been widely studied, and the discrete element method (DEM) has been frequently used to model the associated soil behavior and then to explore the associated underlying mechanisms. However, several difficulties involved in modeling still remain unsolved. To resolve these difficulties, a new approach based on the effect of the microfracturing of asperities is proposed in this study for the DEM modeling of the sand aging or creep process through several aging cycles of associated reduction in the mobilized friction resistance at particle contacts and subsequent particle rearrangement to reach a new equilibrium state. This approach can be easily incorporated into different contact models and DEM simulations of the loading, unloading, and/or reloading processes, in either drained or undrained conditions, before and/or after aging. This new approach is proven effective because the DEM simulations incorporated with this new approach can satisfactorily reproduce the experimental observations in the triaxial creep process, drained and undrained recompression after aging, and 1D secondary compression and rebound. The simulation results also indicate that, based on the stress–force–fabric relationship, the contribution from the contact normal anisotropy to the deviatoric stress q gradually increases, whereas the contribution from the tangential force anisotropy becomes less during triaxial creep under a constant q. Moreover, the contacts between particles are gradually away from the state where the frictional resistance is fully mobilized, and then become more stable. During the subsequent triaxial recompression after creep, the aged samples exhibit enhanced soil stiffness, which is also found to be associated with the evolution of the invariants of the anisotropy tensors. It is worthwhile noting that the aging or creep effects on the microstructural changes, e.g., the invariants of the anisotropy tensors, can be gradually erased

This paper aims to use a tomographic image system to obtain the spatial distribution of P-wave velocity in soil and then to derive the associated shear modulus (Gmax). The derivation is based on the relationship between Vp and Vs and the fact that the measured Poisson's ratio μ is fairly independent of the stress state. This tomographic image technique is also utilized in a jacked model pile test to explore the underlying mechanism of friction fatigue. The experimental results demonstrate that the shear modulus (Gmax) in the soil surrounding the pile decreases after the pile is subjected cyclic loading. Such a decrease may lead to the decrease of the ultimate radial stress on the pile shaft and eventually to reduce the ultimate shaft resistance, i.e., the result of friction fatigue during pile installation. © 2015 Taylor & Francis Group.

Driven and jacked piles are widely used in modern building practice, nevertheless, the mechanisms of pile installation and pile setup in sand still remain unclear or even unexplored at present. The main objective of this research is to carry out a comprehensive study in order to enhance understanding on pile installation and pile setup and to determine the involved micromechanics. To explore the mechanisms of a driven pile setup in dry sand, model driven pile tests were carried out in a custom-made pressurized chamber. A bender element system and tactile pressure sensors were used in parallel to monitor the stiffness and stress changes in the soil surrounding the pile during the tests. The experimental results demonstrate that pile setup is not caused by the increase of at-rest radial stress σps', rather, it is mainly attributed to the increase in radial stress during pile loading Δσpf', as a result of soil aging (or creep). Pile installation pushes the surrounding soil to the side, thereby imposing additional loading on the soil inside the influence zone. This loading action initiates an associated aging (or creep) process during the setup period and the aging effects ultimately give rise to an increase in Δσpf' and the pile shaft resistance. The measurements also reveal that the increase in aging-induced soil stiffness is due to the contact normal forces among soil particles gradually becoming more homogenized during the setup period. This suggested setup mechanism can explain the absence of pile setup in bored piles (non-displacement piles) — the loading action induced by the pile installation is too insignificant to trigger the required aging process. In addition, in a similar logic, it can elucidate why the setup rate is higher in large-displacement piles than in small-displacement piles. To examine the soil responses in terms of the strain and stress distributions, the stress paths, the movement of particles and contact force m

Static and dynamic liquefaction may produce different levels of damages on the infrastructures and various numbers of casualties. Understanding the associated initiation mechanisms therefore is very important for geotechnical engineers to predict and manage the relevant hazards. In this study, a series of DEM simulations of undrained triaxial compression tests and a cyclic undrained triaxial test are carried out to examine the initiation mechanisms from a micromechanical perspective. The simulation results demonstrate a continuous decrease in the mean of contact normal forces and a continuous increase in the associated COV as the shear strain increases during undrained compression of a loose sample. Such kind of evolution trend found in the contact force distribution was not observed in the simulations of undrained compression tests on medium dense to dense sand samples. However, the same behavior can be detected in the cyclic undrained triaxial tests on a dense sample, i.e., a continuous decrease in the mean of contact normal forces and a continuous increase in the associated COV. Hence, such an evolution tendency of contact force distributions can serve as a basis to explain the initiation mechanisms of the static and dynamic liquefaction from a micromechanical point of view.

In this study, model pile tests were carried out in a custom-made pressurized chamber to explore the mechanisms of a driven pile setup in dry sand. A bender element system and tactile pressure sensors were used in parallel to monitor the stiffness and stress changes in the soil surrounding the pile during the test. The experimental results demonstrate that pile setup is not caused by the increase in at-rest radial stress sigma rp. Instead, it is mainly attributed to the increase in radial stress during pile loading sigma rp, as a result of soil aging (or creep). Pile installation pushes the surrounding soil to the side, thereby imposing additional loading on the soil inside the influence zone. This loading action initiates an associated aging (or creep) process during the setup period; the aging effects ultimately give rise to an increase in both sigma rp and pile shaft resistance. The measurements also reveal that the increase in aging-induced soil stiffness is due to contact normal forces among soil particles gradually becoming more homogenized during the setup period. This suggested setup mechanism could explain the absence of pile setup in bored (nondisplacement) pilesthe loading action induced by pile installation is too insignificant to trigger the required aging process. Using a similar logic, the mechanism could elucidate why the setup rate is higher in large-displacement piles than in small-displacement piles. (C) 2014 American Society of Civil Engineers.

Understanding the soil responses during monotonic jacking can help improve the estimation of axial capacity in pile foundation design. To examine the soil responses in terms of strain and stress distributions, stress paths, the movement of particles and contact force mobilization during monotonic jacking, a three-dimensional simulation of a centrifuge model pile test was performed using the discrete element method. The simulation results show great resemblance to the measurements from the centrifuge model pile test. The distribution of incremental deviatoric strain d epsilon(q) indicates the formation of a shear band in the shoulder of the 'nose cone' that is beneath the pile base. Based on this, the soil mass can be divided into three zones, i.e., the 'nose cone' (zone I), the shear band (zone II) and the surrounding soil beyond the shear band (zone III). The distributions of the radial stress sigma(r)', hoop stress sigma(theta)', vertical stress sigma(v)' and shear stress tau(rv)' are in different modes and high-gradient distributions of mean stress p' and deviatoric stress q are found in zone II. As for the stress path, the surrounding soil experiences four phases in terms of changes in p' and q during monotonic jacking, and the change in the soil stress state decreases with increasing distance from the pile centerline. From a microscopic point of view, the particles in the shear band rotate strongly in a similar direction, fully mobilizing the frictions at the associated contacts. This provides a micromechanical explanation for the soil failure during monotonic jacking.

In this study, model pile tests were carried out in a custom-made pressurized chamber to explore the mechanisms of a driven pile setup in dry sand. A bender element system and tactile pressure sensors were used in parallel to monitor the stiffness and stress changes in the soil surrounding the pile during the test. The experimental results demonstrate that pile setup is not caused by the increase in at-rest radial stress sigma rp. Instead, it is mainly attributed to the increase in radial stress during pile loading sigma rp, as a result of soil aging (or creep). Pile installation pushes the surrounding soil to the side, thereby imposing additional loading on the soil inside the influence zone. This loading action initiates an associated aging (or creep) process during the setup period; the aging effects ultimately give rise to an increase in both sigma rp and pile shaft resistance. The measurements also reveal that the increase in aging-induced soil stiffness is due to contact normal forces among soil particles gradually becoming more homogenized during the setup period. This suggested setup mechanism could explain the absence of pile setup in bored (nondisplacement) pilesthe loading action induced by pile installation is too insignificant to trigger the required aging process. Using a similar logic, the mechanism could elucidate why the setup rate is higher in large-displacement piles than in small-displacement piles. (C) 2014 American Society of Civil Engineers.

Understanding the soil responses during monotonic jacking can help improve the estimation of axial capacity in pile foundation design. To examine the soil responses in terms of strain and stress distributions, stress paths, the movement of particles and contact force mobilization during monotonic jacking, a three-dimensional simulation of a centrifuge model pile test was performed using the discrete element method. The simulation results show great resemblance to the measurements from the centrifuge model pile test. The distribution of incremental deviatoric strain d epsilon(q) indicates the formation of a shear band in the shoulder of the 'nose cone' that is beneath the pile base. Based on this, the soil mass can be divided into three zones, i.e., the 'nose cone' (zone I), the shear band (zone II) and the surrounding soil beyond the shear band (zone III). The distributions of the radial stress sigma(r)', hoop stress sigma(theta)', vertical stress sigma(v)' and shear stress tau(rv)' are in different modes and high-gradient distributions of mean stress p' and deviatoric stress q are found in zone II. As for the stress path, the surrounding soil experiences four phases in terms of changes in p' and q during monotonic jacking, and the change in the soil stress state decreases with increasing distance from the pile centerline. From a microscopic point of view, the particles in the shear band rotate strongly in a similar direction, fully mobilizing the frictions at the associated contacts. This provides a micromechanical explanation for the soil failure during monotonic jacking.

Static and dynamic liquefaction may produce different levels of damages on the infrastructures and various numbers of casualties. Understanding the associated initiation mechanisms therefore is very important for geotechnical engineers to predict and manage the relevant hazards. In this study, a series of DEM simulations of undrained triaxial compression tests and a cyclic undrained triaxial test are carried out to examine the initiation mechanisms from a micromechanical perspective. The simulation results demonstrate a continuous decrease in the mean of contact normal forces and a continuous increase in the associated COV as the shear strain increases during undrained compression of a loose sample. Such kind of evolution trend found in the contact force distribution was not observed in the simulations of undrained compression tests on medium dense to dense sand samples. However, the same behavior can be detected in the cyclic undrained triaxial tests on a dense sample, i.e., a continuous decrease in the mean of contact normal forces and a continuous increase in the associated COV. Hence, such an evolution tendency of contact force distributions can serve as a basis to explain the initiation mechanisms of the static and dynamic liquefaction from a micromechanical point of view.

In this study, model pile tests were carried out in a custom-made pressurized chamber to explore the mechanisms of a driven pile setup in dry sand. A bender element system and tactile pressure sensors were used in parallel to monitor the stiffness and stress changes in the soil surrounding the pile during the test. The experimental results demonstrate that pile setup is not caused by the increase in at-rest radial stress sigma rp. Instead, it is mainly attributed to the increase in radial stress during pile loading sigma rp, as a result of soil aging (or creep). Pile installation pushes the surrounding soil to the side, thereby imposing additional loading on the soil inside the influence zone. This loading action initiates an associated aging (or creep) process during the setup period; the aging effects ultimately give rise to an increase in both sigma rp and pile shaft resistance. The measurements also reveal that the increase in aging-induced soil stiffness is due to contact normal forces among soil particles gradually becoming more homogenized during the setup period. This suggested setup mechanism could explain the absence of pile setup in bored (nondisplacement) pilesthe loading action induced by pile installation is too insignificant to trigger the required aging process. Using a similar logic, the mechanism could elucidate why the setup rate is higher in large-displacement piles than in small-displacement piles. (C) 2014 American Society of Civil Engineers.

Understanding the soil responses during monotonic jacking can help improve the estimation of axial capacity in pile foundation design. To examine the soil responses in terms of strain and stress distributions, stress paths, the movement of particles and contact force mobilization during monotonic jacking, a three-dimensional simulation of a centrifuge model pile test was performed using the discrete element method. The simulation results show great resemblance to the measurements from the centrifuge model pile test. The distribution of incremental deviatoric strain d epsilon(q) indicates the formation of a shear band in the shoulder of the 'nose cone' that is beneath the pile base. Based on this, the soil mass can be divided into three zones, i.e., the 'nose cone' (zone I), the shear band (zone II) and the surrounding soil beyond the shear band (zone III). The distributions of the radial stress sigma(r)', hoop stress sigma(theta)', vertical stress sigma(v)' and shear stress tau(rv)' are in different modes and high-gradient distributions of mean stress p' and deviatoric stress q are found in zone II. As for the stress path, the surrounding soil experiences four phases in terms of changes in p' and q during monotonic jacking, and the change in the soil stress state decreases with increasing distance from the pile centerline. From a microscopic point of view, the particles in the shear band rotate strongly in a similar direction, fully mobilizing the frictions at the associated contacts. This provides a micromechanical explanation for the soil failure during monotonic jacking.

Static and dynamic liquefaction may produce different levels of damages on the infrastructures and various numbers of casualties. Understanding the associated initiation mechanisms therefore is very important for geotechnical engineers to predict and manage the relevant hazards. In this study, a series of DEM simulations of undrained triaxial compression tests and a cyclic undrained triaxial test are carried out to examine the initiation mechanisms from a micromechanical perspective. The simulation results demonstrate a continuous decrease in the mean of contact normal forces and a continuous increase in the associated COV as the shear strain increases during undrained compression of a loose sample. Such kind of evolution trend found in the contact force distribution was not observed in the simulations of undrained compression tests on medium dense to dense sand samples. However, the same behavior can be detected in the cyclic undrained triaxial tests on a dense sample, i.e., a continuous decrease in the mean of contact normal forces and a continuous increase in the associated COV. Hence, such an evolution tendency of contact force distributions can serve as a basis to explain the initiation mechanisms of the static and dynamic liquefaction from a micromechanical point of view.

In this study, model pile tests were carried out in a custom-made pressurized chamber to explore the mechanisms of friction fatigue for jacked piles in sand. Tactile pressure sensors were used to monitor the changes in stresses and interparticle contact forces in the soil surrounding the pile during cyclic jacking. The measurements suggest that the influence of the cyclic jacking process on the radial and hoop stresses of the surrounding soils at a certain depth can be categorized into two stages, i.e., Stage-1 where the pile penetration depth z < the transition depth z* and Stage-2 where z > z*. The transition depth z* is associated with the maximum stationary radial stress σrs' or hoop stress σθs' obtained during the whole pile penetration process. In Stage-1, the values of stationary radial stress σrs', ultimate radial stress σru', and their difference \Δσr'\ = \σru' - σrs'\ continue to increase with increasing penetration depth z. In Stage-2, the values of σrs', σru', and \Δσr'\ decrease with increasing z and the associated soil structure gradually becomes unstable. Similar responses are also observed in the measured hoop stress. It is also found that z* decreases linearly with decreasing distance to the pile centreline. Based on these experimental findings, a simple model is proposed to explain friction fatigue. In response to continuous jacking cycles and increasing pile penetration depths, the base gradually reaches the transition depth of the soil elements surrounding the pile one by one, followed by the elements gradually entering Stage-2. Hence, the soil structure in the nearer zone becomes more unstable, and the associated ultimate radial stresses gradually decrease, which in turn leads to decreases in the ultimate radial and shear stresses on the pile shaft, causing friction fatigue.

This paper aims to propose a three-dimensional DEM procedure to simulate the centrifuge model pile tests and then to examine the involved penetration mechanisms. The simulation results show great resemblance to the measurements of selected centrifuge model pile test and can give accurate estimations of the resistance. The distribution of the incremental deviatoric strain indicates that the strain localization leads to the formation of shear band in the shoulder of the nose cone that is beneath the pile base. The contour map of shear stress also indicates that the shear stress concentrates in a narrow zone inside the regime of shoulder and implies the associated principal stress rotation during pile installation. In addition, the associated particle movement and contact force distribution provide micromechanical evidence to the development of shear band as well as the rotation of principal stress inside the same regime during monotonic installation of the pile. © 2014 Taylor & Francis Group.

This paper aims to propose a three-dimensional DEM procedure to simulate the centrifuge model pile tests and then to examine the involved penetration mechanisms. The simulation results show great resemblance to the measurements of selected centrifuge model pile test and can give accurate estimations of the resistance. The distribution of the incremental deviatoric strain indicates that the strain localization leads to the formation of shear band in the shoulder of the nose cone that is beneath the pile base. The contour map of shear stress also indicates that the shear stress concentrates in a narrow zone inside the regime of shoulder and implies the associated principal stress rotation during pile installation. In addition, the associated particle movement and contact force distribution provide micromechanical evidence to the development of shear band as well as the rotation of principal stress inside the same regime during monotonic installation of the pile. © 2014 Taylor & Francis Group.

This paper aims to propose a three-dimensional DEM procedure to simulate the centrifuge model pile tests and then to examine the involved penetration mechanisms. The simulation results show great resemblance to the measurements of selected centrifuge model pile test and can give accurate estimations of the resistance. The distribution of the incremental deviatoric strain indicates that the strain localization leads to the formation of shear band in the shoulder of the nose cone that is beneath the pile base. The contour map of shear stress also indicates that the shear stress concentrates in a narrow zone inside the regime of shoulder and implies the associated principal stress rotation during pile installation. In addition, the associated particle movement and contact force distribution provide micromechanical evidence to the development of shear band as well as the rotation of principal stress inside the same regime during monotonic installation of the pile. © 2014 Taylor & Francis Group.